Growth and characterisation of cobalt doped zinc oxide 3

3.2 Sample characterization techniques The samples had been characterized by XRD, TEM & EELS, XPS, AES & Ultra-violet photoemission spectroscopy UPS, SQUID, optical transmission spectro

3.1 Sample preparation

3.1.1 Thin film deposition

Zn1–xCoxO (x = 0.05 – 0.33) thin films were deposited on (0001) sapphire (

-Al2O3) substrates with a size of 1 × 1 cm, in a high vacuum chamber with a base pressure < 1 × 10–7 Torr, using a combination of radio-frequency (RF) and direct current (DC) magnetron sputtering Sintered ZnO, Al2O3 and Co materials were used

as the sputtering sources for ZnO, Al and Co, respectively The Al2O3 target was used

to dope the samples with Al for better conductivity All the samples were sputtered in

an atmosphere of pure Ar gas at a pressure of 5 mTorr Prior to deposition, the substrates were first degreased using acetone and then reversely sputtered by 20 mTorr

Ar gas in the pre-cleaning chamber A series of experiments had been carried out to optimize the substrate temperature (from room temperature to 600 oC) and sputtering powers for ZnO (50 W – 200 W) and Al2O3 (20 W – 50 W) Films of high structural quality and resistivity of 1.3 m cm were obtained at a substrate temperature of 500oC, ZnO sputtering power of 150 W and Al2O3 sputtering power of 30W Under these conditions, the deposition rate of ZnO:Al was ~ 4.8 nm/min Subsequently, Co was

doped into ZnO:Al by using the above-mentioned optimum deposition condition, while varying either the Co sputtering power from 3 to 50 W for co-doped samples or the Co sputtering duration from 10 to 98 s for -doped samples at a sputtering power of 10 W For the co-doped samples, Co was added to ZnO:Al by co-sputtering Co with ZnO and

Al2O3 On the other hand, in -doped samples, Co was doped “digitally” into the ZnO:Al host matrix with a nominal thickness of 0.1 nm, 0.25 nm, 0.5 nm and 1 nm, respectively As the actual thicknesses might be different from the nominal ones because the latter were determined from the thickness of Co deposited at room temperature, hereafter the Co sputtering durations were used instead of nominal thicknesses in the discussion of the second group of samples which were 10 s, 25 s, 49

s and 98 s, respectively The δ-doping was repeated for 60 cycles with each cycle

consisting of a 2.38 nm ZnO:Al spacer and a δ-doped Co layer The thickness of Co

was chosen such that the average Co composition will be roughly the same as those of four co-doped samples obtained at a Co sputtering power of 3 W, 8 W, 16 W and 32 W, respectively Table 3-1 summarizes the details of all the samples that have been investigated systematically in this work The thickness of all the films prepared is around 200 nm

Table 3-1 Details of the samples under study

Co composition or sample structure

3.1.2 Fabrication of Hall bars

For electrical characterizations, Hall bars with a length of 324 m and a width of

80 m were fabricated for each sample using a direct laser writer (see Fig 3-1) A 3-1 eight-contact Hall bar configuration was used to measure both the longitudinal and Hall voltages The same sample had also been used to measure the MR and dynamic conductance curves As shown in Fig.3-1, contacts 1 and 2 were current probes, and 3,

1-3-4, 5 and 6 were voltage probes (V3,4, V5,6: longitudinal voltage; V4,5: Hall voltage)

The process flow of Hall bar fabrication is illustrated in Fig.3-2 A MicroTech laser writer was used to pattern both the Hall bar and electrode pads An Al2O3 layer was used as a hard mask to pattern the Hall bar via ion-milling Al was used as the electrode material to obtain an ohmic contact with ZnO An additional layer of Au was used to prevent the Al electrodes from oxidation and also for ease of bonding the lead wires

Figure 3-2 Process flow of Hall bar fabrication

3.2 Sample characterization techniques

The samples had been characterized by XRD, TEM & EELS, XPS, AES & Ultra-violet photoemission spectroscopy (UPS), SQUID, optical transmission spectroscopy, MCD, MR and Hall effect As most of them are standard techniques, in what follows those which were directly relevant to DMS or more specifically to ZnO:Co are described briefly

Al2O3 Substrate

Resist

c) Write Hall bar pattern

3.2.1 Optical transmission spectroscopy

As all the samples in this study were grown on transparent sapphire substrates, optical transmission spectroscopy is one of the most convenient yet powerful techniques to study ZnO:Co Here, our interest was in two wavelength regimes, i.e., the near bandgap region of ZnO:Co and the visible region (1.8-2.4 eV) due to d-d transitions of Co2+ ions substituting Zn in ZnO The former provides useful information on how the bandgap varies with Co composition at low doping and whether there was any new phases emerging at high doping levels In the latter case, optical absorption of Co2+ in ZnO shows characteristic d-d transition peaks for tetrahedrally coordinated Co2+ in ZnO at 571, 618 and 665 nm for 4A2(F) → 2A1(G),

4A2(F) → 4T1(P), and 4A2(F) → 2E(G) transitions, respectively,1,2 as shown in Fig 3(a) Fig 3.3(b) shows a typical optical absorption curve with the d-d transitions as marked Thereafter, the observation of these dips in the optical absorption curves had been used frequently in literature to provide “evidence” that Co has substituted Zn in ZnO However, as will be pointed out in this thesis, the observation of these characteristic peaks did not rule out the possibility of the formation of other phases; instead it only meant that, in some regions, Co substituted Zn in the host matrix It should not be used to imply that the Co2+ observed was indeed responsible for the ferromagnetic behaviour observed in magnetic measurements

3-Figure 3-3 (a) Visible absorption bands of ZnO:Co 2+ at 4.2 K[After P Koidl, 1977, Ref 1]; (b) Typical optical absorption curve in Co-doped ZnO with d-d transitions marked [After I Ozerov, 2005, Ref 3]

3.2.2 Magnetic circular dichroism (MCD)

Although optical transmission could be used to show the replacement of Zn by

Co through d-d transitions, it could not be used to study magnetic properties of doped ZnO samples The MCD method, on the other hand, is a useful technique to characterize this material as discussed below The MCD is a magneto-optical technique which detects the difference in optical absorption between right and left circularly polarized light with the application of a magnetic field in the light propagation direction (see Fig.3-4).4 Its basic principle is as follows

Co-Figure 3-4 Schematic diagram of incident, reflected and transmitted light of a MCD setup

(a)

(b)

In the absence of a magnetic field, the transmitted light intensity is given by,

I = I o exp [-k(E).L] (3.1) where k (E) is the optical-absorption coefficient at energy E , Io the input light intensity and L the sample thickness When a magnetic field is applied, it causes Zeeman splitting of the energy levels, leading to a difference in optical-absorption coefficient between light with two different polarizations (σ+ and σ-), resulting in the MCD effect

The magnitude of MCD, expressed as the angle change (θ) per unit light

propagation distance is given by,

)2()(),

(4

E k E k where k

k − = ±∆

π

where ∆E is the Zeeman splitting energy given by a first order perturbation treatment

of the sp-d interaction, and k+ and k- are the optical absorption coefficient for the σ+

and σ- polarizations, respectively As ∆E is small, Eq (3.2) can be re-written as

following:

dE

E dk

E ( )4

180 ∆

−

=π

∆E is small; but, in a DMS system, the ∆E is greatly enhanced due to the sp-d

interactions As the MCD signal is proportional to the average magnetization, one can easily observe the hysteresis, if any, in DMSs

In the MCD measurements out of ZnO:Co, discussion will be focused in two energy regions, i.e., the absorption edge (near 3.4 eV) and the d-d transition region (1.8

- 2.0 eV) Shown in Fig.3-5 are some typical MCD spectra for ZnO doped with different TMs For ZnO without any TM doping, a weak MCD signal can be observed near the absorption edge at 3.4 eV, although not seen clearly in the figure When Sc, Ti,

V and Cr are introduced into the ZnO matrix, there is no observable change in the MCD spectra, indicating the absence of sp-d exchange interactions in these TM-doped ZnO In comparison, for Mn, Co, Ni and Cu doped ZnO, a clear MCD signal can be observed near the absorption edge at 3.4 eV, indicating the presence of strong sp-d interaction In addition to this pronounced MCD signal, small MCD structures have also been observed near 2 eV in Co-doped ZnO, which correspond to the fingerprints

of the d-d inter-ionic transitions of tetrahedrally coordinated Co2+ as determined from the optical transmission studies

Figure 3-5 MCD spectra of Zn1-xTMxO at 5K for various transition metal doping [After K Ando, 2001, Ref 5]

The biggest advantage of using MCD over other magnetometer-based techniques is that it is energy sensitive, thus enabling the measurement of magnetic hysteresis curves at different photon energies The typical M-H curves determined by MCD are shown in Fig 3-6 In this specific case, there is no hysteresis in pure ZnO (Fig 3-6(a)); though a clear hysteresis is observed in Co-doped ZnO (Fig 3-6(b)) However, as will be discussed in this thesis, the observation of a hysteresis by MCD does not necessarily mean that the material under study is ferromagnetic It will be shown how the MCD taken at different photon energies can help establish the mechanism of ferromagnetic ordering in Co-doped ZnO, in combination with other techniques

Figure 3-6 MCD hysteresis loop obtained at 300 K with energy of 3.4 eV for (a) pure ZnO and (b) doped ZnO [After J R Neal, 2006, Ref 6]

Co-3.2.3 Hall effect

Next, turning the focus to the Hall effect, which is often used in semiconductor

research to determine carrier concentration, mobility and resistivity The fundamental

principle of the Hall effect lies in the Lorentz force exerted on moving charges by an applied magnetic field A magnetic field applied perpendicularly to a current flow would cause a deflection of charge carriers, leading to the buildup of a voltage, i.e., Hall voltage, across the sample, as shown in Fig 3-7(a) The Hall voltage VH is

proportional to the applied magnetic field B and the current I, i.e., VH = -IB/ned, where

d is the sample thickness, e is the electron charge and n is the carrier concentration The corresponding Hall resistance can be written as RHall= (RH/d) B, here RH is the Hall coefficient

The Hall effect in a magnetic materials is more complicated as compared to the non-magnetic counterpart The Hall resistance in a ferromagnet is generally given

by

⊥+

d

R B d

R

where Ro is the ordinary Hall coefficient (equivalent to RH in the non-magnetic case),

Rs the anomalous Hall coefficient and M⊥ the magnetization of sample in the direction which is perpendicular to the current Here, the first term is due to ordinary Hall effect (OHE), while the second term originates from the so-called AHE The AHE has two

proposed mechanisms: skew scattering and side-jump, as illustrated in Fig 3-7 (b) In skew scattering, when electrons are scattered by a magnetic impurity, the angle of trajectory of the deflected electrons is spin-dependent On the other hand, in the side jump, there is a lateral displacement of wave-packet after scattering In magnetic materials, these cause a net spin current and a transverse component in the charge current, which leads to the AHE

Figure 3-7 (a) Schematic of Hall effect, (b) Schematic diagram of skew scattering and side jump contributing to the AHE

The presence of AHE is often considered as one of the strong evidences for intrinsic ferromagnetism in DMSs (see a typical result for Co-doped TiO2 shown in Fig 3-8). 7 - 10 However, considering the fact that the AHE has also been reported in ferromagnetic clusters,11 granular materials12 - 14 and inhomogeneous DMS in the hopping transport regime,15 the observation of AHE alone cannot support the claim that the DMS under study is a ferromagnet of intrinsic origin, unless it is correlated with ferromagnetism observed by other means and furthermore, secondary phases and precipitates must be absent in the sample As will be discussed later in this thesis, in the present case, Zn1-xCoxO samples with x < 0.2 show only OHE As Co concentration increases, the AHE appears in samples with x ≥ 0.25 The onset

composition at which the AHE starts to appear also coincides with the composition at which the Co-rich phase becomes dominant and Co clusters start to appear Therefore, the AHE in this case is due to extrinsic origin instead of intrinsic ones

Figure 3-8 Magnetic-field dependence of (a) magnetization and (b) Hall resistivity ( xy) for rutile Co:TiO2– at room temperatures [After J S Higgins, 2004, Ref 8]

There are two commonly used sample geometries in the Hall effect measurement: van der Pauw and Hall bar, with the commonly used geometries as shown in Fig 3-9 In the van der Pauw geometry, four ohmic contacts are formed on

an arbitrary shaped sample, as shown in Fig 3-9 (a).16 The sample has to have uniform thickness, with square and circle being the most commonly used van der Pauw geometries Also, in this geometry, the contacts are formed on the circumference of the sample and the contact pad area should be relatively small compared to the sample surface area In order to measure Hall resistance or resistivity, magnetic field is applied perpendicular to the sample and two sets of voltages need to be taken to determine the sheet resistance of the sample One resistance measurement is carried out along a vertical edge of the sample, while the other is taken along a horizontal edge More measurements or a negative magnetic field can be applied in order to average the results In the van der Pauw geometry, calculations do not rely on contact spacing, but only on sample thickness, making sample preparation easy However, there are two main disadvantages for the van der Pauw geometry Firstly, error due to contact area size and their placement can be quite significant.17 And secondly, as a few voltage readings need to be taken before calculations can be carried out, it takes a slightly longer time to finish the measurements

In the Hall bar geometry, six to eight contacts are required and measurements are sensitive to sample geometry, as shown in Fig 3-9 (b) Current flows through the sample along the long axis of the rectangular-shaped sample, with external magnetic field applied perpendicular to the sample As current is passed through the sample in known dimensions, only one voltage reading is required to in calculations of carrier concentrations, mobility and resistivity This enables readings to be completed in a much shorter time compared to the van der Pauw geometry; however, sample