Fe3O4 Nanoparticles - Nano

Dec 1, 2011 - Citation data is made available by participants in Crossref's Cited-by Linking service. ... Enhancement of Magnetization through Interfa...
0 downloads 0 Views 2MB Size
Letter pubs.acs.org/NanoLett

Tuning Exchange Bias in Core/Shell FeO/Fe3O4 Nanoparticles Xiaolian Sun,† Natalie Frey Huls,*,‡ Aruna Sigdel,† and Shouheng Sun† †

Department of Chemistry, Brown University, Providence, Rhode Island 02912, United States Material Measurement Laboratory, National Institute of Standards and Technology (NIST), Maryland 20899, United States



S Supporting Information *

ABSTRACT: Monodisperse 35 nm FeO nanoparticles (NPs) were synthesized and oxidized in a dry air atmosphere into core/shell FeO/Fe3O4 NPs with both FeO core and Fe3O4 shell dimensions controlled by reaction temperature and time. Temperature-dependent magnetic properties were studied on FeO/Fe3O4 NPs obtained from the FeO NPs oxidized at 60 and 100 °C for 30 min. A large exchange bias (shift in the hysteresis loop) was observed in these core/shell NPs. The relative dimensions of the core and shell determine not only the coercivity and exchange field but also the dominant reversal mechanism of the ferrimagnetic Fe3O4 component. This is the first time demonstration of tuning exchange bias and of controlling asymmetric magnetization reversal in FeO/Fe3O4 NPs with antiferromagnetic core and ferrimagnetic shell. KEYWORDS: FeO nanoparticles, controlled oxidation, core/shell FeO/Fe3O4, exchange bias

T

magnetization dynamics of core/shell systems. Therefore, the ability to systematically vary the magnetic properties via chemical control over the particle morphology could have important consequences for the understanding of exchange bias at the nanoscale level. In the previous study on FeO/Fe3O4 NPs, the NPs were made by exposing 22 nm FeO NPs to ambient environment for 10 days and had no control over FeO core and Fe3O4 shell dimensions.12 To better understand the exchange bias behavior, a controllable oxidation method is desirable to tune the dimension of core/shell structure and thus the magnetic properties. Recently, by decomposing Fe(acac)3 directly in oleylamine and oleic acid, we succeeded in making nearly monodisperse FeO NPs with sizes tunable from 14 to 100 nm. These FeO NPs, especially those larger than 20 nm, can be oxidized into Fe3O4 in a more controlled way, providing a unique NP system for studying exchange bias within FeO/ Fe3O4 structure. Here, we used 35 nm FeO NPs as a model to study how controlled oxidation can be used to tune the dimension of core/shell FeO/Fe3O4 NPs and the exchange bias present in the core/shell structure. Monodisperse 35 nm FeO NPs were synthesized by reductive decomposition of Fe(acac)3 in oleic acid and oleylamine at 300 °C.13 Once the FeO NPs were formed, the reaction solution was cooled down to a predetermined temperature (25, 60, or 100 °C), and dry air was introduced to the reaction system. Due to the size (35 nm) of the NPs, this oxidation process could be readily controlled, and core/shell FeO/Fe3O4 NPs with tunable FeO core size and Fe3O4 shell

he inherent electronic effects at the interface of core/shell nanoparticles (NPs) have been applied extensively to tune catalytic, optical, and magnetic properties.1−5 For magnetic core/shell NPs, an interesting consequence is the exchange coupling across the core/shell interface that is frequently seen in the form of exchange bias, a horizontal shift in the hysteresis loop accompanied by an increase in coercivity after cooling in a magnetic field.5 Despite the fact that exchange bias is most often studied in thin films for technological applications in the magnetic hard drive industry, its historical roots are in small particles where it was first observed in Co particles coated with a native surface CoO layer.6 The conventional exchange coupling between an antiferromagnet (AFM) and a ferromagnet (FM) or ferrimagnet (FIM) occurs at the interface where cooling in an external field from above the Néel temperature magnetizes the FM along the surface spin direction of the AFM.7 The AFM then acts as a pinning layer, leading to a unidirectional anisotropy or a preferred direction of magnetization in FM. The result of this unidirectional anisotropy is that it is much more difficult to switch the magnetization of the FM in the direction opposite to the cooling field than it is to switch it back to the cooling field direction. This is manifest as a shift in the magnetization versus field (M−H) curve and is often accompanied by an increase in the coercivity. This phenomenon has been observed in core/shell NPs of Co/ CoO,6,8 CrO2/Cr2O3,9 and the so-called “inverted” AFM core/ FIM shell systems of MnO/Mn3O410 and the FeO/Fe3O4 presented herein.11,12 While still poorly understood in two-dimensional systems, exchange bias in zero-dimensional systems is even more complicated due to the high surface area to volume ratio in NPs and consequently the large role the interface plays in © 2011 American Chemical Society

Received: October 3, 2011 Revised: November 29, 2011 Published: December 1, 2011 246

dx.doi.org/10.1021/nl2034514 | Nano Lett. 2012, 12, 246−251

Nano Letters

Letter

diffraction from Fe3O4 formed from surface oxidation of FeO during NP purification and separation processes. After controlled oxidation, the NPs show a more prominent Fe3O4 feature with the peak related to the fcc FeO becoming less intense and broader, indicating that the size of the FeO core is reduced. The TEM and XRD analyses further reveal that the structural conversion from FeO to Fe3O4 is temperature dependent. Longer reaction time or higher reaction temperature under dry air yields a larger fraction of Fe3O4. The average (200) correlation size of the as-synthesized FeO was estimated with Scherrer’s formula to be 13.5 nm and decreased to 11 and 8.5 nm after oxidation in air at 60 and 100 °C, respectively. Correspondingly, the estimated (220) correlation size of Fe3O4 increased from 6.5 and 8.5 to 9.4 nm as the oxidation temperature increased from room temperature, 60 to 100 °C. The conversion of FeO to Fe3O4 is further characterized by magnetization value changes of the oxidized NPs. For magnetic measurement, samples were prepared by placing the powder in the sample holder and adding epoxy to the holder. The setting of the epoxy around the nanoparticles prevented physical motion of the powders in response to the magnetic field. Figure 3a shows the room temperature hysteresis loops of the assynthesized FeO NPs and the oxidized NPs obtained from different reaction temperatures. Due to the presence of a thin surface oxidation layer, the as-synthesized FeO NPs show a small magnetization value of 30 Am2/kg. Once further treated in air, the magnetization values are increased. At 120 °C reaction temperature, this magnetization increase depends also on oxidation time (Figure 3b). From Figure 3b, we can see that the magnetization value reaches its maximum after 90 min to 75 Am2/kg . This implies that the 35 nm FeO NPs can be fully oxidized into Fe3O4 NPs after 90 min oxidation in air at 120 °C. Therefore, to ensure the formation of core/shell FeO/ Fe3O4 NPs, the FeO NPs should be treated at temperatures below 120 °C for less than 90 min. We note that none of the loops appear to be saturated at 1.5 T in Figure 3a, which can be explained by the presence FeO contributing a paramagnetic background. Magnetization as a function of temperature was measured in an external field of 5 mT. Figure 4 shows the zero field cooled (ZFC) and field cooled (FC) curves for the as-synthesized (Figure 4a) and oxidized (Figure 4b,c) samples. The magnetization is small and nearly constant for temperatures below the Néel temperature (TN ≈ 192 K) in the antiferromagnetic state with a sharp increase in magnetization occurring around TN. The peak occurring above TN at around 225 K likely coincides with the thermal instability of the magnetization of the native Fe3O4 ferrimagnet present at the surface of the particles. For the NPs treated at 60 °C, the magnetization as a function of temperature profile appears nearly the same, but a slight kink in the ZFC curve at around 120 K emerges, which coincides with the well-known Verwey (TV) transition of Fe3O4. Such a signature of the Verwey transition in Fe3O4 NPs is not uncommon and has been reported before in other literature pertaining to Fe3O4 NPs.14,15 However, this is the first time this particular feature is observed in the FeO/Fe3O4 core/shell NP system, which implies that 60 °C treatment either greatly increases the amount of Fe3O4 present on the NP shell or the Fe3O4 present has much better crystallinity than reported in other systems. Notice that while the Néel temperature remains the same for the FeO NPs oxidized at 60 °C, the peak associated with the thermal instability of the Fe3O4 appears

thickness were made for magnetic studies. Dry air condition seemed to be necessary in the oxidation reaction, as FeO NPs were sensitive to humidity and tended to decompose into Fe and Fe3O4 before their full oxidation to Fe3O4. Under dry air atmosphere, the FeO NPs could be fully oxidized to Fe3O4 NPs when the solution was heated at 120 °C for 90 min. Figure 1 displays representative transmission electron microscope (TEM) images of the as-synthesized FeO NPs

Figure 1. TEM images of (a and b) the as-synthesized 35 nm FeO NPs; (c) FeO/Fe3O4 NPs obtained from the controlled oxidation of the 35 nm FeO NPs at 100 °C for 30 min; and (d) HRTEM image of a part of one representative 32 nm FeO/Fe3O4 NP as-synthesized NP.13

and those after controlled oxidation at 100 °C for 30 min. The as-synthesized FeO NPs have a truncated octahedral shape with a mean size of 35 ± 2 nm. The partially oxidized NPs have similar morphology to the as-synthesized FeO NPs but show a contrast in the core and shell regions, indicating the formation of core/shell structure. High resolution TEM (HRTEM) of the FeO/Fe3O4 NPs has likewise been analyzed.13 Figure 1d is a representative image of a part of a single 32 nm FeO NP obtained from slow oxidization under air at room temperature for 30 min. The core/shell FeO/Fe3O4 structure is readily seen in the image with (200) lattice fringes shown in FeO core and (220) lattice fringes observed in the Fe3O4 shell structure. The FeO NP oxidation process was monitored by X-ray diffraction (XRD) (Figure 2). The as-synthesized FeO NPs

Figure 2. XRD patterns of the as-synthesized FeO NPs and FeO/ Fe3O4 NPs formed by controlled oxidation of FeO NPs at different temperatures.

show typical (111), (200), (220), (311), and (222) diffraction peaks of the face centered cubic (fcc) wüstite structure. A small peak at 34.5° can also be seen. This peak belongs to the (311) 247

dx.doi.org/10.1021/nl2034514 | Nano Lett. 2012, 12, 246−251

Nano Letters

Letter

Figure 3. (a) Hysteresis loops of the as-synthesized FeO NPs and those oxidized at different temperatures for 30 min. (b) Magnetization values of samples oxidized at 120 °C for different times.

interaction between the Fe2+ ions in the Fe3O4 octahedral sites also allows for a larger remnant magnetization in contrast to the antiferromagnetic FeO, which in its pure form should have little to no remnant magnetization beyond the interfacial contribution. It is interesting to note that the coercivity decreases with increasing Fe3O4 content. For the as-synthesized FeO NPs, we expect a thin native Fe3O4 shell to form (as evidenced by the peak in the magnetization versus temperature curves). The anisotropy in this thin shell will be dominated by surface effects, therefore leading to a larger coercivity (μ0HC = (90 ± 6) mT, where the uncertainty is due to the existence of flux trapped in the superconducting magnet). As the amount of Fe3O4 in the particles increases, the particles’ coercivities take on behavior associated more closely with bulk Fe3O4 (μ0HC = (88 ± 6) and (58 ± 6) mT for the samples treated at 60 and 100 °C, respectively). M−H curves were also measured after cooling the particles from room temperature (T > TN) in a field of 1 T in order to determine the existence of exchange coupling at the FeO/ Fe3O4 interface. A large exchange bias was observed for all samples. This is also consistent with reports of FeO interacting with a native iron oxide shell, however as we show in Figures 5b−d, the exchange coupling has features that are qualitatively different from what has been previously reported in FeO/Fe3O4 NPs.11,12 Figure 5b shows the M−H curves for the assynthesized NPs taken at 2 K in the FC (filled circles) and ZFC (open circles) conditions. What is immediately noticeable is a large shift along the H-axis toward the negative field, indicative of the unidirectional anisotropy induced by the field cooling. This shift is so large that both the descending and ascending coercive fields are negative. There is also a prominent vertical shift in the M−H curve such that both the descending and ascending remnant magnetizations are positive. A vertical shift is often observed in exchange coupled NP systems and can be explained by the presence of uncompensated spins at the AFM/ FIM interface. When the particles are cooled without a field, these spins are aligned antiferromagnetically with the core and contribute no net magnetization. However, after applying a field above the Néel temperature, these spins are aligned with the field but still pinned by the AFM, thus maintaining a preferred direction of magnetization. That the negative maximum magnetization is the same in the ZFC and FC conditions, while the positive maximum magnetization is much higher in the FC case, which further attests to the preferential alignment of the uncompensated interfacial spins in the field cooled case. These uncompensated interfacial spins, which experience the competition between exchange and Zeeman

Figure 4. Magnetization as a function of temperature taken in a field of 5 mT for (a) as-synthesized FeO particles; (b) particles oxidized at 60 °C; and (c) particles oxidized at 100 °C.

much broader and occurs at a higher temperature. This is consistent with a distribution of shell thickness (and thus anisotropy) as well as dipolar interactions occurring between neighboring particles. Lastly, Figure 4c shows the magnetization as a function of temperature for the FeO sample oxidized at 100 °C. Here the kink associated with the Verwey transition is much more prominent due to the higher volume fraction of Fe3O4 present. The blocking peak also associated with the Fe3O4 present in the particles is shifted to temperatures higher than the measurement limit (above 350 K), which for NPs of this size implies that the Fe3O4 in the powder is strongly interacting. Overall, the three magnetization versus temperature profiles indicate increasing contribution of the Fe3O4 to the volume fraction with increasing oxidation temperature and suggest that the ferrimagnetic oxide species is Fe3O4 (evidenced by the Verwey transition). Next, magnetization versus field (M−H) measurements were performed on all samples at a range of temperatures. Figure 5a shows the results of the M−H curves for all samples taken at 2 K. The results are fully consistent with increasing Fe3O4 content after controlled oxidation. The maximum magnetization values (measured at 2 T) increase for increasing oxidation temperature as the ferrimagnetic Fe3O4 increasingly contributes to the overall magnetization. The positive exchange 248

dx.doi.org/10.1021/nl2034514 | Nano Lett. 2012, 12, 246−251

Nano Letters

Letter

Figure 5. (a) ZFC magnetization as a function of field for as-synthesized NPs (squares); NPs oxidized at 60 °C (circles); and at 100 °C (triangles) taken at 2 K. (b−d) M−H for the NPs after cooling in 1 T (filled circles) to 2 K compared to the ZFC case (open circles) for the as-synthesized NPs (b), the NPs oxidized at 60 °C (c), and at 100 °C (d).

Information) at which point the thermal energy can help reverse the magnetization at the interface enough so that coherent reversal can be achieved along the descending branch. Figure 5c shows the field cooled M−H curve for the FeO sample treated at 60 °C for 30 min. Qualitatively it looks similar to the as-synthesized sample in that there are large shifts along both H- and M-axes. The magnetization reversal asymmetry is still prominent as well. What is different is that the coercivity and the exchange field (HE) are lower for this sample (μ0HC = (222 ± 6) and (226 ± 3) mT) than for the as-synthesized sample (μ0HC = (310 ± 6) and (256 ± 3) mT). We can understand this decrease in HC and HE with oxidation treatment by noting two important factors in determining exchange anisotropy: (1) Since exchange bias is an interfacial phenomenon, by decreasing the size of the AFM core through oxidation, the AFM/FIM interface has effectively been decreased thus weakening the exchange anisotropy; and (2) the ability of the AFM core to effectively pin the FIM spins in the shell is directly proportional to the antiferromagnetic anisotropy energy, which scales with core volume. Thus a smaller core will possess a smaller anisotropy energy which will in turn be easier for the FM to overcome when reversing its magnetization. Figure 5d compares the ZFC and FC M−H curves for the sample oxidized at 100 °C for 30 min. There is still clearly an exchange bias present (μ0HE (33 ± 3) mT) as well as a vertical shift of the loop (see inset), though it is not as pronounced as for the as-synthesized sample and the sample treated at 60 °C. The asymmetry in the magnetization reversal is still present as well, though it should be noted that whereas the onset of the asymmetry occurred at fields well below the coercive field in the as-synthesized and 60 °C treated samples, for the sample oxidized at 100 °C, the onset of asymmetry occurs at fields higher than the coercive field, as the descending branch approaches negative saturation. This is likely due to the lower pinning energy of the AFM core, allowing more of the FIM

energies, along with the well-known surface anisotropy from the Fe3O4 spins, are responsible for the slow approach to saturation and irreversibility present close to the maximum field measured. This magnetic behavior, along with the AFM behavior of the FeO, is also responsible for the lack of saturation seen in all low-temperature loops. Therefore, in our analysis we refer not to the saturation magnetization but to the maximum magnetization measured in each magnetic state. Another interesting feature of the field-cooled M−H curve for the as-synthesized sample is its asymmetrical magnetization reversal. In this sample, the descending branch of the M−H curve shows a much slower approach to negative saturation. When the field is reversed from the negative magnetization state and begins to increase, the shape of the loop is more consistent with the sample in the ZFC condition. This asymmetric magnetization reversal has not to our knowledge been reported before in this system and indicates that the switching from magnetic alignment parallel to the cooling field to antiparallel to the cooling field occurs via domain wall nucleation and propagation, while the switch from magnetization antiparallel to the cooling field to parallel to the cooling field can be achieved primarily via coherent magnetization rotation. This makes sense because in the descending branch, the magnetization reversing from the positive state needs not only to reach the coercive field value of the Fe3O4 shell but also to overcome the additional exchange anisotropy that favors parallel alignment of the FIM spins with uncompensated AFM spins. The high magnetostatic energy that results from the relatively easy rotation of the outermost spins compared to the strong pinning of the uncompensated interfacial spins results in domain wall nucleation and propagation occurring in the descending branch. But when the magnetization starts to reverse from the negative magnetization state, the uncompensated AFM spins add with the Zeeman energy to reverse the FIM spins, leading to the breaking of asymmetry and the decreasing of the coercive field. This asymmetrical magnetization reversal persists up to 75 K (Figure S1, Supporting 249

dx.doi.org/10.1021/nl2034514 | Nano Lett. 2012, 12, 246−251

Nano Letters

Letter

shell to switch using magnetization rotation before nucleating a domain wall to overcome the magnetostatic energy. These three examples illustrate that not only does the coercivity and exchange field decrease with increasing oxidation temperature (and thus increasing (decreasing) shell (core) dimensions) but also that the relative dimensions of the core and shell determine the dominant reversal mechanism of the FIM component. For particles where the shell is thinnest, the competition between magnetostatic and exchange energies is highest leading to domain wall nucleation at lower fields. For particles where the shell is thickest, magnetization rotation is the dominant reversal mode for the shell until the very highest fields when the magnetostatic energy becomes large enough that a domain wall forms. Figure 6 is a comparison of the coercive fields for the three samples as a function of temperature in the ZFC (Figure 6a)

bias does persist up to the FeO Néel temperature for all samples. What is interesting to note is the different temperature evolution as a function of oxidiation temperature. The assynthesized sample and the sample oxidized at 60 °C both start at low temperatures with very high exchange field which decreases rapidly with temperature. The sample oxidized at 100 °C has a small low-temperature exchange field that decreases very slowly with temperature. Since all the samples have the same total diameter, it is clear that the sample dependent behavior is a manifestation of the competition between the exchange anisotropy at the interface and the temperaturedependent anisotropy of the Fe3O4. In effect, the anisotropy and temperature stability of the shell grow at the expense of the pinning strength of the core. It is worth noting that unlike other exchange bias studies of partially oxidized FeO, this is the first in which the core and/or the shell dimensions could be varied, and it is also the first FeO/Fe3O4 study where the findings report asymmetric magnetization reversal. Intriguingly, these results are qualitatively very similar to a recent Monte Carlo simulation study,16 which shows that for an asymmetrical NP with varying AFM core and FM shell dimensions (yet constant total radius), the onset of magnetization reversal does occur at a sufficiently large core diameter and becomes less pronounced as the AFM core size decreases. Similarly, the largest horizontal shift was also observed with the largest AFM core as well, also in agreement with these results. In conclusion, monodisperse 35 nm FeO/Fe3O4 core/shell NPs were synthesized through controlling oxidation under a flow of dry air. The core and shell dimensions were tunable via controlling the oxidation temperature and time. The magnetic properties of the FeO/Fe3O4 particles, which were oxidized to varying the core shell extent, were systematically studied. Exchange bias was observed for all samples. The coercivity and exchange field decreased with increasing oxidation temperature (and thus increasing (decreasing) shell (core) dimensions), which is due to the reduction of AFM/FIM interface area and AFM core anisotropy energy. Asymmetric magnetization reversal in this FeO/Fe3O4 system was first observed. The study provides a deeper understanding of exchange bias in AFM/FIM core/shell nanostructure, which may be important for guiding the design and fabrication of magnetic nanodevices for information storage applications.

Figure 6. Temperature evolution of coercivity (HC) in the ZFC condition (a), FC condition (b), and exchange field (HE) (c) for the as-synthesized NPs (squares) and NPs oxidized at 60 °C (circles) and at 100 °C (triangles). The error in panel (a) is due to flux trapped in the superconducting magnet. Similar error bars are present in panels (b) and (c) but are smaller than the size of the symbols.



and FC (Figure 6b) conditions as well as the exchange field (Figure 6c) as a function of temperature. That there is a large increase in coercivity after field cooling (more than a factor of 3 for the as-synthesized sample at low temperatures) indicates that a large uniaxial anisotropy is induced after field cooling as well. Several models for exchange biased thin films theorize that surface roughness (i.e., uncompensated interfacial spins) contribute to the uniaxial anisotropy which is consistent with the large vertical shift as well. At higher temperatures, the sample dependence of the coercivity decreases and the values begin to converge. However, the as-synthesized sample actually has a small coercivity (μ0HC = (20 ± 6) mT) at 175 K, which then falls to 0 at 200 K due to the thermal fluctuations in the shell overcoming the anisotropy of the Fe3O4. Interestingly, in the FC case, the as-synthesized sample has a coercivity of (19 ± 6) mT at 200 K, indicating that the added uniaxial anisotropy from the uncompensated spins is enough to significantly increase the thermal stability of the shell. Figure 6c shows the exchange field as a function of temperature for all three samples. As expected, the exchange

ASSOCIATED CONTENT

S Supporting Information *

Additional figures (Figure S1). This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION Corresponding Author *E-mail: [email protected]; [email protected].



ACKNOWLEDGMENTS The work was supported in part by DOE/EPSCoR DE-FG0207ER36374.



REFERENCES

(1) Yoon, T.; Lee, H.; Shao, H.; Weissleder, R. Angew. Chem., Int. Ed. 2011, 50, 4663−4666. (2) Wang, C.; Vliet, D.; More, K.; Zaluzec, N.; Peng, S.; Sun, S.; Daimon, H.; Wang, G.; Greeley, J.; Pearson, J.; Paulikas, A.; 250

dx.doi.org/10.1021/nl2034514 | Nano Lett. 2012, 12, 246−251

Nano Letters

Letter

Karapetrov, G.; Strmcnik, D.; Markovic, N.; Stamenkovic, V. Nano Lett. 2011, 11, 919−926. (3) Crut, A.; Juvé, V.; Mongin, D.; Maioli, P.; Fatti, N.; Vallée, F. Phys. Rev. B 2011, 83, 205430. (4) Zhu, J.; Shen, Y.; Xie, A; Zhu, L. J. Mater. Chem. 2009, 19, 8871− 8875. (5) Nogués, J.; Sort, J.; Langlais, V.; Skumryev, V.; Suriñach, S.; Muñoz, S. S.; Baró, M. D. Phys. Rep. 2005, 422, 65−117. (6) Meiklejohn, W. H.; Bean, C. P. Phys. Rev. 1957, 105, 104. (7) Nogués, J.; Schuller, I. K. J. Magn. Magn. Mater. 1999, 192, 203− 232. (8) Inderhees, S. E.; Borchers, J. A.; Green, S. K.; Kim, M. S.; Sun, K.; Strycker, G. L.; Aronson, M. C. Phys. Rev. Lett. 2008, 101, 117202. (9) Zheng, R. K.; Liu, H.; Wang, Y.; Zhang, X. X. Appl. Phys. Lett. 2004, 84, 702−704. (10) Berkowitz, A. E.; Rodriguez, G. F.; Hong, J. I.; An, K.; Hyeon, T.; Agarwel, N.; Smith, D. J.; Fullerton, E. E. J. Phys. D 2008, 41, 134007. (11) Kavich, D. W.; Dickerson, J. H.; Mahajan, S. V.; Hasan, S. A.; Park, J.- H. Phys. Rev. B 2008, 78, 174414. (12) Chen, C. −J.; Chiang, R. −K.; Lai, H. −Y.; Lin, C. −R. J. Phys. Chem. 2008, 114, 4258−4263. (13) Hou, Y.; Xu, Z.; Sun, S. Angew. Chem., Int. Ed. 2007, 46, 6329− 6332. (14) Goya, G. F.; Berquó, T. S.; Fonseca, F. C.; Morales, M. P. J. Appl. Phys. 2003, 94, 3520−3528. (15) Redl, F. X.; Black, C. T.; Papaefthymiou, G. C.; Sandstrom, R. L.; Yin, M.; Zeng, H.; Murray, C. B.; O’Brien, S. P. J. Am. Chem. Soc. 2004, 126, 14583−14599. (16) Hu, Y.; Liu, Y.; Du, A. J. Appl. Phys. 2010, 108, 033904.

251

dx.doi.org/10.1021/nl2034514 | Nano Lett. 2012, 12, 246−251