Growth and characterisation of cobalt doped zinc oxide 5

5-3 b and c, showed a well defined perpendicular anisotropy similar to those observed by Dinia 1 and Rode.2 In both Dinia’s and Rode’s Zn0.75Co0.25O samples, the presence of Co clusters

are also often used to characterize the homogeneity, blocking temperature and T c of the material In addition to magnetometer based characterizations, the magnetic properties

of a DMS material can also be evaluated using techniques such as MCD and AHE In this chapter, the results obtained by SQUID and MCD are presented The results obtained by AHE will be discussed in the next chapter together with the transport measurement results

5.2 Characterization by SQUID

The M-H loops had been measured using SQUID for most of the samples in the temperature range of 10 to 300K For the Co3W sample, measurements had also been performed at 2 K and 5 K All the samples measured showed the presence of ferromagnetism, varying from weak ferromagnetism embedded in a strong paramagnetic phase to ferromagnetism with clear hysteresis loop as the Co content increased Note that the strong diamagnetism from the substrate has been subtracted out from the raw data for all results unless it is indicated otherwise Also note that,

throughout this thesis, “ferromagnetism” and “presence of a clear hysteresis in the M-H loop” were used inter-changeably

Figure 5-1 M-H curves (in-plane) for co-doped sample Co3W (Zn 0.95 Co 0.05 O) at 2, 5, 100, 200, 300 and

400 K (a) as determined, (b) corrected for substrate effect.

Before proceeding to discuss the magnetic properties of samples with different

Co compositions, discussions would start with the Co3W (x = 0.05) sample Shown above in Fig 5-1 are both (a) the raw magnetic moment measured by SQUID and (b)

on the temperature As the diamagnetism was almost independent of temperature, all the data had been corrected by subtracting the diamagnetism at 300 K As it is shown in Fig 5-1 (b), the paramagnetic signal increased with decreasing the temperature The paramagnetism might have originated from other Co atoms distributed in the host matrix or from the defects

The in-plane and out-of-plane initial M-H curves for Co3W sample at different temperatures are shown in Fig 5-2 From the initial curves, it was observed that the in-plane anisotropy was more pronounced as temperature increased from 2 K to 50 K In the out-of-plane initial M-H curves (Fig 5-2 (g) and (h)), also observed as temperature increased, there was a change in initial M-H shapes when moving from 2 K to 50 K The changes in M-H curve shapes could be an indication of different origins of ferromagnetism, or multi-phase present in this material, which would be further discussed in transport studies In Fig 5-2 (b) and (d), the in-plane anisotropy observed

at 2 and 50 K agreed well with the observation by Sati et al.3, who had observed a strong anisotropy for Co2+ ions in the ZnO host matrix for very low Co concentration of

x = 0.003 – 0.005 Sati interpreted it as a signature of intrinsic ferromagnetism, where

Co2+ ions in the ZnO lattice were coupled ferromagnetically However, for the lightly doped samples in this thesis, it was still not possible to pinpoint the origin of ferromagnetism from M-H curves alone

Increasing the Co content near the onset of secondary phase formation, x = 0.24 (Fig 5-3 (a)), a switch from out-of-plane anisotropy to in-plane anisotropy as temperature decreased from 300 K to 5 K was observed As the temperature was decreased to 5 K, not much change was observed in the M-H curves in-plane, but a sudden increase in coercivity was observed in the out-of-plane curve This increase was much larger in magnitude as compared to the Co3W sample, but their ferromagnetic origin up to this point was still not clear and would be further explored using other characterization methods

The M-H curves for samples with x > 0.25, Fig 5-3 (b) and (c), showed a well defined perpendicular anisotropy similar to those observed by Dinia 1 and Rode.2 In both Dinia’s and Rode’s Zn0.75Co0.25O samples, the presence of Co clusters had been eliminated via XRD studies and the magnetic anisotropy was attributed to intrinsic properties of Co2+ Although the results obtained were similar to those obtained by Dinia’s and Rode’s, whom believe that the ferromagnetism was intrinsic, the origin of the ferromagnetism of the films studied in this study was not intrinsic in nature This was probably due to the presence of secondary phases as observed from the structural studies and this would be further characterized by transport studies

Figure 5-3 M-H curves (in-plane and out-of-plane) for co-doped samples (a) Co20W (Zn 0.76 Co 0.24 O), (b)

Co 32W (Zn 0.71 Co 0.29 O), (c) Co45W (Zn 0.70 Co 0.30 O) at 10 and 300 K

(a)

(b)

(c)

Fig 5-4 (a) compared the in-plane M-H loops for samples A, B, D, and H at room temperature According to theoretical studies, it had been predicted that the magnetism of the lightly co-doped ZnO samples should be weak but yet ferromagnetic

at room temperature Experimental results from various groups had also made use of this weak magnetism at room temperature, coupled with good structural properties without secondary phases, as an indication of DMS material However, more studies need to be carried out, particularly transport studies to confirm their true origin As the

Co composition was increased, especially passes the onset of secondary phase formation, the shapes of the M-H curves evolved from a weakly ferromagnetic character

to one with a 100 fold increase in magnitude of magnetization and well defined coercivity Caution should again be stressed that presence of ferromagnetism in any film should not be used to conclude that the material was an intrinsic DMS In fact, the magnetic properties of the heavier co-doped samples were dominated by secondary phases, as confirmed from structural studies, rather than behaving as an intrinsic DMS phase

Before reaching the onset of secondary phase or cluster formation, the saturation magnetization of the sample was ~ 0.01 µB/Co atom However, even after passing the onset composition, a further increase of Co composition would not lead to a constant increase of saturation magnetization, which was always below 1 µB/Co atom The rather low value of saturation magnetization suggested that the Co atoms incorporated in the host matrix are not all “magnetically active”, due to presumably the formation of anti-ferromagnetic phases The coercivity, however, was considerably larger than those reported by other groups (typically around 50-200Oe1,10,11), after the Co composition exceeded the onset value of secondary phase formation This agreed well with the

observation in XRD studies that secondary phases and Co clusters started to from at x > 0.25

Looking at the overall trend of magnetic moment and coercivity with Co composition, as shown in Fig 5-4 (b) and (c), it was interesting to observe that both the magnetic moment and coercivity initially increased slowly and when it reached the onset of secondary phase formation, increased sharply, followed by a dip around x = 0.28 The initial magnetic character, in particular the slight increase of coercivity, should be due to isolated Co ions with large anisotropy.3 Weak magnetism of these films was observed as Co was soluble in the ZnO host matrix and also due to the absence of ferromagnetic nanoclusters However, as the Co content was increased to x = 0.25, there was an onset of secondary phase formation and the magnetic moment increased sharply which was due to the presence of ferromagnetic secondary phases In addition, the appearance of a local minimum for coercivity was due to the switching over from in-plane easy-axis anisotropy to out-of-plane easy-axis anisotropy As the Co content was further increased, magnetism and coercivity reached a maximum and then decreased again This could be the indication of dominance of different ferromagnetic secondary phases as Co content increased The sudden increase of the coercivity at x > 0.25 implied that the Co particles started to form a mutually connected magnetic network due to presumably spinoidal decomposition.4 Of course, in addition to Co clusters; there could also be the formation of other secondary phases

Figure 5-4 (a) In-plane M-H curves of co-doped Co3W (Zn 0.95 Co 0.05 O), Co 8W(Zn 0.86 Co 0.14 O), Co 15W (Zn 0.80 Co 0.20 O) and Co 32W (Zn 0.71 Co 0.29 O) at room temperature, (b) Saturation magnetization as a function of Co composition at room temperature, (c) Coercivity as a function of Co composition at 10,

150 and 300 K

0 0.05 0.10 0.15 0.20 0.25 0.30 0.35

0 0.05 0.10 0.15 0.20 0.25 0.30 0.35

As it was discussed in last chapter, the secondary phases could be in the form of

Co, CoO, ZnCo2O4 and/or Co3O4 Bulk CoO, in rock-salt structure, is a known antiferromagnet with a Néel temperature of 297 K, but small CoO nanoparticles can exhibit ferromagnetic behaviour due to frustrated surface spins.5 Another Co oxide,

Co3O4 is also antiferromagnetic, but in the nanoparticle form, can be ferromagnetic below 25 K.6 In a Zn:Co system, ZnCo2O4 can exist as a paramagnetic phase when it is

an n-type semiconductor and ferromagnetic when it is a p-type semiconductor.7 In addition, Co nanoclusters, with a size larger than the superparamagnetic limit, will exhibit ferromagnetic properties However, the measured M-H loops alone could not tell whether the origin of the ferromagnetism was from ZnO:Co system or from these secondary phases mentioned, as the M-H loops obtained were a summation of magnetisation data from all the different phases It is thus crucial to correlate the data obtained with those of different characterization methods

To further study the magnetic properties of the films, FC and ZFC measurements were carried out at 10 – 400 K The curves, as shown in Fig 5-5 (a), (b)

and (c) showed that the T c of all the samples were above 400 K, agreeing well with results obtained by other groups which are listed in Table 5-1 Different magnetic fields were applied for different samples to allow determination of curves while eliminating noise from the data Thus, the magnitude of magnetic moment and also ZFC-FC dependence on magnetic field will not be discussed in this section

As Co clusters were not seen by TEM in the lightly doped sample, the unusually

high T c was likely due to the uncompensated surface spins of antiferromagnetic clusters because the latter has a Neel temperature far around room temperature.5 As Co content

observed, an indication of magnetic clusters.15 The hump around 200 – 300 K, observed

in Fig 5-5 (c), could be an indication of presence of antiferromagnetic CoO, which would be discussed further in the chapter

Although all the samples were ferromagnetic based on the SQUID results, there was no direct evidence to show that the ferromagnetism was of intrinsic nature, i.e., originating from carrier induced ferromagnetic interactions among the Co2+ ions which substitute Zn The results obtained by SQUID were likely a summation of contributions from various sources including isolated Co2+ ions, ZnO:Co ferromagnetic clusters, partially compensated surface spins of antiferromagnetic clusters for lightly co-doped samples and secondary phase clusters for heavily co-doped samples

(a)

(b)

(c)

Table 5-1 T c of various ferromagnetic ZnO:Co samples

Co content (x) Growth method T c (K) Remarks Reference

0 x 0.25 PLD > 350 Ferromagnetic but origin

not confirmed, secondary phase when x > 0.25

8

Co2+

9

0.05, 0.15, 0.25 PLD 280 - 300 Samples with higher

carrier concentration are ferromagnetic

10

Co2+, no metallic Co or CoOx particles

11

0.035 – 0.115 Sputtering > 350 RKKY not applicable, Co

cluster ruled out, indirect exchange interaction proposed

15

5.3 Characterization by MCD

The discussion will now be focused on the results obtained by MCD The measurements were concentrated on two co-doped samples, lightly doped sample Co 8W (x = 0.137) and heavily doped sample Co 32W (x = 0.289), and one delta-doped sample, Co 98s ([ZnO:Al (2.38 nm)/Co(1.0 nm)]×60) For the MCD spectra of ZnO:Co, discussions were concentrated on the absorptions at two particular wavelength regions, one at the band edge (near 3.4 eV) and the other at the d-d transition regime (near 2 eV)

As shown in Fig 5-6 (a)-(c)), the MCD spectra for all the three samples exhibited rather broad peaks in the above regions The broadening of the peaks suggested that the films were highly non-uniform at microscopic scale which could be seen clearly in low magnification TEM images for the heavily doped samples For the lightly co-doped sample, paramagnetic behaviour dominated the MCD spectrum obtained

Fig 5-7 (a) and (b), shows the temperature dependence of MCD for heavily doped samples There seems to be no temperature dependence for these samples, as they had passed the onset of secondary phase formation, indicating that they were less influenced by sp-d interaction As there were no noticeable structures around 2.0 and 3.4 eV, even as temperature was varied, the magnetic properties of this material had no correlation with a DMS form of ZnO:Co This result agreed well with the structural and SQUID results as determined above Thus, magnetic properties for these heavily doped samples were likely not due to sp-d interactions, but more likely from extrinsic origins

-20 -10 0 10 20 30 40

Figure 5-6 MCD spectra of co-doped (a) Co8W (Zn 0.86 Co 0.14 O), (b) Co32W (Zn 0.71 Co 0.29 O) and δ-doped

(c) Co98s ([ZnO:Al (2.38 nm)/Co(1.0 nm)] ×60) at 6 K.

(a)

(c) (b)

-300

0300

-700

-350

0350

Figure 5-7 MCD spectra of co-doped (a) Co32W (Zn 0.71 Co 0.29 O) and δ-doped (b) Co98s ([ZnO:Al (2.38

nm)/Co(1.0 nm)] ×60) at 6 and 300 K

From the hysteresis curve carried out on the lightly co-doped sample near the

d-d transition wavelength, Co8W was purely paramagnetic in nature This contrad-dicted-d the SQUID measurement results where the same sample was found to be weakly ferromagnetic Again, it was difficult to have a direct comparison because the SQUID picked up contributions from all available sources, whereas the MCD only picked up

(a)

(b)

Figure 5-8 Hysteresis curve determined by MCD and SQUID for co-doped Co8W (Zn 86 Co 0.14 O) at 6 and

10 K respectively

For the heavily co-doped sample, sample Co 32W, the MCD showed that ferromagnetic behaviour was dominant Figures 5-9(a) and (b) showed the MCD hysteresis curves of sample Co 32 W obtained at different photon energies, 756, 425,

343 and 325 nm (1.64, 2.92, 3.61 and 3.83 eV), which was also marked in the transmission spectra shown in Fig 4-15 by the “*” symbol, at 6 K and 300 K, respectively Figure 5-9(c) and (d) showed the MCD hysteresis curve of its delta-doped counterpart The MCD curves, for both co-doped and delta-doped samples, were strongly dependent on the photon energy, confirming the inhomogeneous nature of the samples which consisted of ferromagnetic regions of different phases Although the origin of ferromagnetism for all the curves at different photon energies could not be pinpointed, as it would be discussed later in transport measurements, only the MCD curve taken at 2.92 eV (425 nm) for the co-doped sample agreed well with the