This content has been downloaded from IOPscience. Please scroll down to see the full text.
Download details:
IP Address: 140.113.38.11
This content was downloaded on 26/04/2014 at 08:21
Please note that terms and conditions apply.
Fabrication, characterization and studies of annealing effects on ferromagnetism in
Zn1−x Cox O nanowires
View the table of contents for this issue, or go to the journal homepage for more
2006 Nanotechnology 17 5511
(http://iopscience.iop.org/0957-4484/17/21/036)
Nanotechnology 17 (2006) 5511–5518 doi:10.1088/0957-4484/17/21/036
Fabrication, characterization and studies
of annealing effects on ferromagnetism in
Zn
1
−x
Co
x
O nanowires
Z Y Wu
1, F R Chen
1, J J Kai
1, W B Jian
2,4and J J Lin
2,3,41Department of Engineering and System Science, National Tsing Hua University,
Hsinchu 300, Taiwan
2Department of Electrophysics, National Chiao Tung University, Hsinchu 300, Taiwan 3Institute of Physics, National Chiao Tung University, Hsinchu 300, Taiwan
E-mail:[email protected]@mail.nctu.edu.tw
Received 9 June 2006, in final form 3 October 2006
Published 20 October 2006
Online at
stacks.iop.org/Nano/17/5511
Abstract
Diluted magnetic semiconductor Zn
1−xCo
xO (x
0
.11) nanowires with
average diameter of
∼40 nm were prepared by thermal evaporation, followed
by high-energy Co ion implantation. Bombardment by Co ions produced a
good number of structural defects (stacking faults and orientational
variations) in the nanowires. The as-implanted nanowires were paramagnetic.
We performed two types of thermal annealing, one in 1 atm argon flow and
the other in a high vacuum, at 600
◦C, and studied the effects of annealing on
the magnetic properties of these nanowires. Argon annealing removed
structural defects in the nanowires and the nanowires then revealed
ferromagnetic ordering. This result suggests that structure defects are harmful
to the occurrence of ferromagnetism in the Co-implanted ZnO. The structure
of the as-implanted and the annealed nanowires was inspected in detail by
using scanning electron microscopy, energy dispersive x-ray spectroscopy,
maps of electron energy loss spectra, x-ray diffraction, and high-resolution
transmission electron microscopy. Taken together, these studies suggested
that no second phase existed on the scale down to the spatial resolution of
∼0.5 nm. Noticeably, the nanowires even displayed largely enhanced
ferromagnetism after annealing in a high vacuum. A subsequent annealing in
oxygen has also been performed on those vacuum-annealed nanowires to
study the roles played by the O vacancies in determining the ferromagnetic
properties of the nanowires. Our results indicate that both the improved
structural quality and the increased number of O vacancies are key factors for
the occurrence of ferromagnetic ordering in the Zn
1−xCo
xO nanowires.
1. Introduction
Diluted magnetic semiconductors [1] (DMSs) have attracted much attention because of the possibility of introducing the spin degree of freedom into the charge controllable semiconductor devices. The integration of charge and spin in the existing semiconductor technologies, such as transistor, memory, and magnetic recording technologies, has evolved into a new field of fundamental and applied research called
4 Authors to whom any correspondence should be addressed.
‘spintronics’ [2]. A DMS is known to be a good candidate for the application of spintronic devices and scientists are searching for room-temperature (RT) ferromagnetic DMSs. An experimental result of the influence of carrier concentration on ferromagnetism (FM) of the IV–VI DMS material, PbSnMnTe, was reported [3] and indirect exchange via the Ruderman–Kittel–Kasuya–Yosida (RKKY) interaction was proposed to explain the occurrence of the observed FM. The transition temperature of the IV–VI DMS material was, however, too low to be useful in real applications. Munekata
Z Y Wu et al
et al [4] demonstrated the existence of FM in the III–V DMS In1−xMnxAs. Afterwards the new III–V DMS of (Ga, Mn)As
was prepared and it was found to have a Curie temperature (TC) as high as 110 K [5,6]. Recently, Dietl et al [7] have
theoretically explained the 110 K high TC found in the
p-type DMS of (Ga, Mn)As. Furthermore, they have predicted the possible existence of RT FM in GaMnN and ZnMnO. The above-mentioned works have recently triggered intense research activities on DMS materials, both theoretical and experimental.
ZnO is a II–VI DMS host material with a wurtzite structure. It possesses a direct, wide band gap of∼3.3 eV and a large exciton binding energy of∼60 meV. ZnO is transparent and can be over-doped to form conductive films [8, 9] for applications in the display technology. ZnO-based materials are expected to be widely used in optoelectronics [10], solar cells, gas sensors, and nanoelectronic devices in the near future [11]. Co-doped ZnO stands out as one of the possible candidates for RT FM. Experimentally, Ueda
et al [12] and several groups [13–20] have suggested RT or intrinsic FM in Co-doped ZnO films, while other groups have reported either the absence of FM or even the presence of antiferromagnetism [21, 22]. Other groups have argued for extrinsic FM [23] due to the existence of second phases of Co nanocrystals [24,25] and microstructures [26]. Theoretically, a double-exchange mechanism either from the Zener model description [7] of delocalized high-density carriers inducing RKKY-like interactions or from the transition metals resulting in deep impurity band states [27, 28] has been proposed to explain the FM in Zn1−xCoxO. The indirect exchange via
donor defects and bound magnetic polarons [29, 31] was argued to explain the high Curie temperatures in dilute FM oxides. Theoretical models of both n-type [30,31] and p-type doping [32,7] have been given to realize RT FM in ZnO-based DMS.
Nanowire structures of ZnO-based DMSs with diameters smaller than 100 nm, which is a typical thickness of the specimen usually used for transmission electron microscopy (TEM) studies, provides a new platform to study structure effects on physical properties. It is known that different growth conditions, atomic structures (e.g., being crystalline or polycrystalline), and thermal annealing conditions can lead to distinct magnetic properties of Co-doped ZnO. Thus, in this work we performed thermal annealing on our Zn1−xCoxO
nanowires in high vacuum as well as in oxygen to investigate the variations in their magnetic properties, and we utilized high-resolution transmission electron microscopy (HRTEM) to inspect the resulting crystal structure of these nanowires. The Zn1−xCoxO nanowires used in the present work are
∼40 nm in diameter. Various characterization methods, including scanning electron microscopy (SEM), HRTEM, energy dispersive x-ray spectroscopy (EDX), x-ray diffraction (XRD), and electron energy loss spectroscopy (EELS), were carried out to confirm that there does not exist a second phase of Co clusters or any nanocrystal up to the scale of∼0.5 nm. In other words, to within our experimental limit, there are no aggregations, if any exist at all, containing more than 2– 3 Co atoms. The correlations of thermal annealing conditions, structure effects, and O vacancies with the possible occurrence of ferromagnetism in Zn1−xCoxO nanowires are the main
subjects of this work.
This paper is organized as follows. Section2describes our experimental method for the fabrication and characterization of 40 nm diameter Zn1−xCoxO nanowires. Section3contains
our experimental results and discussion of the various magnetic behaviours of the Zn1−xCoxO nanowires and their correlations
with the structure quality and annealing conditions. Our conclusion is given in section4.
2. Experimental details
ZnO powder was placed in a crucible situated at the centre of a quartz tube which was placed in a furnace heated to 950◦C. A glass substrate with gold nanoparticles (∼40 nm in diameter) as catalysts pre-deposited on it was located at the downstream end of the quartz tube where the temperature was kept at 500◦C. The chamber was maintained at 200 Pa with a constant flow of argon. After 8 h, ZnO nanowires with an average diameter of∼40 nm formed on the glass substrate. The as-grown ZnO nanowires were then implanted by 40 keV Co ions with doses of(1–6)×1016cm−2. The implantation was
performed at room temperature by using a tandem accelerator (NEC 9SDH-2). The beam current was kept at 150 nA cm−2 to minimize beam heating.
Thermal annealing processes were performed at 600◦C for 12 h. Such an annealing temperature was chosen because it would not lead to any segregation and clustering of Co atoms in the ZnO nanowires, as was confirmed by previous works [33] and our HRTEM studies. Different annealing atmospheres (either an argon flow of 150 sccm at 1 atm or a high vacuum of 5×10−5Torr) were employed to significantly modify the physical properties of the as-implanted Zn1−xCoxO
nanowires. Annealing in an argon flow not only prevented oxidation of the samples but also led to recovering the lattice ordering. On the other hand, annealing in a high vacuum produced oxygen deficiencies and Zn interstitials, and thus possibly changed the carrier concentration in the nanowires. Moreover, a second annealing in 1 atm of either oxygen or argon atmosphere on those already vacuum-annealed samples was also performed for further investigations of the O vacancy effects on ferromagnetism. All the as-implanted and annealed Zn1−xCoxO nanowires were characterized by using grazing
angle XRD (MAC Science, MXP18), field-emission SEM (JEOL JSM-6330F), and HRTEM (JEOL JEM-2010F). EELS and energy loss elemental mapping of nanowires were take by using an EDX system with a Gatan imaging filter. The magnetic properties of the nanowires were measured by using a Quantum Design SQUID magnetometer.
3. Results and discussion
Most of our as-grown ZnO nanowires lay on the substrate. A top-view SEM image of our nanowires on a glass substrate as schematically displayed in figure 1(a) shows that the as-grown ZnO nanowires are∼40 nm in diameter and 2–3µm in length. Most of the nanowires are fairly straight and they possess a circular cross section. HRTEM was used to examine the crystal structure and atomic columns on a spatial scale down to 0.5 nm. The electron diffraction pattern was used to inspect the periodicity of selected areas. A representative HRTEM image of the as-grown ZnO nanowires 5512
Figure 1. (a) Top-view SEM image of as-grown ZnO nanowires.
(b) HRTEM image and its corresponding electron diffraction pattern (inset) displaying the single-crystalline structure of an as-grown ZnO nanowire.
and its corresponding diffraction pattern with zone axis of [11¯20] are shown in figure1(b). The lattice fringe spacing of 0.52 nm as indicated corresponds to the separation distance between the (001) planes and is in agreement with the lattice constants for the ZnO wurtzite structure. The growth direction of the ZnO nanowire is [0001], and the whole wire comprises a single-crystalline structure.
The as-grown ZnO nanowires were implanted by 40 keV Co ions as depicted in figure2(a). This implantation process for producing Zn1−xCoxO nanowires was chosen because it
ensured a random and uniform distribution of Co ions in the ZnO matrix. In bulk ZnO, the implanted Co ion distribution estimated from the SRIM simulation [34] predicted a peak at 20 nm and a range of∼40 nm. The predicted implantation range, which was near the average diameter of our nanowires, guaranteed a uniform distribution of Co ions in the bombarding direction. The Co concentrations in our nanowires were determined from the EDX results, as will be described below and in figure 3. Alternatively, the thickness of the resulted Zn1−xCoxO nanowire layer can be evaluated in the following
manner. We know that the bombardment of a dose of 1 × 1016cm−2Co ions produced the as-implanted Zn
0.98Co0.02O
nanowires. By assuming all Co ions were implanted in the ZnO matrix without loss, we estimated the thickness of this layer to be ∼120 nm. This evaluated thickness is ∼3 times as large as the implanted Co ion range inferred from the SRIM code simulation. Such a discrepancy could readily result from the geometrical difference between a bulk and a nanowire-formed layer. The whole layer of the ZnO nanowire with a thickness of 4–5µm covered with a very thin Zn1−xCoxO nanowire layer of
∼120 nm is shown in the side-view SEM image in figure2(b).
Figure 2. (a) Schematic illustration of the preparation method for
Co-implanted ZnO nanowires. (b) Side-view SEM image of ZnO nanowires grown on a glass substrate. The estimated thickness of the thin, as-implanted Zn1−xCoxO nanowire-formed layer is depicted in
the image.
Figure 3. (a) EDX spectroscopy of the as-implanted Zn0.89Co0.11O
nanowires. (b) Bright-field image of the as-implanted Zn0.89Co0.11O
nanowires with compositional EELS mapping of the O K edge, Co L edge and Zn L edge.
The Co concentration of the as-implanted Zn1−xCoxO
nanowires was determined from the relative ratio of the Co element among Zn and O elements by using EDX spectroscopy. Measurements on more than 20 different nanowires from the same Co ion-dosed sample were carried
Z Y Wu et al
Figure 4. (a) Room-temperature x-ray diffraction patterns for the
as-grown ZnO nanowires and the as-implanted Zn1−xCoxO
nanowires as indicated. (b) Variation of the a-axis and c-axis lattice constants with Co concentration x .
out to determine the x value. A typical EDX spectrum of the as-implanted Zn0.89Co0.11O nanowire is shown in
figure 3(a). The peaks indicating C and Cu elements came from the background of the copper grid used in the HRTEM measurement. A quantitative analysis of the integrated area in the EDX spectrum reveals the atomic ratio of 11% of Co. The average x values are 4.2±0.9%, 7.9±1.6%, and
11.0±1.8% for Zn1−xCoxO nanowires with doses of 2, 4, and
6×1016cm−2Co ions, respectively. Thesexvalues determined
from the EDX results revealed a fairly linear dependence on the Co doses. Thus, we have approximated the averagex values to be 2, 4, 6, 8, 10, and 11 at.% in our samples dosed with 1, 2, 3, 4, 5, and 6×1016cm−2Co ions, respectively. The
spatial resolution limit of the EDX mapping as was described in our previous work [33] is about 4 nm. The compositional EELS map of the same Zn0.89Co0.11O nanowire is displayed in figure3(b). The bright-field zero-loss EELS mapping with the corresponding image of Co L edge exhibits a uniform distribution of the Co element in the ZnO nanowire. The images of the O K and Zn L edges are also shown in figure3(b) to confirm the data of the EELS map. In the image of the Co L edge, no Co clusters or nanocrystals were found within the spatial resolution limit of about 1.8 nm [35].
The crystal structures of the grown and the as-implanted ZnO nanowires were characterized by grazing angle
Figure 5. Hysteresis loops of the Zn0.89Co0.11O nanowires before and
after annealing in 1 atm argon flow. The data were taken at 2 K.
XRD patterns, and are shown in figure 4(a). An incident angle of 0.5◦ was used to observe the crystal structure of the surface layer of Zn1−xCoxO nanowires. The XRD
pattern for the as-grown ZnO nanowires is also plotted for comparison. The XRD data indicate that all the as-implanted Zn1−xCoxO nanowires are single phased and possess a wurtzite
structure. The ZnO nanowires remain crystalline even under the bombardment from our highest dose of 6×1016cm−2Co
ions. No perceptible peaks showing any second phases of Co clustering or CoO nanocrystals, as were identified in previous results [36], were found in this work. The spatial resolution limit to detect clusters or nanocrystals from the XRD patterns is about 10 nm. Figure4(b) shows a small variation ofaandc
lattice constants for the Zn1−xCoxO nanowires as a function of
thexvalues. We see that changes in thea-axis lattice constant is minute while a tendency of a slight decrease in thec-axis lattice constant is observed. The decrease in c-axis lattice constant is consistent with previous data of polycrystalline Co-substituted ZnO [21] but is in conflict with data of Co-doped ZnO films prepared by pulsed laser deposition [12].
3.1. Structural effects on ferromagnetism
The magnetic properties of Zn1−xCoxO nanowires were
investigated by using a SQUID magnetometer. All of our as-implanted nanowires displayed paramagnetic behaviour at 2 K. That is, there are no detectable hysteresis loops, indicating the absence of any ferromagnetic ordering. Since the mechanism of FM in DMS materials is still unclear, while it is known that different preparation methods can profoundly alter the magnetic behaviours, we were motivated to investigate the structure effect on ferromagnetism in Co-doped ZnO nanowires in this work. The crystal structures of our nanowires were markedly modified after the Co ion implantation process (see below). Thus, we performed thermal annealing of our nanowires in 1 atm argon to restore the crystal order and then measured the magnetizations as a function of the magnetic field. Noticeably, after the argon annealing, we observed profound hysteresis loops at 2 K in all Zn1−xCoxO nanowires. The hysteresis loops for the
as-implanted (closed circles) and then annealed (closed squares) Zn0.89Co0.11O nanowires are shown in figure5for comparison.
Figure 6. (a) SEM image of as-implanted Zn0.94Co0.06O nanowires.
(b) HRTEM image of as-implanted Zn0.89Co0.11O nanowires with
white triangles indicating one type of structural defect, i.e., stacking faults. Inset: an electron diffraction pattern revealing another type of structural defect, i.e., orientational variations.
The appearance of hysteresis loops, a characteristic behaviour of FM, points to domain wall motion in our 40 nm diameter Zn1−xCoxO nanowires. Apart from the nonreversible loops
we can see the linear and reversible component in the field-dependent magnetization. The linear part may originate from some isolated Co ions without ferromagnetic exchange interaction with other magnetic cations. To reduce the effect of the linear component from those paramagnetic impurities we used the same sample for subsequent thermal treatments and magnetization measurements.
Bearing in mind investigating the crystal structure effects on the possible occurrence of FM, the Zn1−xCoxO nanowires,
both before and after annealing in argon, were carefully examined by HRTEM. We first show the SEM and HRTEM images of the as-implanted nanowires in figure6. Right after ion implantation but before argon annealing, the morphology and dimensions of the Zn1−xCoxO nanowires did not change
appreciably, except a bending of the nanowires occurred. The HRTEM image shows a single-crystalline structure with a [0001] growth direction (which is the growth direction for our as-grown ZnO nanowires). The bending effect could also be observed in the [1120]¯ electron diffraction pattern, as shown in the inset of figure6(b) for the as-implanted Zn0.89Co0.11O
nanowires. A small rotation with respect to the zone axis generated deviation angles for the (0110)¯ and (0111)¯ plane sets. The first type of structural defect (called orientational variations) exhibited bending in the SEM image and rotation of plane-set points in the selected area electron diffraction pattern. A second type of structural defect (called stacking faults) was also found, as indicated by the many small triangles in the HRTEM image shown in figure 6(b). The locations of the
Figure 7. (a) Variation of the number of stacking faults per 100 nm
in length along the [0001] growth direction of as-implanted and annealed Zn1−xCoxO nanowires with the Co concentration.
(b) Variation of the deviation angle of the (01¯10) and (01¯11) plane sets along the [11¯20] axis with the Co concentration.
small triangles marking stacking faults were determined from the back and forth fast Fourier transform analysis of the (0001) plane set.
We have quantitatively analysed these two types of structural defect in our nanowires as described below. The numbers of stacking faults in the grown ZnO and the as-implanted Zn1−xCoxO nanowires were counted per 100 nm in
length along the [0001] direction and are plotted as a function of Co concentration in figure 7(a). The number of stacking faults for each Co concentration was evaluated and averaged by counting over 20 Zn1−xCoxO nanowires. For comparison,
the amount of stacking faults in the as-grown ZnO nanowires is also plotted. We found that the Co ion implantation generates a good number of stacking faults. Some of the stacking faults were, however, removed by the thermal annealing process. After 1 atm argon annealing, the number of stacking faults was reduced to the level of 6–7 per 100 nm in length, which was still twice that in the as-grown ZnO nanowires. In particular, the crystalline structure of the Zn1−xCoxO nanowires was now
greatly recovered.
The slight bending of the as-implanted Zn1−xCoxO
nanowires was quantified by deviation angles representing the rotation of plane-set points in the electron diffraction pattern. Figure7(b) shows the deviation angles in degrees as a function of the Co concentration for both the as-implanted and annealed Zn1−xCoxO nanowires. We see that the deviation angles for
the plane sets of (0110) and (01¯ 11) changed by a factor of¯ ∼2 in the as-implanted nanowires, as compared with those in the
Z Y Wu et al
Figure 8. (a) TEM image of Co clusters on surfaces of
carbon-coated ZnO nanowires. Inset: HRTEM image of a Co cluster. (b) EDX spectroscopy of a Co cluster on the surface of
carbon-coated ZnO nanowires.
annealed nanowires. On the other hand, the deviation angle was barely changed for the (0001) plane set. We notice that the argon annealing is much more effective in removing stacking faults than orientation variations. This result suggests that the structure defects due to stacking faults are more detrimental to the occurrence of FM in the as-implanted Zn1−xCoxO
nanowires, as compared with the structural defects due to orientational variations. We also notice that the lattice planes along the [0001] direction are fairly well ordered even after ion bombardment, while those planes in other directions are more disordered.
In addition to the quantitative analysis discussed above, we carefully examined the microstructures of the as-implanted and the annealed Zn1−xCoxO nanowires from the HRTEM
images and the electron diffraction patterns. We found no sign of a second phase due to either Co clustering or CoO nanocrystals in the HRTEM images. Also, no extra diffraction spots other than those characteristic to ZnO have ever been detected from the electron diffraction patterns. Moreover, in order to provide further clarifications, we purposely fabricated carbon-coated ZnO nanowires, as shown in figure 8(a), and then performed Co ion implantation and high-vacuum annealing. In this case, we clearly observed Co metal clustering on the surfaces of the carbon-coated ZnO nanowires. A typical HRTEM image of a 1 nm Co cluster is also shown in the inset of figure8(a). The dotted microstructures observed in HRTEM images were determined to consist of Co atoms, without O element, by using EDX analysis (see figure 8(b)). It should be stressed that when we examined our Zn1−xCoxO
nanowires without carbon coating, we did not find any similar Co clustering on the surfaces.
Our results are consistent with a previous report [37] that the implantation defects were removed upon the annealing procedure and that the annealing does not result into an out-diffusion of magnetic cations. Besides, the results agree well with the concept that a uniform distribution of Co ions in ZnO host favours FM ordering [38,26]. It may imply an absence of FM in some polycrystalline samples [39, 40]. Although a theoretical argument exists that the structural optimization has little effect on the couplings for Zn1−xCoxO [27], we believe
that the crystallinity is crucial for turning on the FM.
3.2. Annealing effects on ferromagnetism
We studied the structure effects on the ferromagnetic ordering through measuring the hysteresis loops in the Zn1−xCoxO
nanowires both before and after annealing in an argon flow. Our results for the structure effects provide important clue for resolving the controversial issue of whether ferromagnetic ordering could occur in crystalline and polycrystalline Co-substituted ZnO [12, 21]. However, as mentioned in the Introduction, ZnO is very sensitive to the ambient air as well as the partial pressure of oxygen. Therefore, we also performed the annealing in a high vacuum of 5×10−5Torr and studied the
magnetizations of the Zn1−xCoxO nanowires. Furthermore, we
then performed a second annealing on those already vacuum-annealed nanowires in either an argon or oxygen atmosphere for further investigations of their magnetic properties.
Our measured hysteresis loops for both the argon- and vacuum-annealed Zn1−xCoxO nanowires are presented in
figure 9(a). The hysteresis loops are much more profound for the x = 0.11 nanowires relative to the x = 0.08 nanowires, clearly demonstrating the formation of enhanced ferromagnetic domains when more Co ions are introduced into the ZnO nanowires. Inspection of figure 9(a) reveals that much stronger FM is found in the vacuum-annealed Zn1−xCoxO nanowires, while the argon-annealed nanowires
revealed weaker FM. From the HRTEM and diffraction pattern analysis, we found that the crystal structure of the vacuum-annealed nanowires reveals no discernible difference from that of the argon-annealed nanowires. Therefore, we believe that another mechanism in addition to the structure effects must contribute to the striking enhancement of FM in the Co-implanted ZnO. Figure9(b) shows a plot of the temperature dependencies of the magnetizations for the vacuum-annealed Zn1−xCoxO nanowires. Noticeably, the magnetization of
the vacuum-annealed nanowires still displays clear hysteresis loops at 10 K (as a matter of fact, a hysteresis loop can even be observed at room temperature, not shown). On the other hand, we point out that the hysteresis loops for the argon-annealed Zn1−xCoxO nanowires quickly diminished with increasing
temperature and totally disappeared above 10 K.
In order to investigate the possible mechanism for the occurrence of enhanced FM in the high-vacuum annealed Zn1−xCoxO nanowires, we also performed a subsequent
second annealing in either an argon or oxygen atmosphere on those vacuum-annealed samples. We found that, after a second annealing in 1 atm argon, both the temperature and magnetic field behaviours of the magnetizations are unchanged. However, if the second annealing is performed in an oxygen atmosphere, the hysteresis loop for the nanowires 5516
Figure 9. (a) Hysteresis loops for argon-annealed and high-vacuum
annealed Zn0.92Co0.08O and Zn0.89Co0.11O nanowires at 2 K. The
magnetization was normalized to that at the field of 1500 Oe for comparison. (b) Hysteresis loops for the high-vacuum annealed Zn0.92Co0.08O nanowires at 2, 5, and 10 K as indicated.
greatly diminishes and the magnetization is notably attenuated (see figure10). The diminished FM in the vacuum-annealed Zn1−xCoxO nanowires followed with an oxygen annealing
immediately implies a decisive role being played by O element in determining the magnetic properties of the nanowires. We believe that the first high-vacuum annealing results in the formation of a good number of O vacancies in the nanowires, while the second annealing in oxygen largely fills those vacancies up again with O. It is established that ZnO is a native n-type material with the dopants generating from either O vacancies or Zn interstitials [41]. Consequently, we may draw the conclusion that the enhanced FM found in the vacuum-annealed nanowires is very likely caused by a change in the electron concentration due to the generation of a large amount of O vacancies. We cannot inspect point defects of O vacancies but can observe planar defects of stacking faults through HRTEM analysis. Taking the results from second annealing together, we conclude that the effect of argon annealing is mainly to remove structure defects only, while high-vacuum annealing causes an extra effect of increasing the number of O vacancies in ZnO.
To trace the origin of the enhanced FM found in the high-vacuum annealed nanowires, we carried out measurements of EELS spectra in the as-implanted, argon-annealed, and high-vacuum annealed Zn1−xCoxO nanowires. The technique of
L3/L2 intensity ratio correlating the EELS spectra with the
valence states [42, 43] was used to determine the change of the oxidation state of Co ions in our nanowires. To determine
Figure 10. Hysteresis loops for the same Zn0.89Co0.11O nanowires
after a first annealing in high vacuum and then a second annealing in oxygen. The data were taken at 2 K.
the numerical ratio of L3/L2, a double arctangent function was
used as a background function for subtracting the background intensity. The intensity ratios thus determined for the as-implanted, argon-annealed, and vacuum-annealed Zn1−xCoxO
nanowires are 3.5, 3.4, and 3.1, respectively. We see that the valence state of Co ions in the argon-annealed nanowires is barely changed from that in the as-implanted nanowires.
The results of annealing effects, pointing out the important roles played by defects, are consistent with recent observation of reversible 300 K FM ordering in Co2+:ZnO thin film [44] and thin film grown by ultrasonic-assisted solution chemical vapour deposition [45]. Double-exchange mechanisms [46] from either Zn interstitials or O vacancies in our observations are theoretically supported [29,27], and it will thus provide adequate description of the FM in the Zn1−xCoxO system.
4. Summary
In this work, we fabricated Zn1−xCoxO nanowires by using
the standard ion implantation technique. The Co ions were uniformly distributed in nanowires, as was inspected by using EELS map under a spatial resolution of ∼1.8 nm. In addition, no second phase of either Co clustering or CoO nanocrystals was found in the nanowires by using HRTEM within our spatial resolution of ∼0.5 nm. The c
lattice constant of the Zn1−xCoxO nanowires decreases with
increasing Co concentrationx. The as-implanted nanowires showed no sign of ferromagnetic ordering at 2 K. In the as-implanted Zn1−xCoxO nanowires, two types of structural
defect (i.e., stacking faults and orientational variations) were found. After annealing in argon, the removal of structural defects and the recovery of structure order introduced FM in the Zn1−xCoxO nanowires, which displayed noticeable
hysteresis loops up to 10 K. We found that the structure defects due to stacking faults are more detrimental to the occurrence of FM. For comparison, we performed Co implantation on carbon-coated ZnO nanowires and found clear Co clusters in HRTEM images. Most noticeably, thermal annealing in high vacuum caused a largely enhanced FM ordering in the Zn1−xCoxO nanowires. A subsequent second annealing on the
vacuum-annealed nanowires was performed to illustrate that O vacancies play decisive roles in determining the occurrence of ferromagnetism. EELS spectra revealed a slight shift toward a
Z Y Wu et al
higher oxidation state for Co ions in the high-vacuum annealed nanowires, relative to that in the as-implanted and the argon-annealed Zn1−xCoxO nanowires.
Acknowledgments
This work was supported by the National Science Council of ROC under Grant Nos NSC 2112-M-009-038, NSC 93-2120-M-009-009 and NSC 94-2120-M-009-010, and by the MOE ATU Program.
References
[1] Furdyna J K 1988 J. Appl. Phys.64 R29
[2] Wolf S A, Awschalom D D, Buhrman R A, Daughton J M, von Moln´ar S, Roukes M L, Chtchelkanova A Y and Treger D M 2001 Science294 1488
[3] Story T, Galazka R R, Frankel R B and Wolff P A 1986 Phys.
Rev. Lett.56 777
[4] Munekata H, Ohno H, von Molnar S, Segm¨uller A, Chang L L and Esaki L 1989 Phys. Rev. Lett.63 1849
[5] Ohno H, Shen A, Matsukura F, Oiwa A, Endo A, Katsumoto S and Iye Y 1996 Appl. Phys. Lett.69 363
[6] Matsukura F, Ohno H, Shen A and Sugawara Y 1998 Phys.
Rev. B57 R2037
[7] Dietl T, Ohno H, Matsukura F, Cibert J and Ferrand D 2000
Science287 1019
[8] Heo Y W, Park S J, Ip K, Pearton S J and Norton D P 2003
Appl. Phys. Lett.83 1128
[9] Li Z Q, Zhang D X and Lin J J 2006 J. Appl. Phys.99 124906
[10] Huang M H, Mao S, Feick H, Yan H, Wu Y, Kind H, Weber E, Russo R and Yang P 2001 Science292 1897
[11] Heo Y W, Norton D P, Tien L C, Kwon Y, Kang B S, Ren F, Pearton S J and LaRoche J R 2004 Mater. Sci. Eng. R47 1
[12] Ueda K, Tabata H and Kawai T 2001 Appl. Phys. Lett.79 988
[13] Lee H J, Jeong S Y, Cho C R and Park C H 2002 Appl. Phys.
Lett.81 4020
[14] Rode K, Anane A, Mattana R, Contour J P, Durand O and LeBourgeois R 2003 J. Appl. Phys.93 7676
[15] Prellier W, Fouchet A, Mercey B, Simon C and Raveau B 2003
Appl. Phys. Lett.82 3490
[16] Ramachandran S, Tiwari A and Narayan J 2004 Appl. Phys.
Lett.84 5255
[17] Wu J J, Liu S C and Yang M H 2004 Appl. Phys. Lett.85 1027
[18] Kittilstved K R, Norberg N S and Gamelin D R 2005 Phys. Rev.
Lett.94 147209
[19] Chen J J, Yu M H, Zhou W L, Sun K and Wang L M 2005
Appl. Phys. Lett.87 173119
[20] Sati P et al 2006 Phys. Rev. Lett.96 017203
[21] Risbud A S, Spaldin N A, Chen Z Q, Stemmer S and Seshadri R 2003 Phys. Rev. B68 205202
[22] Kobayashi M et al 2005 Phys. Rev. B72 201201
[23] Wi S C et al 2004 Appl. Phys. Lett.84 4233
[24] Norton D P et al 2003 Appl. Phys. Lett.83 5488
[25] Park J H, Kim M G, Jang H M, Ryu S and Kim Y M 2004
Appl. Phys. Lett.84 1338
[26] Martinez B, Sandiumenge F, Balcells L, Arbiol J, Sibieude F and Monty C 2005 Appl. Phys. Lett.
86 103113
[27] Sluiter M H F, Kawazoe Y, Sharma P, Inoue A, Raju A R, Rout C and Waghmare U V 2005 Phys. Rev. Lett.
94 187204
[28] Petit L, Schulthess T C, Svane A, Szotek Z,
Temmerman W M and Janotti A 2006 Phys. Rev. B
73 045107
[29] Coey J M D, Venkatesan M and Fitzgerald C B 2005 Nat.
Mater.4 173
[30] Sato K and Katayama-Yoshida H 2001 Japan. J. Appl. Phys.
40 L334
[31] Venkatesan M, Fitzgerald C B, Lunney J G and Coey J M D 2004 Phys. Rev. Lett.93 177206
[32] Spaldin N A 2004 Phys. Rev. B69 125201
[33] Jian W B, Wu Z Y, Huang R T, Chen F R, Kai J J, Wu C Y, Chiang S J, Lan M D and Lin J J 2006 Phys. Rev. B
73 233308
[34] Ziegler J F and Biersak J Phttp://www.srim.org
[35] Egerton R F and Crozier P A 1997 Micron28 117
[36] Kolesnik S, Dabrowski B and Mais J 2004 J. Appl. Phys.
95 2582
[37] Ronning C, Gao P X, Ding Y, Wang Z L and Schwen D 2004
Appl. Phys. Lett.84 783
[38] Martinez B, Sandiumenge F, Balcells Ll, Arbiol J, Sibieude F and Monty C 2005 Phys. Rev. B72 165202
[39] Lawes G, Risbud A S, Ramirez A P and Seshadri R 2005 Phys.
Rev. B71 045201
[40] Bouloudenine M, Viart N, Colis S, Kortus J and Dinia A 2005
Appl. Phys. Lett.87 052501
[41] Look D C, Hemsky J W and Sizelove J R 1999 Phys. Rev. Lett.
822552
[42] Wang Z L, Yin J S and Jiang Y D 2000 Micron31 571
[43] Pearson D H, Fultz B and Ahn C C 1988 Appl. Phys. Lett.
53 1405
[44] Schwartz D A and Gamelin D R 2004 Adv. Mater.16 2115
[45] Khare N, Kappers M J, Wei M, Blamire M G and MacManus-Driscoll J L 2006 Adv. Mater.18 1449
[46] Garc´ıa M A et al 2005 Phys. Rev. Lett.94 217206
