Introduction
III-nitride semiconductor material has a direct wide bandgap, good structural, chemical, and thermal stability, good light transmission characteristics, and low toxicity, which gives it great potential value as a diluted magnetic semiconductor (DMS). DMS materials have attracted considerable interest because of their potential applications in spintronic applications [1,2]. Since the original discovery of magnetism in Mn-doped GaAs at a temperature of 100 K [3,4], III-V based DMSs have been widely investigated [5–8]. Among them, AlN has attracted much attention as a promising DMS because of its stable mechanical properties, wide band-gap, and high Curie temperature (Tc). It is predicted that transition-metal-doped AlN is ferromagnetic above room temperature (RT) [9,10]. Up until now, Mg-, Mn-, Co-, Cr-, Cu-, Mg- and Fe-doped AlN with RT magnetism have been reported [11–15]. However, Ni-doped AlN DMSs are rarely studied because the Ni clusters are easily precipitated in the semiconductors [16]. A series of problems have to be solved to enable actual applications. For example, it is difficult to prepare the RT ferromagnetic materials. Furthermore, the origin of the magnetism in transition metal (TM)-doped AlN DMSs is still debatable. For use in practical applications, Ni-doped AlN DMSs with room temperature magnetism are necessary [17,18]. This work focused on the most intensively investigated III-nitride semiconductor material: AlN films. Theoretical and experimental research was conducted on the Ni-doped AlN diluted magnetic semiconductors.
In this work, we focused on the most intensively investigated III-nitride semiconductor material: AlN films. The Ni-doped AlN diluted magnetic semiconductors were studied by theoretical and experimental methods. The structure, magnetic properties, and mechanism of magnetization of Ni-implanted AlN films were studied. The analysis shows that the Ni-implanted AlN films have a wurtzite structure without the formation of a secondary phase after annealing; this facilitates the transfer of Ni ions from interstitial sites to substitutional sites. The magnetic measurement indicates that the Ni-doped AlN films exhibited magnetism at 300 K. The p-d exchange mechanism is also discussed in detail from the first-principles calculation.
Experimental
AlN films were prepared on sapphire substrates by metalorganic chemical vapor deposition (MOCVD). The samples were grown on c-sapphire substrates in a vertical cold wall MOCVD reactor under a pressure of 40 Torr. Trimethylaluminum (TMAl) and ammonia were used as precursors for the Al and N sources, and H2 was used as carrier gas. First, a thin high-temperature AlN (HT-AlN) film, around 60 nm thick, was deposited on the sapphire substrate via the pulsed atomic layer epitaxy (PALE) technique, with TMAl and ammonia flowing into the growth chamber separately at 1050°C. Subsequently, an AlN intermediate layer with a thickness of 20 nm was deposited at 870°C while TMAl and ammonia were introduced into the chamber simultaneously. Then HT-AlN, around 300 nm thick, was deposited using the same growth conditions as the HT-AlN grown initially. The as-grown AlN film will be designated in this paper as “Sample a”. A high purity (5N) nickel target with a diameter of 10 mm was used for Ni implantation by Metal Vapor Arc (MEVVA) sources at an implantation energy of 100 keV at 300 K for 3 h. The base pressure in the chamber was below 3.5 × 10−3 Pa and the injection dose was 3 × 1017 ion/cm2. Subsequently, the Ni implanted sample was divided into two parts. One was kept to study in the as-implanted state (designated as “Sample b before annealing”) while the other one was annealed at 900°C under a pressure of 840 Pa in the furnace to repair the crystal lattice of the AlN film and to activate the Ni ions (designated as “Sample b after annealing”). The gas flow rate was 1 L/min of N2, and the sample was positioned face-down to avoid the formation of secondary Ni phases such as NiO and Ni2O3. The annealing time was 1 h.
Chemical compositions were measured by X-ray photoemission spectroscopy (XPS, Kratos Amicus, Manchester, UK). Structural properties were determined by X-ray diffraction (XRD, Rigaku D/Max-A, Cu Kα), and magnetic properties were investigated with a vibrating sample magnetometer (VSM) in the temperature range of 900 K. Before the measurement, we used acetone solution, alcoholic solution, and deionized water (in chronological order) to clean the samples and thereby minimize any false artifacts from contaminants.
Results and discussion
Figure 1 shows the XRD spectra for three AlN films. Sample a has a significant diffraction peak at 36.09°, which corresponds to the peak value of AlN (002). There is no other impurity peak in Sample a, which shows that the AlN epitaxial thin film is of high purity. In contrast to Sample a, Sample b has a diffraction peak at 35.97° corresponding to the AlN (002) peak and an additional peak at 44.4° corresponding to the peak of metallic Ni, which indicates that Ni ions were successfully implanted into the AlN epitaxial film and some Ni ions may be present at the interstitial sites of the film after ion implantation. Just as in the case of Sample a, we can only see one obvious peak at 36.05° in Sample b after annealing, without any Ni metal peak, which shows that Sample b after annealing has the wurtzite structure without any other impurity phases, such as NiO and Ni2O3. It is speculated that Ni ions may shift from the interstitial sites to the substitution sites of the crystal lattice.
To study the influence of Ni ion implantation on the lattice constants of AlN thin films, the AlN (002) diffraction peaks in Fig. 1 were analyzed. It is not difficult to find that the AlN (002) diffraction peak in Sample b before annealing shifted to a lower angle than that of Sample a, which indicates that the lattice constant of the AlN film enlarged slightly as the Ni ions were implanted into the AlN film. This is mainly because Ni ions incorporated into the interstitial sites in the AlN film and increased the lattice stress, which led to the lattice expansion. The AlN (002) peak (36.05°) in Sample b after annealing is higher than that (35.97°) in Sample b before annealing, indicating that the lattice of the AlN film has contracted slightly after annealing at 900°C. This result is similar to that reported by Pan et al. [19], who found that Si ions were successfully doped into the substitutional sites in the AlN thin films. In our present experiment, Ni ions transferred from interstitial sites to substitutional sites of AlN after annealing; thus, Ni no longer occupied the interstitial sites in the form of metal particles, causing the shrinkage of the lattice spacing. Compared to Sample a, it can be seen that the AlN (002) diffraction peak of Sample b shifted to the left after annealing. That is to say, the lattice spacing is slightly enlarged. This is because the radius of Ni ions is larger than that of Al ions, and the Ni-N bond is longer than the Al-N bond. Therefore, Ni3+ occupying the location of Al3+ can cause the expansion of lattice constants of AlN film after the annealing treatment. This indicates that the annealing treatment at 900°C is facilitates the transfer of Ni ions from interstitial sites to substitutional sites of AlN and activates the Ni3+ ions.
XPS is largely used for surface characterization, and it can provide valuable information on the chemical state of ions. The XPS measurements were performed with a Kratos Axis Ultra DLD with the monochromatic Al Kα line (1486.7 eV). An analyzer pass energy of 100 eV with a 1 eV energy step was usually used to obtain high resolution photoelectron spectra. Large X-ray spots (500 μm) were used to explore the surface. To further analyze the valence states and the chemical bonding states of Ni and Al, XPS measurements of Sample b before and after annealing were carried out. Figure 2 shows the Ni-2p and Al-2p high-resolution XPS spectra of Sample b. The Gaussian method was used to fit the XPS spectra, and the C-1s line at 285.2 eV was used to calibrate the binding energies. The single peak of Al-2p occurs at 74.1 eV in Fig. 2(a), which corresponds to Al3+-2p3/2 in the AlN wurtzite structure. This indicates that Al ions mainly exist in the Al3+ state, and the chemical bonding states are Al-N, rather than some other state, such as Al-Ni or Ali. These results show that the Ni ion implantation and annealing treatment have little effect on the valence and chemical bonding states of Al in Ni-doped AlN films.
Figure 2(b) shows that the peaks of Ni-2p in Sample b before annealing are at 852.6 and 869.8 eV, which correspond to Ni metal 2p3/2 and Ni metal 2p1/2, respectively. These results show that Ni ions mainly exist in the form of Ni metal particles and the chemical bonding states are Ni-Ni. We believe that Ni ions were successfully implanted into the AlN epitaxial thin film and existed in the interstitial state before annealing. This result is consistent with the results of XRD analysis.
After annealing, the peaks of Ni-2p at 855.2, 860.7, and 873.0 eV can be observed in the XPS spectra of Sample b. Since the peaks at 852.6 and 869.8 eV corresponding to metallic Ni were not observed in Sample b after annealing, the presence of Ni metal clusters can be excluded. Since the Ni-2p3/2 and Ni-2p1/2 peaks of NiO are located at 853.3 and 871.7 eV, respectively [20], which does not correspond to the peaks in Sample b, the chemical bonding states of Ni-O can be ruled out. The distance between Ni-2p3/2 (855.2 eV) and Ni-2p1/2 (873.0 eV) peaks is 17.8 eV, which is distinctly different from that of NiO (18.4 eV). This further indicates that NiO does not existed in Sample b after annealing.
The Ni-2p3/2 (855.2 eV) and Ni-2p1/2 (873.0 eV) correspond well with that for Ni3+-2p3/2 as described in Ref. [21]. To further deduce whether the chemical bonding state of Ni3+ is Ni-O or Ni-N, the magnetization versus magnetic field (M-H) loop measurements for Sample b were conducted, as shown in Fig. 3. The results show that Sample b has high temperature magnetic properties after annealing, whereas Ni2O3 is paramagnetic, indicating that Ni2O3 is not present in Sample b. Moreover, the peak of Ni-2p of Ni2O3 is 855.8 eV, which does not correspond to the peaks in Sample b after annealing; therefore, the presence of Ni2O3 can be ruled out. Hence, it is presumed that the Ni atoms were successfully implanted into the AlN epitaxial thin films, and Ni3+ ions were incorporated into substitutional sites after annealing. Furthermore, the chemical bonding state of Ni3+ is Ni-N in all probability, rather than Ni-O. It is suggested that the annealing treatment can assist Ni ions to transfer from interstitial sites to substitutional sites of AlN and activate the Ni3+ ions, which is consistent with the analysis of the XRD pattern of Sample b after annealing.
To study the magnetic properties of Ni-doped AlN films, the type-b samples were analyzed with a VSM. An external magnetic field of 8000 Oe was applied parallel to the AlN surface at 300 K to measure the magnetization versus magnetic field (M-H) loops of these b-type samples. The diamagnetic properties of pure AlN film were subtracted from the DMS magnetization measurements. As presented in Fig. 3, the apparent hysteresis loops of Sample b were clearly observed, which proves that these films have magnetism at 300 K. The saturation magnetization moment (Ms) and coercivity (Hc) values were also measured.
For Sample b before annealing, Ms is 0.82 emu/g and Hc is 22.8 Oe. From the analysis of the XRD and XPS measurements, it is known that metallic Ni particles exist in Sample b before annealing. The magnetic properties of Sample b before annealing are mainly attributed to Ni metal in the AlN film because of the obvious magnetic properties of metallic Ni. The Ms of Sample b after annealing is 0.36 emu/g and the Hc is 35.29 Oe; in other words, the Ms is weakened after annealing. Since only Ni ions were introduced to the AlN film and there are no Ni clusters existing in Sample b after annealing, the long distance between Ni ions makes it impossible for any direct interaction to occur, which shows that the magnetism is not due to Ni clusters. Simultaneously, because of the high resistivity of AlN films at RT, free-carrier mediated models are not suitable for explaining the magnetism. After annealing, Ni ions transfer from interstitial sites to substitutional sites of AlN and the chemical bonding states are Ni-N. The magnetism may be derived from Ni-N in NiN4 tetrahedron. However, the mechanism of magnetization is still indeterminate, and further research is needed.
It is worth noting that the Hc of Sample b after annealing at RT is higher than that of bulk Ni (6 Oe) [22] and lower than that of Ni-doped AlN DMSs (about 40 Oe) [18]. The coercivity of Sample b after annealing is relatively low, which is more suitable for spin processing. Sample b has high magnetization after annealing at 900°C, which also shows that Ni-doped AlN thin film is a promising DMS material.
To further understand the mechanism of magnetization of Ni-doped AlN, the spin-polarized partial density of states (DOS) of Ni-doped AlN was analyzed from a first-principles study, and was calculated by using the Vienna ab-initio simulation package (VASP). The AlN was created from a 32-atom (2 × 2 × 2) supercell of Al16N16 having the wurtzite structure. One Al atom and one N atom in the supercell were respectively replaced by a Ni atom to study their formation energy. Since the radius and ionicity of Ni is closer to that of Al, than N, it is expected that Ni will be inclined to occupy Al sites in the AlN film. Indeed, the formation energy of NiAl in Ni-doped AlN is 7.02 eV lower than that of NiN, which shows that Ni prefers to occupy the Al site. Figure 4 shows the total DOS of Al16N16 and Al15NiN16, where the Fermi level is set to zero. The majority-spin channel and the minority-spin channel of AlN are completely symmetric in Fig. 4(a), which shows the AlN is nonmagnetic. This result is in keeping with the experimental research. Figure 4(b) shows the total spin DOS of Al15NiN16. The valence band is in the range of −16 ~ −13 eV and −6.6 ~ 0 eV. By comparing the total DOS of Al15NiN16 and Al16N16, it can be found that the d electron of impurity Ni introduces a new impurity level near the Fermi surface. The majority-spin channel is not completely coincident with the minority-spin channel. This indicates that the Ni-doped AlN film is magnetic, which is in accord with our previous analysis of the magnetization measurements.
To further analyze the mechanism of magnetization of Ni-doped AlN, partial DOS of Al-3d, Ni-3d and N-2p are shown in Fig. 5. In the majority-spin channel, the peaks of Ni-3d at −3.40, −2.30, −1.29, −0.86, and −0.34 eV overlap with that of N-2p. In the minority-spin channel, the Fermi level passes through the overlapped band of Ni-3d and N-2p. These characteristics manifest a very strong hybridization between a Ni atom and four nearby N atoms in a NiN4 tetrahedron, which induces the magnetization of Ni and N atoms. A total magnetic moment of 2.99 μB per supercell is calculated; and the local magnetic moment of Ni (1.65 μB) makes the primary contribution. The contribution of the N atom at the top site of the NiN4 tetrahedron is 0.17 μB, which is less than that of the other three N atoms lying in the basal plane of the NiN4 tetrahedron (0.21 μB). This is because the Ni–N bond of the N atom at the top site is longer than that of the other three N atoms. Analyzing the magnetic moments, we found that a majority of the magnetic moments are contributed by the NiN4 tetrahedron. The doped Ni atom hybridizes with four nearby N atoms in a NiN4 tetrahedron; then electrons of the N atoms are spin-polarized and couple with electrons of the Ni atom with strong magnetization, which results in magnetism. Therefore, the p-d exchange mechanism between Ni-3d and N-2p can be the origin of the magnetism.
Conclusions
In summary, we reported on the structure, elemental valences, and magnetic properties of Ni-doped AlN films that were fabricated by ion implantation. The AlN films had been deposited on Al2O3 substrates by MOCVD. The annealing treatment at 900°C facilitated the transfer of Ni ions from interstitial sites to substitutional sites of AlN and activated the Ni3+ ions. The Ni-doped AlN films after annealing have a wurtzite structure without any secondary phase having been formed, as confirmed by the XRD, XPS, and M-H measurements. The RT, Ms and Hc obtained are about 0.36 emu/g and 35.29 Oe, respectively. Sample b exhibited magnetism at 300 K after annealing. A first-principles calculation suggests that the p-d exchange mechanism between Ni-3d and N-2p can be the origin of the magnetism. It is expected that the room temperature ferromagnetic Ni-doped AlN films will have many potential applications as diluted magnetic semiconductors.