Introduction

As the dimension of CMOS devices is down to the nanoscale, metal silicide technology will be even more significant; the substrate of many photonics and microelectronics devices has been silicon. Transition-metal silicides have been studied extensively owing to their outstanding properties, including low resistivity, and great stability [1,2,3,4,5]. For instance, CrSi2, β-FeSi2, and MnSi are suitable as thermoelectric materials due to their narrow energy gap and great thermostability [6]; NiSi, CoSi2, and TiSi2 are frequently utilized as materials of the metal gate for decreasing the resistance [7].

With excellent properties, such as high compatibility and low defect density, one-dimensional nanostructures are promising for current and future microelectronic devices [8], drawing widespread attention not only from academic studies but also from industry applications [9]. For the past few years, growth kinetics of various metal silicide nanowires, including transition-metal silicides and rare-earth silicides, has been studied [10,11,12,13,14].

There are different phases for iron disilicides [15,16,17,18,19], among which, the unusual characteristics of β-FeSi2 is particularly fascinating. As previously reported, β-FeSi2 nanowires were demonstrated to have important applications in the field of communication [20]; unfortunately, over the many years, few have been able to successfully repeat the fabrication of β-FeSi2 nanowires with chemical vapor deposition. For β-FeSi2, the room temperature equilibrium phase, the potential applications in light emitters, and infrared detectors for silicon-based optoelectronics are attributed to its direct bandgap. In this work, we report direct growth and structural characterization of the single crystalline β-FeSi2 nanowires via a chemical vapor deposition method. The as-synthesized β-FeSi2 nanowires exhibited the room-temperature ferromagnetic behavior. Field emission measurements show that the β-FeSi2 nanowires are great field emission materials.

Methods

In this study, we synthesized β-iron disilicide nanowires using chemical vapor deposition with anhydrous FeCl3 powder as a precursor, silicon (100) substrates, and Ar carrier gas but without any catalysts. Silicon substrates were cleaned with 3 %-buffered HF and put in the downstream zone of the furnace; anhydrous FeCl3 powder was placed in an alumina boat upstream from the substrates, the temperature range of which was 750 ~ 950 °C. We carefully varied temperatures, duration time, and the flow rates of carrier gases in order to realize the factors that influenced the growth of β-iron disilicide nanowires. We utilized scanning electron microscopy (SEM) to investigate the morphology of the β-FeSi2 nanowires; X-ray diffraction (XRD) and transmission electron microscopy (TEM) studies were conducted for structural identification. In addition, characteristics such as magnetism and field emission property were measured. The magnetic property measurements of the β-FeSi2 nanowires were conducted by the Superconducting Quantum Interference Device (SQUID) with the VSM option, while field emission property was measured by Kiethly-237.

Results and Discussion

We explored the parameters that could affect the growth of the β-iron disilicide nanowires. Firstly, different gas flow rates were investigated from 50 to 200 sccm as shown in the SEM images of Fig. 1a–c. Figure 1a reveals the gas flow rate at 50 sccm, where we found lots of nanowires with the diameters of 40 nm and lengths of 10 μm. Figure 1b shows the gas flow rate at 80 sccm, where there were some nanowires but the amount was reduced. In Fig. 1c showing the gas flow rate at 120 sccm, there were even fewer nanowires formed. According to the results, the amount of nanowires decreased with the increase of the gas flow rates. When the nanowires grow, the precursor, FeCl3, should be carried to the downstream zone of the tube furnace and react with the Si substrate by a carrier gas. At higher gas flow rates, it may be difficult for the nanowires to grow. Based on the chemical vapor deposition mechanisms, there were generally five steps in the deposition process, namely, (1) diffusion of reactants to the surface, (2) absorption of reactants at the surface, (3) chemical reaction at the surface, (4) desorption of products from the surface, and (5) diffusion of products from the surface. The slowest step determines the rate of the CVD reaction. If (1) or (5) is the slowest step, it is mass transfer-controlled. If (2), (3), or (4) is the slowest step, it can be called “surface reaction-controlled.” At low temperatures and slow gas flow rates, the surface chemical reaction is slower than reactant diffusion; thus, it is surface reaction-controlled. When it is surface reaction-controlled, variation in the film thickness across the wafer in the chamber will depend on the distribution of temperature, and thin film tends to form. However, our purpose is to grow nanowires; therefore, we should avoid surface reaction-controlled reaction. On the other hand, it is mass transfer-controlled at high temperatures and low gas flow rates. When it is mass transfer-controlled, the rate of the top obtained reactants is faster than that at the sidewall since the axial growth is faster than the radial growth; as a result, nanowires tend to form, and thus, we obtained many nanowires with decreasing gas flow rates. Therefore, mass transfer-controlled reaction is necessary for nanowire growth.

Fig. 1
figure 1

SEM images of β-FeSi2 nanowires at different parameters. At different gas flow rates: a 50 sccm, b 80 sccm, and c 120 sccm. At different temperatures: d 750 °C, e 850 °C, and f 950 °C. At different duration times: g 1 h, h 2 h, and i 5 h

Full size image

The second parameter we investigated was different growth temperatures as shown in the SEM images of Fig. 1d–f. Figure 1d reveals the growth temperature at 750 °C, where there were some nanowires but their lengths and diameters were short and small. Figure 1e shows the growth temperature at 850 °C, where we found lots of nanowires with the diameters of 40 nm and lengths of 10 μm. When we increased the growth temperature to 950 °C as shown in Fig. 1f, nanowires became nanorods due to more deposition of precursors. The third parameter we investigated was the duration time; Fig. 1g–i shows the SEM images for 1 h, 2 h, and 5 h. Generally, we found longer nanowires with the increasing duration time. After more than 5 h, the morphology of the nanowires would not change significantly, which may be attributed to the fact that the precursor had been completely consumed.

To identify the structure of the nanowires, we conducted X-ray diffraction (XRD) and transmission electron microscopy (TEM) analysis as shown in Fig. 2. All the peaks in the corresponding XRD spectrum could be indexed to the structure of orthorhombic β-FeSi2 phase as shown in Fig. 2a. Figure 2b is a TEM image showing a single-crystalline β-FeSi2 nanowire; Fig. 2c is the high-resolution TEM image with the inset of the corresponding fast Fourier transform (FFT) diffraction pattern, showing that the β-FeSi2 nanowire has an orthorhombic structure with [200] growth direction and that the interplanar spacings of planes (200) and (111) are 0.493 nm and 0.482 nm, respectively.

Fig. 2
figure 2

a XRD pattern for β-FeSi2 NWs, b a low-magnification TEM image of a β-FeSi2 NW nanowire, and c HRTEM of a β-FeSi2 NW. The inset in c is the corresponding diffraction pattern with [011] zone axis

Full size image

The growth mechanism in our experiment may involve two reactions to produce β-FeSi2 nanowires as shown in Fig. 3. In the first reaction, evaporative FeCl3 was carried to the furnace downstream, reacting with the Si substrate to form β-FeSi2 nanoparticles and by-products of SiCl4 with β-FeSi2 nanoparticles appearing increasingly. In the second reaction, SiCl4 from the first reaction would also react with the precursor of evaporative FeCl3 and form β-FeSi2 and Cl2. With Cl2 carried out by Ar gas, we gradually obtained β-FeSi2 nanowires from both the first and second reactions. The growth mechanism was VS because we did not observe catalyst-like metal droplets at the front end of the nanowire. The synthesis via VLS mechanism requires a catalyst; however, no catalyst was used in the experiment. To further investigate the growth mechanism, we tried hydrogen, which may have a reduction effect; still, no metal catalytic droplet was formed. Therefore, we demonstrate that the growth mechanism was VS.

Fig. 3
figure 3

Schematic illustration of the growth mechanism 1 FeCl3(s) → FeCl3(g); 2 4FeCl3(g) + 11Si(s) → 4β-FeSi2 + 3SiCl4(g); 3 2FeCl3(g) + 4SiCl4(g) → 2β-FeSi2 + 11Cl2

Full size image

The magnetization of β-FeSi2 was interesting with different dimensions. It has been found to exhibit superparamagnetism in nanoparticles, even though no magnetic ordering occurs in bulk [21], while in the case of β-FeSi2 thin film, ferromagnetism was found only at temperatures below 100 K [22]. The ferromagnetic behavior of β-FeSi2 nanowires may be due to the large specific surface area of the nanowire, leading to many unpaired iron atoms on the surface. Additionally, some strain and defects generated during the growth process could be another factor contributing to the ferromagnetism. To examine the magnetic properties of the grown β-FeSi2 nanowires, the magnetic properties were measured using the Superconducting Quantum Interference Device (SQUID) with VSM option.

Figure 4a is the magnetic response only from the silicon substrate, which clearly shows diamagnetic behavior; we subtracted the magnetism of the silicon substrate for all the following magnetism of β-FeSi2 nanowires. The magnetization curve of the β-FeSi2 nanowires was growing in 2 h as shown in Fig. 4b. The nonlinear hysteresis loop curve shows that the β-FeSi2 nanowires exhibited ferromagnetic behavior at room temperature. The coercivity was about 264 Oe. Larger saturation magnetization was found at 2 K because of the decreasing thermal fluctuation. Due to the reduced coordination of the surface iron atoms, or the strain and structural defects in the crystal, β-FeSi2 nanowires grown here were found to be ferromagnetic [23]. Figure 4c shows the magnetization curve of the longer β-FeSi2 nanowires growing in 5 h. From shorter to longer nanowires, the coercivity increased from 264 to 345 Oe at 300 K, and even to 575 Oe at 2 K; saturation magnetization was raised more as well. It was confirmed that the longer nanowires possessed better magnetic properties. Temperature-dependent field cooling (FC) and zero-field cooling (ZFC) magnetization measurements are shown in Fig. 4d, where the magnetization curve did not drop to zero, showing that the curie temperature of β-FeSi2 NWs was higher than room temperature. The ZFC and FC curves of β-FeSi2 NWs did not overlap; the temperature of curve separation is called blocking temperature (Tb), indicating that a large magnetic anisotropy energy barrier distribution existed [24]. When the temperature was lower than Tb, the magnetic anisotropy energy was larger than the thermal fluctuation. As a result, grains were blocked and not impacted by the applied magnetic field; thus, the magnetization was observed.

Fig. 4
figure 4

a Magnetization measurements of the Si substrate. b Magnetization measurements of the shorter β-FeSi2 nanowires at 2 K and 300 K. c Magnetization measurements of the longer β-FeSi2 nanowires at 2 K and 300 K. d Temperature-dependent magnetization of the β-FeSi2 nanowires

Full size image

To explore the field emission properties, we conducted the field emission measurements for the β-FeSi2 nanowires. The sample was measured in a vacuum chamber at ~ 10-6 torr. Figure 5 shows the current density (J) - field (E) plot with β-FeSi2 nanowires of different lengths. According to the Fowler–Nordheim (F–N) plot and the Fowler–Nordheim equation:

$$ J=\left(\mathrm{A}{\ss}^2{E}^2/\varphi \right)\exp \left(-\mathrm{B}{\varphi}^{3/2}/\ss \mathrm{E}\right), $$
Fig. 5
figure 5

The field emission plot of β-FeSi2 NWs with different dimensions. The inset shows the corresponding ln(J/E2)-1/E plot

Full size image

where J is the current density, E is the applied electric field, and φ is the work function; the inset reveals the ln(J/E2)-1/E plot. The field enhancement ß was calculated to be 1060 from the slope of ln(J/E2) = ln(Aß2/φ)-Bφ3/2/ßE, and ß increased from 1060 to 2367 with the increasing length of nanowires, demonstrating that longer β-FeSi2 NWs had better field emission properties than shorter ones, and that β-FeSi2 NWs could be great field emission materials.

Conclusions

β-FeSi2 nanowires were successfully synthesized with a CVD method. Processing parameters, including temperature, gas flow rate, and duration time were investigated for their effect on the nanowire growth. The growth mechanism has been proposed. Unlike bulk and thin-film β-FeSi2, the as-synthesized β-FeSi2 nanowires exhibited room-temperature ferromagnetic behavior. Field emission measurements demonstrate the β-FeSi2 nanowires as potential field emitting materials.