1 Introduction

Diluted magnetic semiconductor (DMS) materials have attracted much attention arising from the feasibility of manipulation of both charge and spin degree of freedoms in a single material system [1–4]. Electronic devices employed for data processing in computers such as field-effect transistors in central processing units are made from semiconductor materials; while the memory devices employed for non-volatile data storage in computers such as hard drives are generally made from ferromagnetic materials. Diluted magnetic semiconductor is a kind of material with promising properties of both semiconductors (with a tunable conductance under gate bias) for logic computation and ferromagnetic materials (with a controllable spin polarization) for information storage. With this motivation, research on DMS materials nowadays has become one of the major areas of spintronics [5–8].

Two main factors limit DMS practical applications for devices. One is the Curie temperature of the DMS material; the other is whether the ferromagnetism in the DMS material originates from free-carrier mediation (sometimes referred to as ‘intrinsic’) or purely from localized secondary phases of magnetic dopants such as clusters or precipitates (sometimes referred to as ‘extrinsic’). If the ferromagnetism is not mediated by free carriers, the spin polarization cannot be carried by the free carriers; hence, it is generally not applicable for spintronic devices. The carrier-mediated ferromagnetism has been confirmed in several well-known ‘canonical’ DMS materials, such as Mn-doped GaAs (GaAs:Mn) [9, 10] and Mn-doped InAs (InAs:Mn) [11, 12], however, all these materials show Curie temperatures far below room temperature. On the other hand, above-room-temperature Curie temperatures have been observed in some DMS materials but the origin of the ferromagnetism is still controversial, especially in oxides and nitrides. ZnO is one of this type of DMS materials.

ZnO is a wide bandgap semiconductor material with a direct bandgap of ∼3.37 eV at room temperature [13], which has promising applications for optoelectronic devices [14] due to its large exciton binding energy (∼60 meV) and direct bandgap [15]. ZnO has attracted lots of interest for DMS research initially due to its predicted above-room-temperature Curie temperature based on theoretical calculations [16–18]. Starting from near the end of the last century growing efforts have been focused on ZnO DMS research and above-room-temperature ferromagnetism in ZnO DMS has been widely observed in experiments [19–56]. Up to the present more than 1300 papers have been published in the ZnO DMS area according to ISI-Web-of-Knowledge statistics. However, the origin and mechanism of the ferromagnetism in ZnO DMS are still under intense debate. Recently, the research on ZnO DMS has been primarily focused on the study for clarifying the origin and mechanism of the ferromagnetism in ZnO DMS.

Since the discovery of room-temperature ferromagnetism in ZnO, quite a few review papers have been published on ZnO DMS and other diluted magnetic oxides [4, 57–70]. Compared to previously published comprehensive reviews on ZnO DMS, this brief perspective paper does not go through details on all the aspects in the ZnO DMS area (such as different preparation techniques, growth parameters, transition-metal dopants, etc.), but focuses primarily on the recent progress on three topics: (1) spin coherence in ZnO, (2) free-carrier type and concentration-dependent magnetic properties in ZnO DMS, and (3) ferromagnetism in undoped and non-transition-metal-doped ZnO. The three topics are important for ZnO DMS research. Spin coherence is a critical physical parameter for ZnO DMS device applications. The latter two help to further clarify the origin and mechanism of the ferromagnetic ordering in ZnO DMS and facilitate future development of precisely controllable ZnO DMS.

2 Spin coherence in ZnO

Practical spintronic devices require a long spin-coherence time in the hosting material at room temperature. ZnO material has a wide bandgap and a weak spin–orbit coupling [71], which is likely to lead to a long spin-coherence time. Compared to the large number of publications on ZnO DMS (>1300 papers), the number of publications on spin-coherence study in ZnO is not much (<40 papers), but in recent years it has increased a lot in both experiments [72–84] and theory [85–87]. The representative experimental results of spin coherence in ZnO are summarized in Table 1.

Table 1 Experimental studies of spin coherence in ZnO
Full size table

The seminal work on spin-coherence study in ZnO was done by Ghosh et al. using time-resolved Faraday rotation (TRFR) spectroscopy [72]. Spin-coherence times of 20 ns at 30 K in ZnO bulk single crystals and 2 ns at 10 K and ∼188 ps at 280 K in pulsed laser deposition (PLD) grown thin films were observed. Previously, the spin coherence could only be observed and investigated in high-quality ZnO materials such as ZnO bulk single crystals [72, 77, 81] and epitaxial ZnO thin films growth by molecular-beam epitaxy (MBE) and PLD [72, 75, 77]. Recently, the spin coherence has also been successfully studied in ZnO nanoparticles/quantum dots [73, 76, 78–80], spin-coated sol–gel ZnO thin films [82], and ZnO DMS materials doped with Co and Mn [80, 82]. Figure 1a shows the TRFR spectroscopy of spin-coated sol–gel ZnO thin films at 300 K and 10 K. The measurements were performed at a magnetic field of 1.4 T with laser wavelengths of 375 nm (at 300 K) and 368 nm (at 10 K) [82]. The red solid lines and gray dotted lines show the experimental data and the exponentially damped sinusoidal fits, respectively. The spin-coherence time of the spin-coated ZnO thin film is estimated to be ∼1.2 ns at 300 K. Most of the spin dynamics studies have been focused on the electrons in ZnO. Several studies on the hole [77, 84] and exciton [84] spin dynamics have also been reported.

Fig. 1
figure 1

Spin-coherence in ZnO. (a) Spin coherence of spin-coated sol–gel ZnO thin films. Time-resolved Faraday rotation spectroscopy of the ZnO thin film at 300 K and 10 K with a magnetic field of 1.4 T. The red solid lines and gray dotted lines show the experimental data and the exponentially damped sinusoidal fits, respectively. The spin-coherence time is ∼1.2 ns at 300 K. (b) Annealing effect on spin coherence of ZnO. Time-resolved Kerr rotation spectroscopy of ZnO single crystal (top, red) and after rapid thermal annealing in N2 ambient at 500 °C (middle, blue) and 800 °C (bottom, green). The measurements were performed at 8.5 K with a magnetic field of 90 mT. The circles and solid lines show the experimental data and the exponentially damped sinusoidal fits, respectively. The spin-coherence times are ∼11.2 ns, ∼2.3 ns, and ∼2.0 ns for the ZnO before and after 500 and 800 °C annealing, respectively. (c), (d) Electric field effect on spin coherence of ZnO. Time-resolved Faraday rotation spectroscopy of epitaxial ZnO thin films (c) without (top) or with an in-plane electric field of 4.9 mV/μm (bottom) and (d) at different electric fields of 0–3.6 mV/μm. All the measurements were performed at 20 K with a magnetic field of 0.2 T. The spin-coherence time is enhanced from ∼0.7 ns at zero field to ∼1.6 ns with an electric field of 4.9 mV/μm. [(a) reprinted with permission from Ref. [82]. Copyright the American Chemical Society. (b) reprinted with permission from Ref. [81]; (c)–(d) reprinted with permission from Ref. [75]. Copyright the American Institute of Physics]

Full size image

The spin-coherence property of ZnO can be modified by thermal processing and tuned with external applied bias. Yang et al. reported that the spin coherence in a ZnO single-crystal wafer sample was degraded after a rapid thermal annealing in N2 ambient [81]. Figure 1b shows the time-resolved Kerr rotation (TRKR) spectroscopy of ZnO single crystal (top, red) and after rapid thermal annealing in N2 ambient at 500 °C (middle, blue) and 800 °C (bottom, green). The measurements were performed at 8.5 K with a magnetic field of 90 mT. The open circles and solid lines show the experimental data and the exponentially damped sinusoidal fits, respectively. The spin-coherence times are ∼11.2 ns, ∼2.3 ns, and ∼2.0 ns for the ZnO before and after 500 and 800 °C annealing, respectively. The degraded spin-coherence time is attributed to the annealing-induced surface conducting layer [81]. Ghosh et al. reported that an in-plane electric could enhance the spin coherence in ZnO [75]. Figure 1c shows the TRFR spectroscopy of epitaxial ZnO thin films without (top) and with an in-plane electric field of 4.9 mV/μm (bottom). The spin-coherence time is enhanced from ∼0.7 ns at zero field to ∼1.6 ns with an electric field of 4.9 mV/μm. Figure 1d shows the TRFR mapping at different electric fields of 0–3.6 mV/μm. Both the measurements in Figs. 1c and d were performed at 20 K with a magnetic field of 0.2 T. The capability to electrically manipulate spin coherence is promising. ZnO spin lifetime was also analyzed based on a magnetic tunnel junction device with ZnO as the spacer layer [88].

3 Free-carrier type and concentration-dependent magnetic properties in ZnO

Whether the magnetic properties (e.g. magnetization, Curie temperature, anomalous Hall effect, etc.) of a DMS material show free carrier concentration dependent behavior is an important and straightforward approach to study the origin and mechanism of the ferromagnetism in the DMS. Clear carrier concentration dependent magnetic properties have been demonstrated in ‘canonical’ DMS materials such as GaAs:Mn and InAs:Mn [9–12], while in other DMS materials it is still controversial, for example nitride and oxide DMS. In recent years, preliminary results of carrier concentration dependent magnetic properties have been reported in ZnO DMS materials [89–104]. Generally, two approaches are employed for tuning the carrier concentration in a semiconductor, one is chemical doping and the other is electrostatic doping. Both ways have been attempted to tune the carrier concentration in ZnO DMS and study how the magnetic properties behave. Experimental results of carrier concentration dependent ferromagnetism in ZnO DMS materials are summarized in Table 2.

Table 2 Carrier concentration dependent ferromagnetism in ZnO diluted magnetic semiconductors
Full size table

3.1 Chemical doping

Clear carrier concentration dependent magnetization in ZnO DMS was first observed in Co-doped ZnO with a donor co-dopant (for tuning the electron concentration) of Al [91] and Ga [94, 99], and (Mn,Ga) co-doped ZnO [95, 105]. Yang et al. studied the above-room-temperature ferromagnetism in a series of ZnO:(Mn,Ga) [95, 105] and ZnO:(Co,Ga) [94] thin film samples with various free electron carrier concentrations. Figure 2a shows the magnetic field dependent magnetization at 300 K of four ZnO:(Mn,Ga) thin film samples with electron carrier concentration of 1.2×1020 cm−3 (Mn-A), 4.7×1019 cm−3 (Mn-B), 8.4×1018 cm−3 (Mn-C), and 3.5×1018 cm−3 (Mn-D) [95]. The top and bottom insets show the magnetic field dependent magnetization at 10 K of Mn-A and the temperature-dependent magnetization at 2000 Oe of Mn-A, respectively. Figure 2b shows the relation between the saturated magnetization (M S) and electron carrier concentration (n) of the four ZnO:(Mn,Ga) thin film samples shown in Fig. 2a. Clear electron carrier concentration dependence of the magnetization is observed [95]. The inset shows the x-ray diffraction (XRD) pattern of a ZnO:(Mn,Ga) thin film. The orientation of the thin film is along (\(11\bar{2}0\)) direction with the peak full-width-at-half-maximum (FWHM) of ∼0.29∘. Transmission electron microscopy (TEM) studies were performed on a ZnO:(Mn,Ga) thin film sample with no secondary phase observed [105]. Figure 2c shows the TEM studies of a ZnO:(Mn,Ga) thin film sample [105]. Figure 2c (i)–(iii) show the cross-sectional TEM and high-resolution TEM (HRTEM) images near the ZnO:(Mn,Ga)-sapphire substrate interface region, and HRTEM image of the ZnO:(Mn,Ga) thin film, respectively. The top and bottom insets in (iii) show the fast Fourier transform pattern and diffraction pattern of the ZnO:(Mn,Ga) thin film [105]. Figure 2d shows the magnetic field dependent magnetization at 300 K of four ZnO:(Co,Ga) thin film samples with electron carrier concentration of 8.3×1019 cm−3 (Co-A), 3.6×1019 cm−3 (Co-B), 4.0×1018 cm−3 (Co-C), and 1.3×1018 cm−3 (Co-D) [94]. The top inset shows magnetic field dependent magnetization of sample Co-A at 700 K. The bottom inset shows the temperature-dependent magnetization of sample Co-A at 0.2 T. A high-temperature extrapolation fitting (M∼M 0(T C−T)1/2) indicates the Curie temperature (T C) is around 950 K. Figure 2e shows the relation between the saturated magnetization (M S) versus the electron carrier concentration (n) of the four ZnO:(Co,Ga) thin film samples shown in Fig. 2d. The top and bottom insets show the HRTEM image of a ZnO:(Co,Ga) thin film and the cross-sectional HRTEM image near the ZnO:(Co,Ga)-sapphire substrate interface region [94]. The observed magnetization values are large in the study shown in Figs. 2a and 2d, while comparable and even larger magnetization values in ZnO DMS have been reported elsewhere [106, 107]. The free electron concentration dependence of the anomalous Hall effect (AHE) was also observed in ZnO:(Co,Ga) thin films [99]. Figure 2f shows the magnetic field dependent magnetization (blue) and AHE (black) in (i) ZnO:Co and (ii) ZnO:(Co,Ga) thin films. The ZnO:Co and ZnO:(Co,Ga) thin films show electron concentration (n) of 4.0×1019 and 5.7×1020 cm−3, saturated magnetization (M S) of 0.35 and 0.85 μ B/Co2+, and anomalous Hall conductivity (σ AHE) of 0.002 and 0.078 Ω−1 cm−1, respectively.

Fig. 2
figure 2

Carrier concentration dependent magnetic properties in ZnO DMS thin films. (a) Magnetic field dependent magnetization at 300 K of four ZnO:(Mn,Ga) thin film samples with electron carrier concentration of 1.2×1020 cm−3 (Mn-A), 4.7×1019 cm−3 (Mn-B), 8.4×1018 cm−3 (Mn-C), and 3.5×1018 cm−3 (Mn-D). Top inset: magnetic field dependent magnetization of sample Mn-A at 10 K. Bottom inset: temperature-dependent magnetization of sample Mn-A at 2000 Oe. (b) The plot of the saturated magnetization (M S) versus the electron carrier concentration (n) of the four ZnO:(Mn,Ga) thin film samples shown in (a). Inset: The XRD pattern of a ZnO:(Mn,Ga) thin film. The orientation of the thin film is along (\(11\bar{2}0\)) direction with a peak FWHM of ∼0.29∘. (c) TEM analyses of a ZnO:(Mn,Ga) thin film sample. (i)–(ii) cross-sectional TEM and HRTEM images near the ZnO:(Mn,Ga)-sapphire substrate interface region, and (iii) HRTEM image of the ZnO:(Mn,Ga) thin film. Top inset: fast Fourier transform pattern. Bottom inset: diffraction pattern. (d) Magnetic field dependent magnetization at 300 K of four ZnO:(Co,Ga) thin film samples with electron carrier concentration of 8.3×1019 cm−3 (Co-A), 3.6×1019 cm−3 (Co-B), 4.0×1018 cm−3 (Co-C), and 1.3×1018 cm−3 (Co-D). Top inset: magnetic field dependent magnetization of sample Co-A at 700 K. Bottom inset: temperature-dependent magnetization of sample Co-A at 0.2 T. A high-temperature extrapolation fitting indicates the Curie temperature (T C) is around 950 K. (e) The plot of the saturated magnetization (M S) versus the electron carrier concentration (n) of the four ZnO:(Co,Ga) thin film samples shown in (d). Top inset: HRTEM image of a ZnO:(Co,Ga) thin film. The scale bar is 2 nm. Bottom inset: cross-sectional HRTEM image near the ZnO:(Co,Ga)-sapphire substrate interface region. The scale bar is 3 nm. (f) Magnetic field dependent magnetization (blue) and anomalous Hall effect (black) in (i) ZnO:Co and (ii) ZnO:(Co,Ga). The ZnO:Co and ZnO:(Co,Ga) thin films show electron concentration (n) of 4.0×1019 and 5.7×1020 cm−3, saturated magnetization (M S) of 0.35 and 0.85 μ B/Co2+, and anomalous Hall conductivity (σ AHE) of 0.002 and 0.078 Ω−1 cm−1, respectively. [(a)–(c) reprinted with permission from Refs. [95, 105]; (d)–(e) reprinted with permission from Ref. [94]; and (f) reprinted with permission from Ref. [99]. Copyright the American Institute of Physics]

Full size image

In Dietl’s theory [16], ZnO DMS is predicted to have an above-room-temperature Curie temperature when doped with 5 % Mn and within a 3.5×1020 cm−3 hole concentration p-type environment. In early theoretical work by Sato and Katayama-Yoshida, it was predicted that the ZnO:Mn tends to be ferromagnetic in p-type environment while ZnO:Co tends to be ferromagnetic in n-type environment [17, 108]. Both consistent [42, 47, 51] and inconsistent (ferromagnetic ZnO:Mn in n-type environment) [30, 35, 55] experimental results have been reported. Further efforts need to be focused on ZnO DMS to clarify the carrier type mediation effect.

3.2 Electrostatic doping

Compared to the chemical doping approach, electrostatic doping is a ‘cleaner’ way to tune the carrier concentration when studying the magnetic properties of DMS materials, due to no additional complexity introduced from the dopants. For example, it has been argued that Ga and Al (donors of ZnO) are likely to introduce additional unknowns to the magnetic properties of ZnO DMS materials because Ga and Al may also form compounds with the magnetic dopants (e.g. Mn, Co, etc.) in addition to their role of tuning electron carrier concentration in ZnO. However, additional fabrication process required for electrostatic doping such as dielectric layer and gate electrode metal deposition takes more effort compared to chemical doping. In ZnO DMSs, only a couple of electrostatic doping studies [92, 98] have been reported so far on investigation of carrier concentration dependent magnetic properties. Gate-bias-dependent magnetization [92] and anomalous Hall effect [98] were reported in ZnO:Co DMS thin films.

4 Ferromagnetism in undoped and non-transition-metal-doped ZnO

Whether the origin of the ferromagnetism is from the secondary phases from the magnetic dopants is a long debate in ZnO DMS. Different from ferromagnetic transition metals Fe, Co, and Ni, transition metals V, Cr, and Mn as well as Ti and Cu do not show any ferromagnetism in their own elemental forms; hence, these transition metals were initially considered as better dopants for ZnO DMS research. However, due to the multi-valence states, ferromagnetism in non-ferromagnetic nanoparticles, or their oxide compounds, recently it has been argued again that dopants V, Cr, Mn, Ti, and Cu in ZnO may introduce complexity and are not ideal dopants when studying the origin and mechanism of the ferromagnetism in ZnO DMS. Recent observations of ferromagnetism in non-transition-metal (e.g., C, Ga, Li, B, etc.)-doped ZnO and undoped ZnO pave an additional way to study the origin and mechanism in ZnO DMS. These results are discussed in this section.

The ‘unexpected’ magnetism in undoped ‘non-magnetic’ oxide materials was first observed in HfO2 thin films by Coey’s group [109, 110]. The ferromagnetism was attributed to the defects such as oxygen vacancies. Since then, observation of ferromagnetism has been discovered in various undoped oxide materials, including CeO2 [111], Al2O3 [111], In2O3 [111, 112], SnO2 [111], and TiO2 [112, 113]. Recently, ferromagnetism has been reported in undoped and non-transition-metal-doped ZnO [100, 111, 114–122] such as C-doped ZnO (ZnO:C) [115, 119], and Ga-doped ZnO (ZnO:Ga) [118]. The representative results of room-temperature ferromagnetism reported in undoped ZnO and non-transition-metal-doped are summarized in Table 3. The origins of the ferromagnetism in these studies are mostly attributed to the intrinsic defects (e.g. oxygen vacancy), defect complex, or mediation of the intrinsic defects in ZnO, except for ZnO:C, in which the origin of ferromagnetism was attributed to Zn–C system [115]. The defect-mediated mechanism of ferromagnetism in ZnO DMSs is different from the carrier-mediation or bound magnetic polaron [45] mechanisms, but it suggests that a precise control of defects in ZnO is also likely to be an alternative approach to design the magnetic properties in ZnO DMSs.

Table 3 Ferromagnetism in undoped and non-transition-metal-doped ZnO diluted magnetic semiconductors
Full size table

Xu et al. studied the magnetic properties of undoped ZnO thin films [117]. The ZnO thin films were grown on a-plane sapphire substrates using PLD at 570 °C in N2 ambient under a pressure of 0.3 mbar. The ZnO thin film shows above-room-temperature ferromagnetism. The Curie temperature is above 300 K indicated from the field-cooled and zero-field-cooled curves [117]. Figure 3a (i) shows the magnetic field dependence of the magnetization of the ZnO thin film at 5 K and 290 K with a ferromagnetic hysteresis loop observed at both temperatures. Figure 3a (ii) shows the anomalous Hall resistivity of the ZnO thin film at various temperatures of 5, 20, 50, and 290 K. Two additional interesting observations were reported [117]. First, another undoped ZnO thin film grown at the higher temperature of 655 °C (with all other growth parameters the same) did not show an evident ferromagnetic hysteresis loop, which is attributed to fewer amount of defects in the undoped ZnO at higher temperature. Second, Cu-doped ZnO thin films grown under the same growth conditions as the ferromagnetic undoped ZnO did not show ferromagnetism in the magnetization measurement; however, they unexpectedly showed an evident AHE in the transport measurements. The authors then concluded that the AHE should not be used as an indication of intrinsic ferromagnetism. The AHE [123] was previously considered as a sign for intrinsic ferromagnetism in ZnO DMS materials [68, 94, 98, 99, 103, 105, 115, 124–131], but now it is more widely accepted that the mechanism of AHE determines whether it is intrinsic (e.g. intrinsic deflection mechanism) or extrinsic (e.g. side jump mechanism, skew scattering mechanism) [123].

Fig. 3
figure 3

Ferromagnetism in transition-metal-free ZnO. (a) Room-temperature ferromagnetism in undoped ZnO thin films. (i) Magnetic field dependence of the magnetization of the ZnO thin film at 5 K and 290 K. Ferromagnetic hysteresis loop observed at both temperatures. (ii) Anomalous Hall resistivity of the ZnO thin film at 5, 20, 50, and 290 K. (b) Room-temperature ferromagnetism in ZnO:C thin films. (i) The temperature dependence of the normalized saturation magnetization [M S(T)/M S(5 K)] for ∼1 % (solid circles) and ∼2.5 % (open circles) ZnO:C thin films. Inset: magnetic field dependent magnetization of the ∼1 % ZnO:C thin film at 300 K with a hysteresis loop observed. (ii) Field-dependent Hall effect of the ∼1 % ZnO:C thin film at 10, 80, and 300 K. Inset: anomalous Hall effect curve after subtraction of the linear background from the normal Hall effect. [(a) reprinted with permission from Ref. [117]. Copyright the American Institute of Physics. (b) reprinted with permission from Ref. [115]. Copyright the American Physical Society]

Full size image

Pan et al. studied the magnetic properties of ZnO:C thin films [115]. The ZnO:C thin films were grown on c-plane sapphire substrates using PLD at 400 °C and a pressure <10−7 torr. The ZnO:C samples show above-room-temperature ferromagnetism. Figure 3b (i) shows the temperature dependence of the normalized saturation magnetization [M S(T)/M S(5 K)] for ∼1 % (solid circles) and ∼2.5 % (open circles) ZnO:C thin films. The inset in Fig. 3b (i) shows the magnetic field dependent magnetization of the ∼1 % ZnO:C thin film at 300 K with a hysteresis loop observed. The Curie temperature is higher than 400 K indicated by the temperature dependent saturation magnetization curve shown in Fig. 3b (i). The undoped ZnO control sample shows non-magnetic properties. In the two ZnO:C samples under study, the larger C concentration sample (∼2.5 % measured by secondary ion mass spectroscopy) shows a stronger magnetization (7.1 emu cm−3 at 300 K) than the smaller C concentration sample (∼1 %; 5.1 emu cm−3 at 300 K). No ferromagnetism was observed in carbon powder, sintered ZnO/C targets, and the sapphire substrate. The AHE was also observed in the ZnO:C samples with a consistent shape of hysteresis loop with the magnetization hysteresis loop. Figure 3b (ii) shows the field-dependent Hall effect of the ∼1 % ZnO:C sample at 10, 80, and 300 K. The inset in Fig. 3b (ii) shows the AHE curve after the linear background (from the normal Hall effect) was subtracted. Density functional theory (DFT) calculations based on a supercell with 18 f.u. wurtzite ZnO were employed to discuss the experimental results. The magnetism in ZnO:C arises from the carbon substitution for oxygen forming a Zn–C system in the ZnO environment. The magnetic moment is 2.02 μ B per carbon dopant predicted by the DFT calculation, which is in good agreement with the experiments.

Room-temperature ferromagnetism has also been reported in Li-doped ZnO [49, 132–135] and B-doped ZnO [136, 137], with the origin of the ferromagnetism attributed to the cation vacancy [132] and the induced magnetic moment of oxygen atoms in the nearest-neighbor sites to B–Zn vacancy pairs [136], respectively.

5 Concluding remarks and outlook

Above-room-temperature ferromagnetism has been widely observed and reported in ZnO DMS materials so far. Currently the centerpiece of ZnO DMS research is to further clarify the origin and mechanism of the ferromagnetism in ZnO DMS materials. To investigate the role of free-carrier concentration in the magnetic properties of ZnO DMS materials is an indispensable approach in this direction. Different from the ‘canonical’ DMS material GaAs:Mn [2, 138] (in which Mn itself is not only a magnetic dopant but an acceptor dopant introducing holes as well), in ZnO DMS materials the magnetic dopants such as Mn and Co do not provide additional electrons or holes in significant amounts, because all the common magnetic dopant elements are generally not shallow donors/acceptors in ZnO at all. Hence, in order to tune the carrier concentration in ZnO DMS using the chemical doping approach, a second dopant element in addition to the magnetic dopant element is required such as Ga [94, 95, 99] and Al [91]; however, the donor/acceptor dopants introduce additional complexity at the same time of supplying free carriers for the mechanism analyses, because the role of the free carriers introduced by the donor/acceptor dopants has to be clearly distinguished from other possible contributions from the donor/acceptor dopants themselves, especially when these dopants compound with magnetic dopants or introduce defects in ZnO showing evident magnetic property changes. For example, ferromagnetism has been reported in ZnO:Ga (without any magnetic dopants) recently [118], as discussed in the previous section. We need to pay additional attention to make a conclusion that the magnetic property is dependent on the free-carrier concentration rather than the concentration of the donor/acceptor dopants. Additionally, to achieve a reliable p-type environment in ZnO with controllable free-hole concentration is also indispensable for the carrier mediation mechanism study of ZnO DMSs, which is still a challenge using the chemical doping method. Most of the free carrier concentration dependent magnetic property studies in ZnO DMSs are under electron environment so far.

Compared to the chemical doping method, electrostatic doping is a ‘cleaner’ approach to tune the carrier concentration in a semiconductor material controllably with the applied gate voltages. Although preliminary results on carrier concentration dependent magnetization [92] and anomalous Hall effect [98] under electrostatic doping have been reported in ZnO DMS materials, a lot more work still needs to be done to reach solid conclusions comparable to the achievements from similar experiments in ‘canonical’ DMSs GaAs:Mn [2, 9, 10, 139, 140] and InAs:Mn [11, 12]. Furthermore, how the Curie temperature changes with carrier concentration has not yet been carefully studied in ZnO DMS materials. No experimental results showing a clear conclusion have been reported so far to the best knowledge of the author. These studies under electrostatic doping are important to ZnO DMS research and worthwhile to be focused on in the near future. Recent discovery of the ionic-liquid gating technique [141–146] paves the way for a novel electrostatic doping approach with orders-of-magnitude higher free-carrier concentration achievable than the electrostatic doping using conventional solid gate-dielectrics. Exciting results have been reported in several DMS materials using ionic-liquid gated electrostatic doping recently [144, 145]. Studies on ionic-liquid-gated ZnO DMS devices are likely to further clarify the carrier-mediation mechanism therein due to the opportunity of a larger range of tunable free-carrier concentration.

Finally, it is necessary to point out here that ZnO is a semiconductor material with many species of native defects and impurities with quite a few of them behaving as shallow donors [13, 147, 148]. Quite different from classical semiconductors such as Si and GaAs, ZnO materials prepared using various growth methods from different groups generally show significant differences in physical properties including native defect/impurity species and density, mobility, crystallinity etc. Sometimes the difference could be as large as many orders of magnitude; for example, the electron concentration in undoped ZnO prepared by various methods can range as large as from 1015 cm−3 to 1020 cm−3. When these ZnO materials are employed for DMS study, it is not a surprise that distinct results are reported even with similar magnetic doping conditions. It is a significant difficulty hindering the progress in further clarification of the controversial issues in ZnO DMS research. Once high-quality undoped ZnO materials with low density of native defects and impurities, showing comparable background electron concentration to bulk crystals [149–154], as well as negligible physical property difference from one group to the other within the research community, can be widely prepared in the future, a more reliable platform and basis for ZnO DMS research is established. Research results of ZnO DMS achieved based on this platform can be more reasonably and reliably compared from one group to the other for analyses.

It is still a long way to go with tremendous efforts towards practical device application of the above-room-temperature ferromagnetism demonstrated in ZnO DMS materials, although reported preliminary results of some prototype devices such as ZnO DMS magnetic tunnel junctions [155] have shown promising interest and potential application.