Growth and characterisation of cobalt doped zinc oxide 6

CHAPTER 6 ELECTRICAL TRANSPORT PROPERTIES OF Co-DOPED ZnO 6.1 Introduction As discussed in Chapter 3, all the samples had been fabricated into Hall bars so as to carry out systematic s

CHAPTER 6 ELECTRICAL TRANSPORT PROPERTIES OF Co-

DOPED ZnO

6.1 Introduction

As discussed in Chapter 3, all the samples had been fabricated into Hall bars so

as to carry out systematic studies on the transport properties of Co-doped ZnO The electrical transport measurements that had been carried out include temperature-dependence of resistivity, carrier density, differential conductance, MR and Hall effect The differential conductance measurement for DMS is new and is only performed for the first time in this work (to the best of our knowledge) As the differential conductance is very sensitive to local environment felt by the carriers, it is a powerful technique to characterize the uniformity of ZnO:Co Some preliminary results on the study of ZnO:Co-superconductor junctions were also presented The results of electrical transport properties will be presented first, followed by the discussion of origin of ferromagnetism in ZnO:Co, in combination with other results discussed in the previous chapters

6.2 Differential conductance studies

Inhomogeneity and disorder exist virtually in all types of materials, in particular in the material system under study, as seen from the structural properties presented and discussed in Chapter 3 These inhomogeneity and disorder would affect the electrical transport properties in a profound way In most of the work reported so

far for ZnO:Co, the electrical transport properties only include the resistivity, carrier concentration and Hall effect As average parameters, however, these quantities could not reveal directly how the structural properties were related to the transport properties

As inhomogeniety and disordering induce potential fluctuations, and in the extreme case form sub-regions of totally different characteristics, the effects should be reflected more apparently in the dynamic conductance as a function of the bias voltages For instance, in a system in which metallic particles were embedded in a semiconductor host matrix, one should expect the appearance of Schottky junction behaviour when the density of nanoparticles exceeds a certain threshold value or when the current path was confined in a narrow region Based on these considerations, the dI/dVxx versus

Vxx curves for all the samples were measured at different temperatures The results were shown in Figs 6-1 (a)-(h), respectively Here the discussion was focused on the following two aspects: (1) the shape of the conductance curve and (2) the size of the zero-bias anomaly (ZBA) Before proceeding to discuss the data, the effect of current induced heating will be commented on When the bias current increased, it was unavoidably that the sample would be heated up to a temperature which was higher than the cryostat temperature As all the samples under study exhibited a dirty metal behaviour, the local heating might lead to a higher conductance at a higher bias As all the samples under study had almost the same thickness and the Hall bars were of same dimension, the temperature rise should be inversely proportional to the resistivity of the sample when the bias voltage was at the same level Bearing this in mind, the results from the dI/dV curves will now be discussed

Shown in Fig 6-1(a) are the dI/dV curves for Al-doped ZnO without doping As can be seen from the figure, the differential conductance increased with the bias voltage at low temperature and becomes almost a constant above 30 K As the

Co-ZBA disappeared at the presence of a magnetic field, it should originate from localization in this sample As the sample was further doped with Co, both the shape and ZBA of the dI/dV curve changed remarkably As shown in Figs 6-1(b) - (d) (x

weak-=0.05, 0.14 and 0.2), the differential conductance curves for x < 0.2 had roughly a “V” shape With a further increase of Co composition, the low-bias curve evolved gradually into a “U” or parabolic shape (Fig.6-1 (e)-(h)) The V-shape could be understood as being caused by electrical-field assisted “de-trapping” of carriers localized in shallow potential wells On the other hand, the U-shape and parabolic curves could be attributed to the transport across grain boundaries which could be of either a tunnel junction or nanoscale heterojunctions / Schottky junctions,1 depending

on the electrical characteristics of the secondary phases Although current induced heating may also affect the shape of the dI/dV curve, it should not be the dominative mechanism because the resistivity of the lightly doped samples was almost the same as that of heavily doped samples (to be discussed shortly)

Now we turn to the temperature-dependence of ZBA for different samples, as shown in Fig 6-2 ((a)-(h)) In order to have a meaningful comparison, the ZBA was defined as the conductance ratio between V = 2 V and 0 V, i.e., dI/dVxx(2V)/ dI/dVxx(0) The ZBA, in principle, can appear in many different situations In a 4-point probe measurement configuration like the one used in this study, the influence of sample-electrode contact could be neglected and, apart from weak localization and heating effect, the ZBA was mainly caused by electrostatic potential disorder in the sample The difference between weak localization and electrostatic potential induced carrier localization could be readily differentiated from the dependence of ZBA on an applied magnetic field The ZBA for ZnO:Al disappeared completely at an applied field of 1T perpendicular to the sample surface, while those for samples with Co

doping were insensitive to the external field The size of ZBA served as an indicator of carrier localization strength and the shape of the differential conductance curve helped

to identify the carrier transport mechanism in individual samples Although currently there is no theoretical model available to explain quantitatively the shape of differential conductance curves observed here, clear changes in both the size of ZBA and shape of the conductance curve as the Co composition was varied had been observed

Focussing on the ZBA at 4.2 K, its value initially increased with the Co composition and reached a maximum at x = 0.25 beyond which it decreased again to unity at high Co composition As discussed in Chapter 4, the Co atoms were uniformly dispersed in the host matrix in samples with x < 0.2 When x increased to 0.25 at the onset of secondary phase formation, there was a significant increase in the degree of disorder inside the samples A further increase of Co composition would in turn lead to the enhancement of ordering due to formation of percolated secondary phases or Co clusters Therefore, the ZBA was expected to increase with the Co composition below the onset point of secondary phase formation and decreased again after it reached a maximum value It was worth noting that the shape of the dI/dV curves was also different at the two sides of the onset composition These results demonstrated clearly the usefulness of this technique to establish the relationship between the structural properties with the electrical transport properties As would be shown later, the ZBA was also correlated with the carrier localization in Co-doped samples

-2 -1 0 1 2 7.58

7.60 7.62 7.64

2.45 2.50 2.55 2.60 2.65

14K 14K

4.2K 4.2K

0 100 200 3000.97

0.99

1.01

0.991.001.01

Figure 6-2 Zero bias anomaly as a function of temperature for (a) Al-doped ZnO and various co-

doped samples, with x values indicated.

6.3 Temperature and Co composition-dependence of resistivity

From the dI/dV curves at different temperatures for samples with different Co compositions, one could readily obtain the dependence of resistivity on both temperature and the Co composition The dependence of resistivity on the Co composition is shown in Fig 6-3 at 4.2 K The resistivity increased slowly with the

increase of Co composition for x < 0.25 After reaching a maximum at x = 0.25, it decreased with a further increase of the Co composition The trend was almost the same as that of the dependence of ZBA on Co composition, as shown in the inset This again suggested that the degree of disorder was highest at x = 0.25, the onset composition of formation of secondary phases

Figure 6-3 Resistivity versus Co composition at 300 K; Inset shows zero bias anomaly dependence on

Co composition at 4.2 K

Fig 6-4 shows the zero-bias resistivity as a function of temperature for samples with different Co compositions As could be seen from the figure, all the samples at low and high Co compositions exhibited a typical “dirty metal” behaviour, while those in-between behaved like an insulator due to strong disorder caused by the onset of phase separation On the other hand, as shown in the inset, the Al-doped ZnO sample exhibited a weak semiconductor-like behaviour with the resistivity being almost independent of temperature Comparing Fig.6-4 with Fig.6-1, one immediately realized that the dI/dV curves provided more insights into the electrical conduction mechanism

of the samples

Co composition (%)

Figure 6-4 Resistivity versus temperature for various co-doped samples; Inset shows resistivity versus temperature for Al-doped ZnO

6.4 Hall effect

The presence of AHE is considered as one of the strong evidences for intrinsic ferromagnetism in DMSs.2,3 However, considering the fact that AHE had also been reported in ferromagnetic clusters,4 granular materials5-7 and inhomogeneous DMS in the hopping transport regime,8 the observation of AHE alone cannot support the claim that the DMS under study is a ferromagnet of intrinsic origin, unless it was correlated with ferromagnetism observed by other means Most importantly, secondary phases and precipitates must be shown to be absent in the sample

Figure 6- 5 Hall voltage as a function of applied magnetic field for (a) Al-doped ZnO and various doped samples, with x values indicated, at 4.2K and 300 K, current applied 0.1mA

co-Fig 6-5 shows the Hall voltage as a function of the applied field for different samples at different temperatures For non-magnetic samples, it was well known that the offset due to contact misalignment could be corrected by calculating the Hall voltage as [Vxy(B)-Vxy(-B)]/2 However, for magnetic samples, the offset was dependent on the magnetic field due to the strong MR effect Therefore, all the data shown in Fig 6-5 had been corrected for the MR effect However, the correction for the nonmagnetic field dependent offset by calculating the Hall voltage as [Vxy(B)-Vxy(-B)]/2 had not been carried out because the sample contains an antiferromagnetic phase; any operation involving [Vxy(B)-Vxy(-B)]/2 may lead to a wrong Vxy – B loop which would make it difficult to compare the AHE loop with the M-H loop measured by SQUID However, [Vxy(B)-Vxy(-B)]/2 at the maximum field had been used as the Hall voltage to calculate the carrier concentrations (to be presented shortly)

As shown in Figs 6-5 (a)-(c), the Zn1-xCoxO samples with x < 0.2 showed only OHE As the Co concentration increased, the AHE appears in samples with x > 0.25 The onset composition at which AHE started to appear also coincided with that of the appearance of secondary phases, i.e., at x = 0.25 in sample F All other samples (G-L) with x > 0.25 exhibited very clear AHE characteristics, as shown in Fig 6-5 (d)-(f) at 4.2K and Fig 6-5 (g)-(i) at 300K for samples F, G and J, respectively Therefore, now

it could be concluded that the presence of AHE in films with x > 0.25 was due to ferromagnetic secondary phases and not due to intrinsic ferromagnetism Although clear hysteresis loops had been observed for all the samples by SQUID, there was a fundamental difference between the samples with x < 0.25 and those with x 0.25 In the former, the magnetic properties came from magnetic regions which were electrically “isolated” from each other On the other hand, the magnetic regions in

samples with x 0.25 had reached the electrical percolation threshold, leading to the observation of AHE

By now ferromagnetism in samples with x 0.25 had been observed by SQUID, MCD and AHE Naturally it would be interesting to know if there is any correlation between the ferromagnetism observed by different techniques One of the possible ways to know this was to compare the hysteresis curves obtained by different techniques which were shown in Fig 6-6 for sample H For the case of using MCD, it had been shown the loops obtained by light with different photon energies As could be seen from the figure, the hysteresis curves obtained by SQUID, Hall and MCD measured at 2.92 eV (425 nm),9 were almost identical and matched well each other in shape However, the MCD curves measured at other photon energies were obviously different from those measured by SQUID and AHE As discussed in the MCD results

in previous chapter, these results suggested that the dominant phase in this sample indeed consists of Zn-incorporated CoO, which had formed an electrically percolated network The curves obtained from MCD at wavelengths near the d-d transition do not fit the SQUID or Hall results This might be understood from the fact that the d-d transitions originate from Co2+ ions of which not all of them contributed to ferromagnetism which would be picked up by SQUID and Hall effect For example, those inside the antiferromagentic clusters contributed to d-d transitions, but they did not contributed to SQUID and Hall signals

Figure 6-6 SQUID, AHE and MCD (at various energies) M-H curves for co-doped sample Co32W (Zn 0.71 Co 0.29 O) at 300 K

6.5 Temperature-dependence of carrier concentration

The temperature-dependence of carrier concentrations derived from Hall measurements were shown in Figs 6-7(a)-(h) The large fluctuation of carrier concentration for low-resistivity samples was caused by the small OHE signal However, it could be seen very clearly that carrier localization indeed occured at low temperature, < 50 K, in particular in samples with Co composition with x < 0.25 The high-temperature over low-temperature carrier density ratio agreed well with the same ratio of ZBA for different samples This suggested strongly that carriers were localized

in potential wells at low temperature and become de-trapped as temperature increased However, the carrier localization did not lead to ferromagnetic ordering in low Co composition samples

Figure 6-7 Carrier concentration of (a) Al-doped ZnO and various co-doped samples, with x values indicated, as a function of temperature

6.6 Magnetoresistance

As showed above, carrier localization indeed occured in ZnO:Co at low temperature; however, there was no intrinsic ferromagnetic ordering in lightly doped samples A question naturally arises here: is there any sp-d interaction in these samples?

The MR curves at different temperatures would be able to answer this question To this end, the MR had also been measured at various temperatures, from 1.5 to 300 K, in the field range of -6T to 6T, using the same Hall bars that had been used for differential conductance and Hall effect measurements MR is defined as the ratio between difference in resistance with and without applied magnetic field over resistance at the maximum applied field The MR was more sensitive to Co composition than AHE at low doping levels; thus it allowed us to study s,p-d interactions in samples even if AHE is absent For all the measurements, a bias current as small as possible (limited

by the signal-to-noise ratio) had been used, though the possible influence of heating from current flow, especially at low temperatures, could not be excluded

The MR curves (Figs 6-8(a)-(h)) for samples with x < 0.25 were very similar

to those reported in literature.10-14 In Fig 6-8 (a), the MR behaviour of Al-doped ZnO films without Co is shown A small negative MR was observed, decreasing with temperature, which was characteristic of weak localization.10,11,13 With doping of Co (x

≤ 0.2), a positive MR appeared at intermediate field values which was superimposed

with a negative MR at both low and high applied magnetic field At low temperature, the field at which the MR changed from positive to negative increased with Co composition but decreased with temperature Above 10 - 50 K (depending on Co composition), the MR became negative in the entire field range for samples A, B, C, D and F (Figs 6-8(b)-(e)) The negative MR near zero field exhibited a similar field dependence as that of ZnO:Al; therefore it could be understood as being originated from the destruction of quantum corrections due to weak-localization With the further increase of magnetic field, the spins of Co2+ ions would become increasingly aligned and large s-d interaction would lead to a splitting of the conduction band into spin-up and spin-down sub-bands The spin splitting of conduction band enhanced electron-

electron interactions in a disordered system which leads to a positive MR.11,13 Detailed simulation had been carried out to understand the MR behaviour of lightly doped ZnO:Co (sample 3W and 8W) based on the effect of the field-induced giant spin-splitting on disorder-modified electron-electron interactions.15 In addition to the field dependence, an anisotropy in the MR had been observed, which was due to the anisotropy of Co moment in the ZnO host matrix

When the field increased further, a negative MR appeared due to possibly the increasing alignment of electron spins with those of Co2+ ions11 or the formation of bound magnetic polarons.13 As the Co composition increased further, the positive MR became dominant in a much wider field range and, at x = 0.25, the MR continued to be positive even up to 6T As it was shown in the inset of Fig 6-8e, starting from x = 0.25, the negative MR peak at low field was no longer a single peak; instead it showed clear hysteresis The negative MR was relatively insensitive to temperature and became dominant over the positive MR above 70 K For samples with x > 0.25, the dominance

of negative MR with hysteresis became even more apparent (Figs 6-8(f)-(h)) and the positive MR was no longer observable in the entire field region below 6T In order to focus on the details of MR at low field, in Figs 6-8 (f) – (h), the MR in the range of -2T to 2T was presented The samples with x > 0.25 showed a typical MR curve for granular-like material at low field superimposed with a slowly changing negative background The onset Co composition of such MR behaviour again coincided with the Co composition at which the Co-rich phase became dominant and Co clusters started to appear

Figure 6-8 MR of (a) Al-doped ZnO and various co-doped samples, with x values indicated, from 4.2 –

50 K as a function of magnetic field, applied perpendicular to sample plane; Inset in (e) shows hysteresis in MR behaviour for sample Co25W (Zn 0.75 Co 0.25 O) for applied field of -1 to 1 T at 50 K

Figure 6-9 MR dependence on temperature of (a) Al-doped ZnO and various co-doped samples, with x values indicated, from 1.6 – 300 K

The temperature dependence of MR for various samples is shown in Fig 6-9

As could be seen from the figure, significant changes in MR occured when temperatures were lower than 50 K This corresponded well with the temperature-dependence of carrier concentration shown in Fig 6-7, indicating that the MR was partially caused by the redistribution of carriers at Zeeman splitting bands

6.7 ZnO:Co-Nb junctions

As discussed above, although all the samples exhibited “ferromagnetism” in SQUID measurement, the AHE effect was only observed in heavily doped samples which were attributed to an extrinsic origin Another possible way to study the magnetic properties of ZnO:Co was to form junctions with a superconductor The electrical conduction at the interface between a normal metal and a superconductor was carried out by a process called AR In this process, an electron-like quasiparticle from the normal metal, having energy of which was smaller than the cannot enter the superconductor and was retroflected as a quasihole At the mean time, a Cooper pair was transferred into the superconductor electrode In general, the electron transport in an N/SC junction could be described by the BTK model, which calculated the current-voltage relation of a point contact between a normal metal and a superconductor by assuming that there existed a δ-shape energy barrier Z at the

interface

In general, the AR was greatly suppressed at the interface between a ferromagnet and a superconductor due to the large exchange energy in FM It was generally believed that Cooper pairs can only traverse through a ferromagnet with a distance on the order of l = D E/ ex , where D if the diffusion constant of electrons in the ferromagnet, the Plank constant and Eex the exchange energy As Eex was large, l was usually less than 1 nm Strijkers et al have developed a modified BTK model to

describe the electrical transport properties of the FM/SC interface by taking into account both the barrier at the interface and polarization of electrons in the ferromagnetic layer.16 Shown in Fig 6-10 were the normalized conductance versus voltage curves for a point-contact FM/SC junction with different barrier height Z and polarization of electrons P in the FM layer As seen from the figure, for a constant Z,

the conductance below the superconducting gap decreased with increasing P When P

= 1, the electrical conduction was completely suppressed in the gap region

The suppression of electrical conduction across the FM/SC interfaces was mainly caused by the fact that there was a lacking of quasiparticles with opposite spin directions to form Cooper pairs Therefore, it had been proposed and verified that the

AR at the FM/SC interfaces would be enhanced if the electron spins in the FM region were not well-aligned such as those in domain walls17,18 or granular materials.19 In these cases, the electrons in one region would be able to pair up with retro-reflected holes neighboring regions with a distance which was shorter than the Cooper pair coherence length Therefore, this type AR was often termed CAR or non-local Andreev reflection.20 Bearing these basic facts in mind, we will discuss below the experiments and results of ZnCoO/Nb junctions

-6 -4 -2 0 2 4 6 0.0

0.5 1.0 1.5 0.0 0.5 1.0 1.5 0.0

0.5

1.0

1.5

0.0 0.5 1.0 1.5 0.0

0.5

1.0

0.5 1.0 1.5 2.0

Although both the original and modified BTK models were only valid for a point contact, we had adopted a large-size contact because our samples were inhomogeneous in nature and it was difficult to deduce the polarization ratio from these measurements Instead, we were only interested in how junctions with different

Co compositions were different from one another To study the magnetic properties of the samples, 2 superconducting Nb pads, with a size of 1.5 mm × 1.5 mm and a spacing of 0.8 mm, as shown in Fig 6-11, were deposited on the ZnO:Co thin film samples via sputtering Measurements were then carried using a two-probe configuration on four samples: D (x = 0.20), G (x = 0.27), J (x = 0.30) and L (x = 0.33),

in the temperature range of 1.4 K- 10 K, i.e., below the critical temperature of Nb The resistance between the two electrodes consisted of the contributions from two ZnCoO/Nb contacts and the centre portion of ZnCoO between the two electrodes As the ZBA for the four samples was very small (see Fig 6.2), the dependence of dynamic conductance, if any, should mainly originate from the contacts

Al 2 O 3 substrate ZnO:Co

1.5mm 0.8mm 1.5mm

Al 2 O 3 substrate ZnO:Co