Shuai Ninga and
Zhengjun Zhang*b
aState Key Laboratory of New Ceramics and Fine Processing, School of Materials Science and Engineering, Tsinghua University, Beijing 100084, P. R. China
bKey Laboratory of Advanced Materials (MOE), School of Materials Science and Engineering, Tsinghua University, Beijing 100084, P. R. China. E-mail: zjzhang@tsinghua.edu.cn
First published on 24th November 2014
Undoped ZrO2 thin films are prepared on 〈100〉 Si substrates by reactive DC magnetron sputtering using a Zr target. By controlling the oxygen partial pressure during the deposition process, we can successfully control the phase structure of the as-deposited film, which can be tetragonal, monoclinic or a mixture of them. A magnetic property measurement reveals that phase-dependent d0 ferromagnetism exists in ZrO2 thin films. Specifically, only tetragonal ZrO2 thin films can be room-temperature ferromagnetic. Photoluminescence measurements, X-ray photoelectron spectroscopy analyses and post thermal annealing experiments suggest the d0 ferromagnetism in undoped tetragonal ZrO2 films is mainly driven by oxygen vacancies.
Zirconia, as one of the most important ceramic materials with widespread applications, was predicted theoretically to be a new DMO candidate,12 and has attracted great attention on its RTFM. A number of theoretical calculations and experimental studies have shown that doping,13,14 phase structure15,16 and intrinsic point defects17,18 are all likely to influence its RTFM. As doping could create a large number of point defects19 and modify significantly the electronic structure of the host materials, it is hard to make clear, at the presence of dopants, whether intrinsic or extrinsic defects dominate the RTFM of ZrO2. Therefore, it is of importance to clarify the origin of RTFM in undoped ZrO2 for the development of ZrO2 based dilute magnetic materials.
ZrO2 has three crystalline structures, namely the monoclinic, tetragonal and cubic phases. The monoclinic phase (space group P21/c) is stable thermodynamically at RT. It transforms into the tetragonal phase (space group P42/nmc) at a temperature of ∼1400 K, and into the cubic phase (space group Fm3m) at ∼2570 K.20 Thus it is a big challenge to obtain the high temperature phases at RT. The studies on doping ZrO2 with trivalent or divalent cations suggest that the creation of enough oxygen vacancies is helpful to stabilize the high temperature phase at RT.21 We therefore selected reactive DC magnetron sputtering technique to prepare ZrO2 thin films to control the phase structure for the investigation of the details of RTFM in undoped zirconia.
In this work, we utilized reactive DC magnetron sputtering to prepare undoped ZrO2 thin films, whose phase structure could be monoclinic, tetragonal or a mixture of them by adjusting the oxygen partial pressure during deposition process. We have also performed thermal annealing treatment to modulate the phase structure and defect level on as-deposited films. Taking into account of structure analysis, defect characterization and magnetic property measurement, we found a phase-dependent room-temperature d0 ferromagnetism existing in undoped ZrO2 thin films, and convinced that the ferromagnetism in tetragonal ZrO2 films was mainly driven by oxygen vacancy.
The structure of samples was characterized by X-ray diffraction (XRD, Rigaku D/max 2500 PC) using Cu Kα (λ = 0.154 nm) radiation, at the θ–2θ coupled mode with a scan step of 0.02°. The microstructure of the as-deposited films was also identified by high-resolution transmission electron microscopy (HRTEM, JEOL JEM 2011) and the selected area electron diffraction (SAD) with a working voltage of 200 kV. The magnetic property was obtained at RT by a superconducting quantum interface device (SQUID) magnetometer (MPMS SQUID-VSM), with the magnetic field direction parallel to the surface of samples. The mass of each sample was measured by induced-coupled-plasma atomic emission spectra (ICP-AES) to normalize the magnetization. The photoluminescence (PL) spectra were measured by a Raman spectrometer (LabRam Aramis) at RT using a 325 nm He–Cd laser. The composition and chemical states of samples were examined by X-ray photoelectron spectroscopy (XPS, ESCALAB 250Xi). Before the measurements, the samples were sputtered by Ar+ for 5 minutes to avoid influence of any surface contamination. The XPS spectra were obtained using Al Kα (1486.6 eV) and calibrated using the carbon 1s core level at ∼284.6 eV.
Fig. 3 shows M–H curves of samples S218, S317 and S416 respectively. It should be mentioned that during the whole operations and measurements procedure, nonmagnetic tweezers, fixtures and containers were used to avoid any unintentional ferromagnetic contamination.22 For comparison, the M–H curves were all normalized to the mass correspondingly. One sees that the films exhibited a phase-dependent RTFM behavior, i.e. a clear hysteresis loop was observed only when the tetragonal phase was present in the films (in S218 and S317), with a coercivity of ∼60 Oe shown in the inset of Fig. 3, while the ferromagnetic was not achieved in the film consisting of only the monoclinic phase (S416). Furthermore, one can also see that the film containing more tetragonal phase has a larger saturated magnetization (MS).
Fig. 3 M–H curves of ZrO2 films deposited at different oxygen partial pressures; inset shows the enlarged hysteresis loop of S218. |
Fig. 4(a) depicts a typical XPS spectrum of S218. The absence of peaks at ∼638.8, 706.8, and 777.9 eV (see inset of Fig. 4(a)) excludes the possible contamination of magnetic elements like Mn, Fe, and Co in the samples. The chemical states of Zr and O in the films were investigated by XPS analysis. Fig. 4(c)–(e) show the Zr 3d core level spectra of the three samples. Fig. 4(f)–(h) show the corresponding O 1s core level spectra. For all the samples, the Zr 3d core level spectra can be well fitted by two pairs of peaks, i.e. A (∼178.3 eV) and B (∼180.5 eV) ascribed to 3d5/2 and 3d3/2 of Zr0, and C (∼182.3 eV) and D (∼184.5 eV) ascribed to 3d5/2 and 3d3/2 of Zr4+ respectively. The O 1s core level spectra can be well fitted by two Gaussian peaks, i.e. one at a binding energy of ∼530.0 eV (peak a) ascribed to O2− in the lattice sites, and the other at a higher binding energy of ∼531.0 eV (peak b) due to O2− in the oxygen-deficient regions in the ZrO2.8 Fig. 4(b) plots the area proportion of peaks C and D (black triangle) related to Zr4+ and the area proportion of peak b (blue circle) related to oxygen vacancy, respectively. It is seen that all the three samples are oxygen-deficient and that the number of oxygen vacancy reaches a maximum in S218 with a tetragonal structure, and a minimum in S416 with a monoclinic structure.
Fig. 5(a) shows the PL spectra of the three samples measured at RT. One sees that all PL spectra can be well fitted by two Gaussian peaks centered at ∼410 nm (peak I) and ∼520 nm (peak II), respectively. According to the literature, peak I is assigned to oxygen vacancy, and peak II is ascribed to lower valence cation (Zr3+ or other doping cations).23–25 The intensity of peak I is the highest in S218 with a tetragonal structure, and decreases monotonically in S317 and S416, in which the content of tetragonal phase get decreased. Fig. 5(b) plots the normalized MS of all the samples versus the integrated intensity of peak I, where one observes a good positive linear relationship. It indicates that oxygen vacancies play an important role in the origin of the RTFM in the films.
Fig. 5 (a) Room-temperature PL spectra of all samples. (b) The saturated magnetization versus the integrated area of peak I from PL. |
The XPS and PL results indicate that the RTFM in undoped ZrO2 depends on not only phase structure but also oxygen vacancy within the films. To clarify their respective effects, two series of annealing experiments were carried out: (1) anneal the S218 (a tetragonal phase with a large number of oxygen vacancies) at 350 °C for 1 h in air, to reduce oxygen vacancies in the annealed sample (denoted as A218); (2) anneal the S416 (a monoclinic phase with negligible oxygen vacancy) at 350 °C for 1 h in Ar to introduce more oxygen vacancy in the annealed sample (denoted as A416). XRD analysis of S218 and A218 indicates that no structure change in the film is induced by thermal annealing shown in Fig. 6(a). The Zr 3d core level XPS spectra of S218 and A218 are shown in Fig. 6(c), from which one observes a significant increase in the proportion of peaks related to Zr4+ after air annealing. Fig. 6(d) compares the O 1s core level XPS spectra of two samples, where one sees a significant decrease in the number of oxygen vacancies in A218 comparing with that of S218. Fig. 6(b) shows the comparison of the normalized M–H curves of A218 and S218. One sees that, due to the striking reduction of oxygen vacancies in A218, the RTFM almost disappears. It suggests that the number of oxygen vacancies in tetragonal ZrO2 is very crucial in enhancing the RTFM. As for A416 annealed in argon, XRD shown in Fig. 7(a) indicates no observable change in phase structure. Fig. 7(c)–(d) compare the XPS spectra of Zr 3d and O 1s core level in S416 and A416 respectively, from which a decrease in the proportion of peaks related to Zr4+ and an increase in the number of oxygen vacancies was observed. However, as shown by Fig. 7(b), RTFM was still not achieved in A416 with a monoclinic structure, although its oxygen vacancy level is close to that in S218 with a tetragonal structure. Thus, it suggests that phase structure of ZrO2 film plays a dominant role in the RTFM driven by oxygen vacancy in undoped ZrO2 thin films.
Fig. 6 (a) XRD pattern; (b) M–H curves; (c) high resolution XPS of Zr 3d core level; (d) high resolution XPS of O 1s core level of S218 and A218. |
Fig. 7 (a) XRD pattern; (b) M–H curves; (c) high resolution XPS of Zr 3d core level; (d) high resolution XPS of O 1s core level of S416 and A416. |
There are lots of theoretical studies on the defect-driven RTFM mechanisms corresponding to the magnetic ordering in DMOs. Generally, two models can explain the phenomenon to some extent, i.e. the percolation model of bound magnetic polaron (BMP)26 and charge transfer ferromagnetism (CTF).27 In the BMP model, the percolation threshold for magnetic ordering is determined by the vacancy level. Specifically, more oxygen vacancies help to produce more BMPs and yield a larger overall volume occupied by BMPs. When the density of oxygen vacancy exceeds a certain threshold, it will result in an overlap of BMPs thus enhancing ferromagnetic behavior.28 However, it could not explain why RTFM was not achieved in A416 with a monoclinic structure and a number of oxygen vacancies but achieved in S218 with a tetragonal structure. In the CTF model, the basic idea is that a narrow, structured local density of states NS(E) is associated with defects, but the Fermi level will not normally locate at a peak in NS(E). A local charge reservoir such as oxygen vacancy occupied with electrons or cations of different charge states, provides a possibility for electron transfer to raise the Fermi level to a peak in NS(E), leading to Stoner splitting of NS(E) and triggering the ferromagnetism.29 Put simply, the CTF not only depends on the vacancy levels; it is also associated with the band structure. As the electronic band structure of tetragonal ZrO2 is significantly different from that of the monoclinic one,21,30,31 the phase-dependent RTFM induced by oxygen vacancy in undoped ZrO2 films observed here may be explained by the CTF model.
This journal is © The Royal Society of Chemistry 2015 |