Since its discovery in Co²⁺-doped TiO₂ in 2001,¹ high-Curie temperature (T_C) ferromagnetism has been observed in a number of diluted magnetic oxides (DMOs),² and elucidation of its origins has been cited as among the most important problems in magnetism to have emerged in several years.^3,4 Identification of the critical parameters governing DMO ferromagnetism has been challenging, however, and often different laboratories have reported incongruous results for seemingly identical materials. For example, high-T_C ferromagnetism in p-type Mn²⁺:ZnO was predicted theoretically,^5-8 but many groups have observed only paramagnetism.^9-11 Since the first claim of room-temperature ferromagnetism in Mn²⁺:ZnO (ferromagnetic saturation moment M_S(300 K) 0.16 _B/Mn²⁺),¹² high-T_C ferromagnetism has been observed in Mn²⁺:ZnO/GaAs films that appeared to correlate with p-type conductivity,¹³ but near-room-temperature ferromagnetism was also claimed for n-type Mn²⁺:ZnO.¹⁴ Recently, ferromagnetism described earlier as intrinsic was suggested to instead arise from manganese oxide impurity phases, such as Zn-doped Mn₂O₃ or Mn₃O₄.^15,16 Here, we report that high-T_C ferromagnetism in Mn²⁺:ZnO can be activated reliably by amine binding and calcination, a process interpreted as incorporating nitrogen into the ZnO lattice as a p-type dopant. These results clearly demonstrate that Mn²⁺:ZnO ferromagnetism is critically sensitive to defects other than the magnetic dopants themselves, offering some insight into the otherwise irreconcilable observations reported previously.

We recently reported high-T_C ferromagnetism in Mn²⁺:ZnO prepared by direct chemical synthesis.¹⁷ To explain the large M_S of 1.35 /Mn²⁺, we speculated that calcination of amines may introduce p-type nitrogen defects into the ZnO lattice. To test this hypothesis, we have performed the split-batch experiment described by Figure 1.¹⁸ A single batch of high-quality 0.20% Mn²⁺:ZnO nanocrystals¹⁷ was split into two equal parts. Half were capped with trioctylphosphine oxide (TOPO, O-capped), and the other half were capped with trioctylamine (N-capped) under identical conditions. Clear colloidal suspensions of the nanocrystals in toluene were spin coated onto fused-silica substrates and calcined aerobically at 500 C for 2 min per coat (~20 coats total, 1 m thickness). Figure 1 shows 300 K magnetization loops collected for the two films. Whereas the sample prepared from O-capped nanocrystals showed little magnetization at 300 K, that prepared from N-capped colloids showed a strong magnetic hysteresis with M_S = 1.22 /Mn²⁺ and a coercivity of 65 Oe. This experiment was repeated several times with various N (dodecylamine or trioctylamine) or O (TOPO or acetate) ligands. Without exception, N-capped colloids yielded ferromagnetic films (13 films, mean M_S(300 K) = 0.76 ± 0.42 /Mn²⁺), whereas O-capped colloids yielded paramagnetic films (10 films, mean M_S(300 K) = 0 ± 0 /Mn²⁺). Calcined powders gave similar results.

Figure 1 Split-batch experiment; 300 K magnetization data for films spin coated from N-capped (·) and O-capped () Mn2+:ZnO colloids taken from the same synthesis batch. See Supporting Information for experimental details.

Figure 2a shows magnetization data measured as a function of temperature for representative ferromagnetic and paramagnetic samples. The data from the paramagnetic sample are reproduced well by a S = ⁵/₂ Brillouin function¹⁷ using the analytical Mn²⁺ concentration determined by atomic emission spectrometry and no variable parameters. This agreement demonstrates that manganese oxides are not formed during calcination since, in that case, the S = ⁵/₂ paramagnetism would be reduced. Indeed, the concentration of Mn²⁺ (0.20%) is ca. 10² times lower than its solid solubility under these conditions (~20%).¹⁹ In addition to paramagnetism, the ferromagnetic sample shows a large temperature-independent moment that persists up to the instrument limit of 350 K. This signal is the ferromagnetism shown in Figure 1, and its temperature independence demonstrates T_C 350 K.

Figure 2 (a) Variable-temperature magnetization data for 0.20% Mn2+:ZnO samples prepared from N-capped (·) and O-capped () colloids, collected at 5000 Oe to reveal the paramagnetism. The solid line in (a) was calculated from the S = 5/2 Brillouin function, as described in the text. (b) MS(300 K) as a function of Mn2+ concentration, represented in units of emu/g (·) and mB/Mn2+ ().

Figure 2b shows the dependence of the 300 K ferromagnetism on Mn²⁺ concentration for a series of samples prepared under identical conditions. These data show that Mn²⁺ is required, but that increasing its concentration above 0.20% actually reduces both M_S/Mn²⁺ and M_S/g, inconsistent with scenarios involving ferromagnetism from segregated manganese oxides. A similar concentration dependence in n-type DMOs has been rationalized by invoking formation of antiferromagnetically coupled TM²⁺ dimers.⁴ From 0.20 to 1.0%, the statistical probability that a Mn²⁺ ion is isolated without any nearest neighbor Mn²⁺ ions decreases by only 16% (from 98 to 82%),²⁰ however, a difference far too small to account for the 97% reduction in ferromagnetism over the same concentration range, arguing against this explanation in the present case. This concentration dependence likely reflects the relative availability of nitrogen⁸ (vide infra) and is the subject of ongoing investigation.

We attribute the activation of ferromagnetism in Mn²⁺:ZnO to the incorporation of uncompensated p-type defects into the ZnO lattice upon amine calcination, a process that has substantial precedence in the literature surrounding p-type ZnO. Successful methods for preparing p-type ZnO have largely relied on introduction of NO, NH₃, N₂, or N₂O during²¹ or after²² vacuum deposition to incorporate nitrogen, believed from single-crystal EPR data to substitute for lattice oxygen as the reduced shallow acceptor, N_O^3-.²¹ Three factors likely contribute to the successful incorporation of uncompensated p-type defects in our Mn²⁺:ZnO samples. (1) Amine capping forms Zn²⁺-N_amine bonds, providing activated nitrogen already bound to lattice cations. (2) Aerobic synthesis and calcination aid suppression of native donor defects, such as O_V, Zn_i, or H.²¹ (3) Amine capping passivates surface defects that could otherwise be trapped during sintering (evident from the diminished green defect emission and enhanced UV excitonic emission in N-capped ZnO nanocrystals¹⁷). We note that addition and calcination of amines after spin coating also enhances the ferromagnetism of the Mn²⁺:ZnO films.

The experimental observations are consistent with a microscopic mechanism involving formation of bound magnetic polarons (BMPs)²³ upon introduction of p-type dopants, analogous to that proposed recently for n-type DMOs.⁴ Magnetic exchange interactions between defect-bound VB holes and Mn²⁺ ions occupying the same space are proposed to align Mn²⁺ spins with respect to one another, forming a BMP. As the defect density is increased, overlapping BMPs coalesce into an extended ferromagnetic domain, as illustrated in Scheme 1.⁴ The hole-Mn²⁺ exchange energy (N₀) arises from oxo(p)-Mn²⁺(d) covalency and can be parametrized using the configuration interaction expression in eq 1,²⁴ in which the resonance integral is proportional to the spatial overlap (Scheme 1), S is the dopant spin, and E^- represents the energy associated with transfer of the VB hole to Mn²⁺ (E⁺ for transfer of a VB electron to Mn²⁺).

Scheme 1

By MCD spectroscopy, the Mn²⁺ Mn³⁺ photoionization transition in Mn²⁺:ZnO has been identified at 24 000 cm^-1,¹⁷ just 4000 cm^-1 below the VB CB excitation, implying that E^- is very small (~4000 cm^-1). The strong hole-Mn²⁺ hybridization that results promotes long-range Mn²⁺-Mn²⁺ exchange coupling (Scheme 1). This analysis is supported by density functional calculations of Mn²⁺:ZnO band structure.^6-8 When p-doped, delocalization of VB holes onto Mn²⁺ ions is extensive and the wave function at the Fermi level possesses substantial manganese character. Hybridization at the Fermi level is the source of all potential spintronics applications of this material.

In summary, we report the discovery that calcination of amines activates high-T_C ferromagnetism in Mn²⁺:ZnO. Perhaps the most important aspect of this result is the clear demonstration that the magnetism of Mn²⁺:ZnO (and by inference, other DMOs) is critically dependent upon factors other than the transition metal dopants themselves. Such factors are likely responsible for the broad range of magnetic properties previously reported.

Acknowledgment

This work was funded by the NSF (DMR-0239325 and ECS-0224138). D.R.G. is a Cottrell Scholar of the Research Corporation. Paul Archer (U.W.) is thanked for preliminary magnetic measurements.