1 Introduction

One-dimensional (1-D) nanomaterials have drawn a lot of interest due to their novel and unique properties and a large number of potential applications such as nanowire field-effect-transistors, nanolasers and nanogenerator [1, 2]. ZnO nanomaterials are among the most important one-dimensional nanomaterials due to their semiconducting, piezoelectric, and biocompatible properties. ZnO nanomaterials have been found to have a wide range of morphologies such as nanorods, nanotubes, nanobelts and nanorings and the shape and size of such nanostructures play an important role for the performance of the devices [3, 4]. There has also been a great deal of attention to ZnO material because of its prospects in optoelectronic applications including chemical sensors, solar cells and optoelectronic devices owing to its direct wide band gap of 3.37 eV at room temperature with a large exciton binding energy (60 meV) [5]. In recent years, ZnO has also become an important material in the search for high Curie temperature (T c) diluted magnetic semiconductor (DMS) after Dietl’s prediction on ZnO:Mn showing ferromagnetism above the room temperature [6]. The presence of room temperature ferromagnetism in V, Cr, Fe, Co, or Ni-doped ZnO systems was theoretically demonstrated by Sato and Katayama-Yoshida [7]. Among all the transition-metal-ions-doped ZnO, especially Mn-doped ZnO, studies have drawn much attention due to the highest magnetic moment of Mn as well as the occupied first half of Mn d band, which forms a stable polarized state [8]. However, there are some experimental works reported on ZnO:Mn with various morphologies, such as bulk [9], thin films [10] and nanoparticles [11], where the magnetism is very sensitive to preparation conditions such as synthesis method, doping content, sintering temperature and annealing environment. For example, Zn0.99Mn0.01O bulk samples prepared by a standard solid-state reaction method exhibited room-temperature ferromagnetism when sintered at a temperature of 500 C; however, the samples sintered at a higher temperature of 900 C displayed a linear response, showing a paramagnetic behavior [12]. Hou et al. showed that the magnetization of Mn-doped ZnO thin films decreased with annealing in oxygen atmosphere, whereas annealing in vacuum gave rise to increase of the magnetization. They explained their data based on the hypothesis that annealing treatments in vacuum increase the number of defects, leading to an enhancement in ferromagnetism, whereas annealing in oxygen leads to almost all defects to disappear, concluding the disappearance of ferromagnetism [13]. According to the study made by Wang et al. on ZnO:Mn nanoparticles grown by an ultrasonic assisted sol–gel process, ferromagnetic ordering increases with increase in Mn concentration up to 2 at.%, while for 5 at.%, ferromagnetism is suppressed and a large paramagnetic effect appears [14] Some studies have also claimed that ferromagnetism in Mn-doped ZnO originates from impurities (manganese oxide or precipitation of secondary phases) or the replacement of Zn2+ by Mn2+ in the ZnO host matrix, while other groups have reported that oxygen vacancies cause ferromagnetic ordering in oxide based DMS [1517]. Thus there is a wide variance between the data reported for samples grown by different methods and for different nano-morphologies. The influence of strain, defects and impurities is very hard to separate in many of the reported studies and thus there is a clear need for studies of Mn-doped high quality (crystalline and optical) ZnO materials.

In the literature, there are diverse techniques to produce transition-metal-doped ZnO materials. These methods include pulsed laser deposition [18], rf magnetron sputtering [19], spray pyrolysis [20], sol–gel [21] and vapor phase transport method (VPT) [22]. Of all these methods, VPT method is much used due to the simple and low-cost equipment and it has been utilized to grow ZnO nanostructures of diverse morphology with excellent crystalline and optical quality [23]. To our knowledge, there are not many studies on ZnO:Mn nanorods grown by VPT method. Our intention is therefore to obtain 1-D and vertically-aligned ZnO-based DMS by this method. In this sense, ZnO nanorods were grown on ZnO buffer layer coated Si substrates by VPT technique and Mn metal was then diffused into these nanorods at three different annealing temperatures in vacuum to achieve DMS with a T c above the room temperature. In addition, we report a relationship between XPS results and magnetic properties of these samples to clarify the origin of the observed room temperature ferromagnetism in our ZnO:Mn nanorods.

2 Experimental details

Vertically aligned ZnO nanorods were deposited on ZnO buffer layer coated Si substrates by a VPT method where ZnO buffer layer was composed of ZnO seeded Si substrates and subsequent CBD (chemical bath deposition) ZnO growth. All details of three stages of growth procedure can be found in [22, 24].

After the nanorods growth, Mn metal was evaporated on the ZnO nanorods by thermal evaporation (Leybold Univex 350) system with a pressure of ∼6×10−6 Torr. The amount of manganese deposited onto ZnO nanorods was monitored by a thickness monitor (Inficon XTM/2) and its thickness was adjusted to ∼5 nm. After the Mn deposition, the samples were annealed in a quartz tube at temperatures going from 500 to 700 C for 8 h with a step of 100 C in a vacuum of ∼10−2 Torr. The crystal structure of the samples was carried out using a Bruker AXS D8 diffractometer with CuK α radiation in the range of 2θ=20–60 with a step of 0.01. The surface morphology and composition were performed with a Zeiss EVOLS 15 scanning electron microscope (SEM) attached with energy dispersive X-ray spectroscopy (EDS). For all SEM and EDS measurements, an acceleration voltage of 20 kV was used. The chemical composition and bonding at the surface were characterized by using X-ray photoelectron spectroscopy (XPS) with Al Kα radiation (1486.6 eV). The C 1s photoelectron peak at 285.0 eV was used as a reference for the charge-correction of binding energies of core level peaks. Magnetization measurements of the samples (MH and MT) were conducted using a Quantum Design Physical Property Measurement System (PPMS) with a vibrating sample magnetometer (VSM) module.

3 Results and discussion

XRD patterns of undoped and ZnO:Mn nanorods annealed at 500, 600 and 700 C for 8 h in vacuum are illustrated in Fig. 1(a)–(d). Only diffraction peaks from wurtzite ZnO (002) planes were observed, showing a highly c-axis preferred orientation perpendicular to the substrate, which is in good agreement with the SEM results shown later. There was not detected any elemental manganese and its oxides such as MnO, Mn2O3, Mn3O4 and ZnMn2O4 in XRD pattern of all the samples, indicating that Mn doping did not significantly alter the wurtzite ZnO structure. The lattice parameter value of c was calculated from the (002) peak of ZnO centered at 34.43 and was found to be 5.21 Å. Within the resolution limit of our XRD system, compared to the undoped sample, there were no detectable peak shifts for the ZnO:Mn samples annealed at 500, 600 and 700 C for 8 h. Furthermore, some diffraction peaks in the pattern can be indexed to the (200), (013), (042) and (240) planes of Zn2SiO4, in accordance with JCPDS card no. 24-1469, meaning that reactions between the Si, SiO2 and ZnO may occur during both the heat treatment of the ZnO seed layer and (more probably) the VPT process conducted at high growth temperature [25].

Fig. 1
figure 1

XRD patterns of nominally undoped ZnO (a) and ZnO:Mn nanorods annealed at 500 C (b), 600 C (c), 700 C (d) for 8 h

X-ray rocking curve (XRC) measurements were performed to search for the crystallinity of both undoped and ZnO:Mn nanorods. Figure 2(a)–(d) shows the XRC data measured between 0 and 30 from the ZnO (002) diffraction peaks for the same samples displayed in Fig. 1. It is seen that the full-width-at-half-maximum (FWHM) value of the undoped ZnO sample is 3.36, implying good alignment of the nanorods perpendicular to the substrate surface. Wang and co-workers obtained a similar FWHM value for their Cu-doped ZnO thin films grown by magnetron sputtering method [26]. On the contrary, the study reported by Yanmei and co-workers showed that undoped ZnO nanorods had a larger FWHM value of 10.4 for hydrothermal synthesized ZnO:Ni samples [27]. Upon Mn evaporation together with annealing, the intensity of the ZnO (002) peak decreases gradually with the increase of annealing temperature with respect to the undoped ZnO. Additionally, the XRC FWHM values of ZnO nanorods increase gradually and reach 6.62 for ZnO:Mn nanorods annealed at 700 C for 8 h, widely consistent with the changes in XRD peak intensities [28]. These results imply that Mn evaporation and the subsequent annealing process caused a broadening in the (002) peak width most probably due to slight alterations in the underlying ZnO buffer layer planarity which affects the alignment of the VPT-grown nanorods.

Fig. 2
figure 2

Rocking curves (of the (002) peak) of nominally undoped ZnO (a) and ZnO:Mn nanorods annealed at 500 C (b), 600 C (c), 700 C (d) for 8 h

The morphological properties of ZnO nanorods grown on ZnO buffer layer coated Si substrates were examined by SEM. Figure 3(a)–(c) illustrates SEM images of ZnO:Mn nanorods annealed at 600 C for 8 h. The plane and 45-tilted view images of the ZnO:Mn nanorods are respectively presented in Fig. 3(a) and (b), indicating that ZnO:Mn samples grew uniformly over a wide area of ZnO buffer layer coated Si substrate. The inset of Fig. 3(a) shows that ZnO:Mn samples had a hexagonal shape with a diameter of ∼100 nm. A cross-sectional image of ZnO:Mn nanorods is given in Fig. 3(c), confirming that ZnO:Mn nanorods were vertically grown on ZnO buffer layer coated Si substrate and all nanorods nearly exhibited similar length of 4.0 μm.

Fig. 3
figure 3

SEM images of (a) top view, (b) 45 tilted view obtained from edge of sample and (c) cross section of ZnO:Mn nanorods annealed at 600 C for 8 h

EDS measurements were made to determine the chemical composition of both undoped and ZnO:Mn nanorods. To search for the homogeneity of the samples, elemental mapping was performed by using EDS. Figure 4(a) illustrates the SEM image of ZnO:Mn nanorods annealed at 600 C for 8 h. EDS mapping of Zn, O and Mn elements is respectively presented in Fig. 4(b)–(d), exhibiting a uniform distribution of all the elements in the sample. Furthermore, it was obtained that the atomic ratio of Mn/Zn+O had a maximum value of ∼1.7 at.% for ZnO:Mn annealed at 600 C for 8 h, whereas the atomic ratios of Mn/Zn+O of all the other ZnO:Mn samples exhibited slightly lower values, as listed in Table 1.

Fig. 4
figure 4

(a) An SEM image and EDS mapping of (b) Zn, (c) O, (d) Mn elements of ZnO:Mn nanorods sample annealed at 600 C for 8 h

Table 1 Actual atomic concentrations of Zn, O or Mn in nominally undoped ZnO and ZnO:Mn nanorods

Analysis of the valence bonding state of undoped and ZnO:Mn nanorods annealed at 700 C for 8 h was performed using XPS and is shown in Fig. 5. XPS measurements present evidence for the incorporation of doped ions into the host lattice and the nature of the incorporation. Figure 5(a) illustrates the full-range XPS survey spectrum and only peaks corresponding to Zn, Mn, O and C are identified. Figure 5(b) indicates the Mn 2p spectrum for ZnO:Mn nanorods annealed at 700 C for 8 h. From Fig. 5(b), the core levels of Mn 2p3/2 and 2p1/2 curves are fitted by Gaussian functions and peak values at 641.52 and 653.43 eV are found for Mn 2p3/2 and 2p1/2, respectively, revealing the presence of Mn2+ ions in the sample rather than metallic Mn (637.7 eV), Mn3+ (642.9 eV) or Mn4+ (645.0 eV) [29, 30], which is consistent with other reports [31, 32]. These XPS data, combined with the XRD results shown earlier, indicate that the Mn ions substitute on the Zn sites of the ZnO lattice. Figure 5(c) illustrates the asymmetric O 1s peak of undoped ZnO nanorods, which is deconvoluted into three sub-peaks at the binding energies of 530.68, 531.65 and 532.10 eV. The low binding energy component (labeled as OL) is ascribed to O2− ions in the wurtzite structure of the hexagonal Zn2+ ion array [33], whereas the high binding energy component (denoted as OH) is related to the presence of loosely bound oxygen on the surface of the ZnO sample belonging to a specific species, such as CO3, adsorbed H2O or O2 [34]. Another peak located at 531.65 eV (marked as OM) is associated with O2− in oxygen deficient regions within the ZnO matrix, showing the existence of some oxygen vacancies in the sample. The fitted contributions of OL, OM and OH peaks are presented in Fig. 5(c)–(d) and the areas of these fitted peaks are denoted by AL, AM and AH, respectively. For undoped ZnO nanorods shown in Fig. 5(c), the ratio of AM/(AL+AM+AH) is determined to be ∼0.21. However, in Fig. 5(d) this ratio increased to ∼0.36 for ZnO:Mn nanorods annealed at 700 C for 8 h, which is a clear indication of the growth of a significant population of Vo defects induced by Mn doping. A similar analysis for ZnO thin films was carried out by Can et al. [35].

Fig. 5
figure 5

(a) XPS survey spectra of nominally undoped and ZnO:Mn nanorod arrays annealed at 700 C for 8 h; (b) shows binding energy spectrum of Mn 2p and Gaussian fitting, (c) and (d) illustrate the binding energy spectra of O 1s with Gaussian fitting for undoped ZnO and ZnO:Mn nanorods, respectively, annealed at 700 C for 8 h

Magnetization loops of all ZnO nanorods were measured at 300 K using a VSM. The effect of Si substrate was eliminated to have an actual magnetization of the samples. The MH loop of a bare Si substrate is shown in the inset of Fig. 6, ascertaining that Si has pure diamagnetic behavior. Figure 6(a)–(d) shows the room temperature MH curves of undoped and ZnO:Mn nanorods annealed at 500, 600 and 700 C for 8 h in vacuum. It can be seen that both undoped and ZnO:Mn samples are ferromagnetic in nature. Ghosh et al. found a similar ferromagnetic behavior for nominally undoped ZnO in their K-doped ZnO nanowires [36]. The reason for the observed ferromagnetism in nominally undoped ZnO nanorods is probably due to the oxygen vacancy defect-induced d 0 ferromagnetism [37, 38]. The nominally undoped ZnO nanorods have a saturation magnetization (M s), remnant magnetization (M r) and coercive field (H c) values of 0.13 emu/gr, 0.02 emu/gr and 210 Oe, respectively. Data from an (otherwise identical) nominally undoped ZnO sample annealed at 700 C for 8 h in vacuum is also presented in Fig. 6(e) where the annealed nominally undoped ZnO nanorods have M s, M r and H c values of 0.13 emu/gr, 0.03 emu/gr and 197 Oe, respectively, indicating that annealing in vacuum alone does not markedly change the M s value of the nanorods with respect to unannealed nominally undoped ZnO. However, after annealing the ZnO:Mn nanorods at 500 and 600 C in vacuum, it was noted that M s values increased to 0.16 and 0.18 emu/gr, respectively. With a further increase in annealing temperature to 700 C, a maximum M s value of 0.19 emu/gr was obtained for ZnO:Mn samples, confirming the definite role of Mn doping in enhancing the ferromagnetism. As compared to unannealed nominally undoped ZnO, the ZnO:Mn samples annealed at 500, 600 and 700 C for 8 h have larger M s values, and specifically the ZnO:Mn sample annealed at 700 C for 8 h has a larger M s value than that of the annealed nominally undoped sample. In all the annealed samples, based on the discussion above there is an appreciable concentration of intrinsic defects. The existence of a high population of intrinsic defects like oxygen vacancy (V o) in the structure, in addition to the Mn doping, causes an increase in the number of bound magnetic polarons (BMPs). According to the BMP theory, magnetic cations (Mn2+), electronic carriers and intrinsic/native defects can form BMPs. An electron associated with a native defect such as an oxygen vacancy will be confined in a hydrogenic orbit of radius rH to form a hydrogen atomic-like structure. BMPs can be formed by the exchange interaction between such electrons and magnetic cations (Mn2+) residing within the orbit radius (r H) of electrons. Ferromagnetism results when the BMPs begin to overlap to form a continuous chain throughout the material. Therefore, a greater density of native defects in conjunction with Mn doping helps to produce more BMPs which yields a greater overall volume occupied by BMPs, leading to the overlap of BMPs, resulting in ferromagnetism [39]. Liu et al. grew Mn-doped ZnO nanowires via chemical vapor deposition at substrate temperatures of 750, 850 and 950 C. Paramagnetic behavior was observed for the samples grown at 750 and 850 C due to a formation of ZnMn2O4, whereas it was found that ZnO:Mn nanowires synthesized at 950 C exhibited room temperature ferromagnetism, attributed to a greater substitution of Mn2+ into the ZnO matrix, leading to ferromagnetic ordering with an M s of 0.25 emu/gr [40]. In our study, compared with nominally undoped (annealed and unannealed) and all the other ZnO:Mn nanorods, ZnO:Mn annealed at 700 C for 8 h has the lowest M r values of 0.01 emu/gr and H c of 36 Oe which are indicative of a small hysteresis area, meaning that the sample displays soft ferromagnetic behavior [16]. We note that in general the observed room temperature ferromagnetism in ZnO:Mn nanorods could also possibly arise from other sources such as inclusions of impurity clusters/phases. In this regard, Mn, MnO, MnO2 and Mn2O3 phases are well-known to be antiferromagnetic with Neel temperatures of 100, 116, 92 and 76 K, respectively, whereas Mn3O4 is ferromagnetic with a Curie temperature of 43 K [13, 41]. However, the origin of ferromagnetism in our samples is not due to the existence of manganese oxides and clusters because these secondary phases were not observed in the ZnO:Mn nanorods either by XRD or by XPS measurements.

Fig. 6
figure 6

Room temperature MH curves of nominally undoped ZnO (a), ZnO:Mn nanorods annealed at 500 C (b), 600 C (c), 700 C (d) for 8 h, and nominally undoped ZnO nanorods annealed at 700 C for 8 h (e). Inset shows the MH loop of a bare Si substrate

Among all the ZnO:Mn nanorod samples, the maximum M s value of 0.19 emu/gr was obtained for Mn diffusion-doped ZnO annealed at 700 C for 8 h, corresponding to the 0.27 μ B/Mn. However, this value is significantly smaller than theoretical value (5.92 μ B/Mn) for a free Mn2+ ion with S=5/2 and g=2, suggesting either (i) that not all the Mn ions are effectively involved in BMP formation, or (ii) the presence of a competition between the antiferromagnetic and ferromagnetic interactions of closely spaced Mn ions. Due to the strong antiferromagnetic interactions between neighboring Mn ions, ferromagnetic ordering can be suppressed. Both these effects may contribute to the observation that the magnetization of Mn ions is much lower than that based on a simple theoretical estimate [42].

We note finally that in the literature some other possible models have been suggested to explain the origin of dopant-related ferromagnetism in transition-metal (TM)-doped ZnO including the double exchange mechanism proposed by Garcia et al., based on the co-existence of Mn3+ and Mn4+ oxidation states in ZnO:Mn [43] and the RKKY (Ruderman–Kittel–Kasuya–Yosida) model where the magnetism results from the exchange interaction between local spin-polarized electrons (such as electrons of Mn2+ ions) and conduction electrons, with the concentration of free carriers playing a crucial role in establishing the ferromagnetism. The double-exchange mechanism did not apply for our case due to the presence of Mn2+ in our ZnO:Mn nanorods. The RKKY mechanism also does not seem to be relevant in explaining the ferromagnetism in our Mn-doped samples due to the expected resistive nature of ZnO:Mn samples compared to undoped ZnO. Reports in the literature for ZnO samples grown by the VPT method (similar to our samples) reveal a carrier concentration in the order of 1017 cm−3 [44] and doping with Mn leading to a decrease in the carrier concentration and an increase in the resistivity. For example, Hong et al. synthesized Mn-doped ZnO thin films by PLD and observed that Mn doping decreased the carrier concentration and increased the resistivity of ZnO films [45]. In addition, Lin et al. produced Mn-doped ZnO thin films by plasma-assisted molecular beam epitaxy and found that the carrier concentration of undoped ZnO (3.23×1018 cm−3) decreased with an increase of Mn doping to 9.5 % (2.14×1015 cm−3) [46].

Based on the entirety of the discussion above we believe that the observed excess room temperature ferromagnetism in the ZnO:Mn samples (compared to annealed nominally undoped samples) is due to Mn-related BMP formation, as explained above. The BMP model is consistent with our XPS results and leads us to conclude that our ZnO:Mn nanorods include a large amount of intrinsic defects (like V o) induced by the Mn doping, which give rise to favorable conditions for the formation of Mn-related BMPs, and hence to room temperature ferromagnetism.

In order to determine the Curie temperature (T c), the magnetization versus temperature (MT) was carried out in the temperature range of 5–300 K at an applied magnetic field of 500 Oe for ZnO:Mn nanorods annealed at 700 C for 8 h, and the data are indicated in Fig. 7. From these data it can be seen that the magnetization was significantly reduced with an increase of the temperature up to 300 K. It is well known that the Curie temperature corresponds to the point where the magnetization drops to zero and the material goes from magnetically ordered to disordered. In light of this it can be said that the Curie temperature for our sample is above the room temperature; however, it is not possible to determine the exact value because it goes beyond the range of our experimental capabilities. Some studies showed that Mn-doped ZnO indicates ferromagnetism with Curie temperatures above room temperature and our result is consistent with these reports [13, 47].

Fig. 7
figure 7

Temperature dependence of magnetization of ZnO:Mn nanorods annealed at 700 C for 8 h

4 Conclusions

In summary, the following points can be concluded from our studies: (i) XRD and SEM results showed that all the samples had wurtzite structure with a well-aligned nanorod morphology; (ii) the XPS results indicated that Mn ions successfully substituted for Zn ions in the lattice and ZnO:Mn nanorods also had a large concentration of native defects (oxygen vacancies) compared to undoped ZnO; (iii) magnetic measurements illustrated that undoped and ZnO:Mn samples showed the room temperature ferromagnetism; (iv) comparing magnetic data with the XPS data, it is concluded that the observed excess ferromagnetism in ZnO:Mn nanorods can be related to the formation of Mn-related BMP, associated with an exchange interaction between Mn2+ ions and oxygen vacancy defect-bound carriers.