Electronic and relating behavior of Mn-doped ZnO nanostructures: An x-ray absorption spectroscopy study

Modern electronic components are steadily decreasing in size to increase efficiency; however, there is a quantum mechanical limit in size as components near nanoscale dimensions. Spintronics (spin + electronics) has emerged as an alternative technology that uses the electron spin rather than its charge as a new way to read and write information during storage and processing. Dilute magnetic semiconductors (DMSs), which favorably combine both ferromagnetic and semiconducting properties, have attracted considerable interest for their potential applications in spintronic devices ever since Dietl et al. theorized that room temperature ferromagnetism (FM) in wide-bandgap semiconductors might exist.¹ Much of the earlier effort toward DMS spintronic devices has focused on the synthesis and characterization of transition metal (TM)-doped II–VI materials, with prominent interest in Mn-doped ZnO, henceforth denoted as Mn:ZnO, due to a proposed and measured Curie temperature above room temperature. However, much controversy surrounds this system as some researchers have observed both paramagnetism (PM) and anti-FM in Mn:ZnO² in addition to both the low temperature (∼45 K) and the room temperature FM previously observed by others.^3,4 Furthermore, it is still unclear whether the FM is intrinsic (carrier-induced) or due to Mn-related secondary phases.^5,6 The disagreements arise, at least in part, from incomplete characterization of the system in terms of morphology and crystallinity. Mn:ZnO systems with the same composition are likely to show different behavior if their morphology and crystallinity vary. While the magnetic behavior of Mn-doped ZnO is continually being pursued,^7,8 interest in Mn-doped ZnO has shifted in recent years to its photocatalytic and relating functionalities pertaining to electron–hole separation following photoexcitation, such as photocatalysis,^9–15 photodegradation of organic molecules,¹⁶ and anti-bacterial¹⁷ properties, among others. Doping may facilitate electron–hole separation and, hence, their reactivity for catalytic processes. Due to vast interest, it is of great importance to explore the relationship between the local environment around Mn and the electronic structure, optical properties, and magnetic properties of the Mn:ZnO system.

In this work, we report the preparation of a series of Mn:ZnO nanostructures using a sol–gel method followed by annealing; their structure and morphology, the x-ray absorption near-edge structure (XANES) at the Mn and O K-edge and the Zn and Mn L_3,2-edges, as well as x-ray excited optical luminescence (XEOL), and magnetic properties were investigated. These experiments will provide considerable information on the electronic structure of the system, such as the oxidation state and coordination of the Mn dopant, as well as how Mn doping affects the overall electronic, optical, and magnetic behavior of the system.

XANES and XEOL are synchrotron techniques. XANES tracks the modulation of the x-ray absorption coefficient at an absorption edge of an element in a chemical environment relative to that of a free atom, of which the photon energy dependent absorption coefficient is monotonic. The modulation in the absorption coefficient arises from core to unoccupied state dipole transitions. The photoelectron probes the neighboring atoms via backscattering; the interference of the outgoing and backscattered electron wave at the absorbing atom produces oscillatory modulations in the absorption coefficient unique to the local structure (identity of the neighboring atom, interatomic distance, local symmetry, and dynamics). Therefore, XANES is sensitive to the local structure of the atom of interest.

XEOL monitors the energy transfer by materials that absorb x rays and convert them to optical photons (near IR to ultraviolet).^18–20 The process is largely based on the inelastic scattering (energy loss) of the photoelectron and Auger electrons resulting from x-ray absorption. The cascade is known as the thermalization of energetic electrons, resulting in electrons at the bottom of the conduction band and holes at the top of the valence band. The recombination of the electrons and holes radiatively via excitons or defect energy levels in the bandgap will produce optical photons associated with the energy of the optical bandgap and defect states, respectively. The branching ratio of the bandgap vs defect emission can often be used to evaluate the crystallinity of the materials of interest.^18,19

Zn_1−xMn_xO nanostructures were prepared by sol–gel synthesis, and a schematic of the process is illustrated in Fig. S1. Stoichiometric amounts (x = 0, 0.01, 0.03, and 0.10) of zinc acetate and manganese acetate (TMs totaling 0.009 mol) along with citric acid (0.009 mol) were dissolved in 30 ml of water and 80 ml of 30% ammonium hydroxide. The mixture was stirred for 6 h to make a sol, which was then placed in a drying oven for 16 h at 90 °C to form the gel. The gel was placed in a high temperature furnace at 400 °C for 4 h where it was cracked to produce the nanopowder as observed under an electron microscope (see below). The nanopowder was then divided into three portions: one portion was heated for an additional 2 h at 600 °C, the second was heated for 2 h at 800 °C, and the third was left as is. ZnO nanoparticles with 0%, 1%, 3%, and 10% Mn were prepared.

The SEM images are displayed in Fig. S2, which show nanoparticles with a clear variation in the particle size ranging from ∼5 to 100 nm. This variation in the particle size is shown to be temperature dependent rather than dopant dependent. The Mn:ZnO samples processed at 400 °C have a very small particle size of ∼5–10 nm, whereas the samples processed at 800 °C are 25–100 nm in size. Upon annealing at 800 °C, it is evident that agglomeration and ripening of the particles occur, resulting in larger particle sizes. There is no distinguishable difference between the 1% and 10% Mn samples processed at the same temperature, which indicates that the dopant concentration does not affect the particle size (Fig. S1).

The x-ray powder diffraction patterns are shown in Fig. S3 where we see that hcp ZnO is the only phase at low Mn concentrations, from 0% to 3%, and at low temperature annealing. Conversely, there are at least two phases present in the Zn_0.97 Mn_0.03O samples at 800 °C as well as in the Zn_0.90 Mn_0.10O samples at 600 and 800 °C. All the nanoparticles contain a wurtzite ZnO (a = 3.25 Å and c = 5.21 Å) phase (JCPDS PDF file no. 01-074-9939) and the secondary phase, when it is present, is of cubic ZnMnO₃ (a = 8.35 Å) (JCPDS PDF file no. 00-019-1461) denoted by the black square. The intensity of the peaks associated with the secondary phase increases with increasing processing temperature and dopant concentration although it remains a minor component. The ZnMnO₃ impurities are a common issue in the synthesis of Mn:ZnO and have been detected by other studies.^21,22 A comparison of the XRD spectra of Mn:ZnO processed at 400, 600, and 800 °C reveals a drastic broadening in the peaks as the processing temperature is decreased. This peak broadening is a common effect in XRD due to a decrease in the particle size.²³ This is tracked with the Scherrer equation,²⁴ showing that the trend is consistent with the SEM observation.

Mn L_3,2-edge, O K-edge, and Zn L_3,2-edge measurements were made on the undulator-based SGM beamline 11ID-1 of the Canadian Light Source (CLS). XANES spectra were recorded in total electron yield (TEY) and fluorescence yield (FLY).²⁵ XEOL spectra were collected using a QE65000 Scientific grade spectrometer (Ocean Optics) and a network CCD camera (Axis 2120), allowing for the collection of the entire optical range (200–900 nm). Mn K-edge and Zn K-edge measurements were conducted at the CLS@APS, sector 20 of the Advanced Photon Source (APS). Spectra were obtained in transmission, total electron yield (TEY), or x-ray fluorescence yield (FLY) as appropriate.

Magnetic measurements, such as the hysteresis loops of selected specimens, were conducted using a SQUID (superconducting quantum interference device) at the Institute for Research in Materials (IRM) at Dalhousie University, Halifax, Canada.

Figure 1 shows the Mn and Zn K-edge XANES of the Mn:ZnO system of different compositions and phases. The Mn K-edge XANES arising from dipole transition from Mn 1s → 4p is shown in Fig. 1(a). A comparison of the Mn K-edge XANES with the ZnO K-edge XANES relative to the threshold (E_o = 0 eV) aligns the XANES features to the same energy scale above the threshold [Fig. 1(b)]. The threshold of Mn:ZnO is similar to that of MnO with Mn in a 2⁺ oxidation state [Fig. 1(c)]. The peak locations A, B, and C in Fig. 1(a) are identical for all the doped samples, indicating that they have a very similar environment albeit with some noticeable broadening at low temperature anneal due to disorder. With increasing processing temperature, the peaks sharpen, indicating an increase in crystallinity.²⁶ The Mn K-edge XANES of samples processed at 800 °C resembles that of the Zn K-edge of ZnO [Fig. 1(b)], showing that Mn is in a similar tetrahedral local environment and substitutes Zn in the ZnO lattice as (Mn) ZnO [Fig. 1(d)].

Previous studies of the Mn K-edge have suggested that Mn substitutes Zn and no secondary phases are present.^26–29 The small pre-edge feature at 6540 eV (peak A) is a result of a weak s → d quadrupole transition associated with p-d hybridization. This, in turn, is only possible if the Mn site does not contain an inversion center, as is present in a tetrahedral configuration such as (Mn)ZnO, or when there is sizable local distortion.³⁰ The pre-edge peak is present in all nanoparticle samples, but close examination shows a decrease in intensity at a processing temperature of 800 °C. This is an indication that a secondary phase, such as cubic ZnMnO₃, possibly forms as some of the Mn converts from a tetrahedral geometry to an octahedral geometry at a high processing temperature (800 °C). Since this new phase is a minor component, the XANES features are still dictated by the Mn presence in the substitutional site.

The Zn L_3,2 edge XANES probes the unoccupied s character in the conduction band since the Zn 3d is full, and the O K-edge probes the unoccupied state of O 2p character in the conduction band. The Zn L_3,2-edge XANES spectra for Zn_1−xMn_xO nanoparticles are shown in Fig. 2 where we see that there is no change in the absorption edge threshold or the location of features in the XANES spectra for systems with 1%–3% Mn at all annealing temperatures. This indicates that the ZnO crystal structure remains essentially intact upon doping. Again, the spectral features sharpen at higher temperature as the crystallinity improves.

However, while the fine structures at the Zn L_3,2-edge remain intact for the 1% sample, there is a noticeable deviation when the Mn concentration reaches 10% [Fig. 2(b)]. This change reveals an increase in the overall disorder of the system resulting from the formation of a secondary phase. The secondary phase contains Zn in the form of a ZnMnO complex, most likely ZnMnO₃ with an octahedral O environment as indicated by the XRD data (Fig. S3). This changes the local environment of some Zn sites in the system from a tetrahedral coordination in ZnO to an octahedral coordination in ZnMnO₃, resulting in the combinatorial Zn L_3,2-edge XANES spectra in Fig. 2(b).

Secondary phase formation is evident at the O K-edge XANES in Fig. 3. From Fig. 3(a), it is apparent that systems with 1%–3% Mn exhibit identical features to ZnO. The XANES spectra of 1% Mn-doped samples processed at 400 and 600 °C and the 3% Mn-doped sample at 400 °C [Fig. 3(a)] show virtually identical features to those of pure ZnO. They resemble the model spectrum obtained by FEFF8 calculations containing tetrahedrally substituting Mn²⁺ in Zn vacancies with additional Zn vacancies in the crystal lattice (MnZn + V_Zn) described in the literature.²⁸ 

A closer look reveals a small pre-edge shoulder at ∼532 eV for samples with the Mn concentration as low as 1% at 800 °C and 3% at 600 °C. This is a signature of Mn in the O_h environment.³¹ This feature grows in the 10% sample under high temperature treatment. Thus, the pre-edge shoulder at 532 eV is present in systems where Mn no longer has a purely substitutional (Zn vacancy) role and forms an oxide complex with a lower binding energy. From the location and shape of the pre-edge feature (Fig. 3), the O in the secondary phase is bonded to either Mn³⁺ or Mn⁴⁺ in the O_h environment via p-d hybridization with the metal d bands. This observation is consistent with the recently compiled system of O K-edge XANES of 3d metal oxide in an O_h environment exhibiting a broad or a split resonance (pending the occupancy of d hole counts) via transition to O 2p states hybridized with the e_g and t_2g 3d orbitals of the transition metal.³¹ The intensity of this pre-edge feature can be used as an indication for secondary phase formation. In this case, the ZnO wurtzite structure remains the dominant component.

The Mn L_3,2 edge is the most sensitive edge in exploring the local properties of Mn since dipole transition allows the probing of the unoccupied states of Mn 3d_5/2,3/2 character in the conduction band. Experimentally, the L-edge is also more chemically sensitive partly because of the high energy resolution of the beamline optic, E/ΔE ∼10 000 (5000 routinely, 0.13 at 640 eV), and partly because of the core–hole lifetime broadening of the Mn L₃-edge (0.2 eV), which is significantly narrower than that of the K-edge (1.12 eV).

Figure 4(a) shows the comparison of the Mn L_3,2-edge XANES of all samples of interest, including MnO, Mn₂O₃, and MnO₂ standards with d⁵, d⁴ and d⁵ mixed valence, and d³ configurations, respectively. Looking at the Mn L_3,2-edge XANES [Fig. 4(a)], it is apparent that while the 1% sample is Mn²⁺ like after 400 and 600 °C anneal, signs of secondary phase formation at 800 °C are evident as are in higher Mn concentrations at lower annealing temperature. Given the 1% Mn:ZnO Mn K-edge XANES as the representative of the MnO₄ tetrahedral moiety of Mn²⁺ in a ZnO matrix [Fig. 1(b)], spectral features differ from this, which is evidence for a secondary phase. Mn XANES in MnO is representative of a Mn²⁺ in an octahedral crystal field surrounded by oxygen [Fig. 4(b)], as is the 1% sample at 400 °C for the tetrahedral environment. It should be noted that the Mn 2p to 3d dipole transition is complicated by the presence of the crystal field and Jahn–Teller distortion, not to mention possible exchange interaction with the surrounding and the proximity of the 2p_3/2 and 2p_1/2 states in energy; thus, the number of peaks observed in the 2p to 3d transition is not always intuitive.

The secondary phase formation in the low Mn concentration systems appears to be temperature dependent as noted above: the 1% Mn samples processed at 400 and 600 °C do not show signs of secondary phases, which are, instead, only present at 800 °C [Fig. 4(a)]. A similar trend can be seen for the 3% Mn samples with secondary phase formation evident in the 600 and 800 °C samples but not in the 400 °C sample. The onset of an impurity phase seen in the Mn L_3,2-edge XANES is the appearance of Mn₂O₃ like features [Fig. 4(c)] although the Mn²⁺ signal is likely buried under a stronger Mn³⁺ signal. The Mn L_3,2-edge XANES trend correlates well with the increase in the pre-edge peak seen in the O K-edge XANES (Fig. 3).

From these results, we propose that Mn²⁺ is first incorporated into the ZnO substitutionally, replacing the Zn²⁺ ion as evident from Figs. 1(b), 3, and 4. However, its local structure is different from that of MnO (Mn²⁺, high spin, coordination number of 6, O_h symmetry) as seen in the relative peak intensities and its resemblance to (Mn)ZnO (high spin). In (Mn)ZnO, the O coordination number is 4 with T_d symmetry, as described in the literature.³² With increasing Mn concentration and processing temperature, the Mn³⁺ state appears, resulting in spectral features similar to those of Mn₂O₃ as seen in Fig. 4(c) and the corresponding O K-edge in Fig. 3(b). The change in Mn valency from 2+ to 3+ and 4+ is frequently seen in the shift to higher photon energy and a decrease in the L₃ to L₂ peak ratios.^4,32–34 Although it is difficult to confirm the presence of the Mn⁴⁺, which is required for ZnMnO₃ formation, ZnMnO₃ is most often found in the form of a mixed phase along with ZnO and spinel ZnMn₂O₄.³² While it is possible that both Mn³⁺ containing ZnMnO₃ and Mn⁴⁺ containing ZnMn₂O₄ impurities exist in the nanopowder and more likely in the near surface region of the nanoparticles. XRD might miss ZnMn₂O₄, which is either too weak to be detected due to its small cluster size or obscured by the dominant ZnO signal. However, it is more probable that mixed valence oxide clusters, MnO_x, are formed upon high temperature annealing. The spectral results are entirely consistent with this notion.

The XEOL spectra are recorded at 1085 eV excitation energy, which is just above the Zn L_3,2-edge. At this energy, the SGM beamline has excellent flux and Zn will be excited preferentially (Fig. S4). Based on the absorption coefficient of Zn, Mn, and O at this energy, Zn absorbs most of the flux and the penetration depth is on the order of 10² nm. More details are displayed in Fig. S4 where the attenuation of the photon for 1% and 10% Mn:ZnO samples is shown.

The XEOL spectra of the Mn:ZnO nanoparticles processed at 800 °C with concentrations from 0% to 10% are shown in Fig. 5(a). They all show a sharp peak at ∼3.2 eV (380 nm), which is the emission from the optical bandgap, and a broad band, and the defect emission, which varies with the dopant concentration in the visible from green to red. The small variation in optical bandgap energy [inset of Fig. 5(a)] is due to the variation in size distribution and quantum confinement. The overall luminescence intensity decreases with the increasing dopant concentration. By normalizing the XEOL spectra to the intensity of the bandgap emission [Fig. 5(b)], one can see that there is a significant change in the overall luminescence and the branching ratio between the defect emission and the bandgap emission upon doping. The addition of Mn decreases the overall luminescence and the defect emission relative to the bandgap emission. A close examination of the bandgap emission [Fig. 5(a) inset] reveals a blue shift of the optical bandgap as the Mn concentration increases (by 40 meV from undoped to 10% doped). This could be due to a change in the binding energy of the exciton as well as quantum confinement noted above. The major difference between the Mn-doped and undoped ZnO systems can be seen in the defect emission; the Mn:ZnO nanoparticles contain defect levels lower in energy than those in the ZnO. The 550 nm defect emission is commonly observed in ZnO nanostructures.^35,36 Incidentally, 550 nm also corresponds to the d–d transition from the ⁴T₁ to ⁶A₁ state of Mn²⁺.³⁷ Thus, energy transfer to Mn²⁺ will quench the ZnO defect emission. The ∼1.8 eV (680) emission most likely corresponds to surface defects on the Mn:ZnO nanoparticles, such as O vacancies near Mn sites that are lower in energy than the O vacancies near Zn sites, resulting in the red shift of the defect band compared to ZnO. Similar surface state emission has been observed in Eu-doped ZnO.³⁶ Furthermore, the luminescence at all dopant concentrations increases in intensity with increasing annealing temperature. An example is shown in Fig. 5(c) for 3% Mn:ZnO [Fig. 5(c)]. A similar enhancement of the luminescence intensity of Mn:ZnO systems by annealing at 800 °C has been previously reported.³⁸ It is almost certainly due to improved crystallinity and the trapping of electrons in the O vacancies. The overall reduction of optical luminescence intensity upon Mn doping shows that Mn doping inevitably reduces radiative electron–hole recombination via the optical channel, which, in turn, enhances electron–hole separation and, hence, its performance in photocatalysis and photodegradation.^9–16 

We now return to magnetism, which triggered the initial flurry of the study of magnetic semiconductors. Figure 6 (top and bottom panels) shows the magnetic hysteresis for the 10% Mn:ZnO and 1% Mn:ZnO processed at 800 and 400 °C, respectively. A list of the magnetic behavior and coercivity, Hc, for all samples is summarized in Table I.

The 10% Mn:ZnO (800 °C) sample displays ferromagnetic ordering at 5 K (Fig. 6, top panel), but no such signature is observed at room temperature. In fact, all samples display a purely paramagnetic state at room temperature. Additionally, the linear shape of the hysteresis loop at low temperature does not rule out the presence of some paramagnetic contribution to the sample. The 1% Mn:ZnO (400 °C) sample displays some degree of ferromagnetic ordering although a superparamagnetic contribution cannot be discounted. Super-paramagnetism (SPM) is common in ferromagnetic materials once the (domain) particle size is <10 nm. The saturation magnetization occurs when H is ∼70 000 Oe for all samples. The coercivity at 5 K for the 1% and 10% Mn:ZnO is ∼110 and ∼225 Oe, respectively. In general, the coercivity increases with increasing processing temperature and Mn concentration. The increase in coercivity is linked to an increase in the domain size in large particles processed at higher temperatures. The samples that contain higher amounts of Mn secondary phases have hysteresis loops with more paramagnetic character (more linear).

Comparing the coercivity in the 3% Mn 800 °C sample with the 10% Mn 800 °C sample, for example, there is a noticeable decrease with increasing Mn secondary phase formation. The formation of secondary phases causes a shift from ferromagnetic ordering toward paramagnetic ordering, decreasing the long-range magnetic order of the system. These results indicate that samples with the majority of Mn as (Mn)ZnO show ferromagnetic behavior, while those containing a higher amount of Mn secondary phase display both ferromagnetic and paramagnetic behaviors at 5 K.

Mn:ZnO nanoparticles prepared via the sol–gel method at various concentrations (1%, 3%, and 10%) and annealing temperatures (400, 600, and 800 °C) have been analyzed. It is found that the as-prepared samples are poor crystalline materials and the crystallinity improves markedly upon annealing as the annealing temperature increases. The improved crystallinity, however, is accompanied by the precipitation of a secondary phase; the higher the Mn concentration, the lower the temperature of the secondary phase, most likely containing MnOx clusters on the surface. Regardless, the dominant phase remains the Mn-doped substitutional phase. The Mn L_3,2-edge and O K-edge are found to be very sensitive to secondary phase formation, and the pre-edge feature in the O K-edge provides semi-quantitative information as to the amount of secondary phase formed. The XEOL result clearly shows that radiative electron–hole recombination is noticeably quenched in Mn:ZnO systems compared to the ZnO nanostructure. This observation strongly suggests that Mn doping will enhance electron–hole separation and, hence, the performance of the system in photocatalysis and photodegradation. Secondary phase formation also leads to a decrease in the coercivity and a shift from FM toward PM. These results can help explain why different Mn concentrations of Zn_1−xMn_xO systems can have different magnetic behaviors. Furthermore, these results help shed light on the issue of why Mn:ZnO systems with identical Mn concentrations show discrepancies in their FM measurements based on the processing temperature. Although further tests are needed to determine the composition of the secondary phase, these results clearly show that both low Mn concentration and low temperature processing are essential if secondary phase formation is to be avoided. These considerations will prove vital to the use of Mn-doped ZnO DMS systems for spintronics and photocatalytic applications.

See the supplementary material for a schematic of the synthesis and SEM and XRD data as well as penetration depth of relevant elements.

Research at the University of Western Ontario (UWO) was supported by NSERC (Grant No. RGPIN-2019-05926), CFI, OIT, and CRC (TKS). The SEM measurement was conducted at Nanofabrication Lab, UWO. M.W.M. acknowledges the warm hospitality of the University of Padova during his visit. Magnetic measurements were made in IMR of Dalhousie University. This research was, in part, conducted at the Canadian Light Source, which was supported by CIHR, NRC, NSERC, and University of Saskatchewan. This research also used resources of the Advanced Photon Source, an Office of Science User Facility operated for the U.S. Department of Energy (DOE) Office of Science by Argonne National Laboratory, and was supported by the US DOE under Contract No. DE-AC02-06CH11357.

M.W.M. participated in the synthesis and did all synchrotron measurements and analyses. L.B. and G.B. participated in the synthesis and laboratory characterization. L.A. and T.-K.S. supervised this work. All authors contributed to the preparation and read the manuscript.

The data that support the findings of this study are available within the article and its supplementary material.