P type transparent conducting CU AL o thin films prepared by PE MOCVD 2

With the increase of growth temperature, more Cu2+ could be regarded as more n-type doping that helped to form energetically favorable acceptor-donor-acceptor complexes, and resulted in

cannot provide accurate information as the core level peak It is still clear that in the CuLMM peak (calibrated), the main peak was at a binding energy around 337.5eV, which was from Cu+, and a small peak was at a lower binding energy around 335.5eV, which was from Cu2+

To decide the percentage of Cu+, the spectra of Cu2p3/2 were fitted by the software

provided with the XPS instrument In the fitting process, the difference in Cu2p3/2

binding energies of Cu+ and Cu2+, the GL (Gaussian-Lorentzian) ratio and the FWHM (Full Width at Half Maximum) were fixed

The peak fitting results are listed in Table 4-6 The Cu+ content was about 80% and the content of Cu2+ increased with the increase of growth temperature This provides further proof that the co-doping theory17 may work in the present films

Table 4-6 The content of Cu + and Cu 2+ calculated from peak fitting results

As discussed in the previous section, Cu2+ can act as an n-type dopant when it replaces

Cu+ With the increase of growth temperature, more Cu2+ could be regarded as more

n-type doping that helped to form energetically favorable acceptor-donor-acceptor

complexes, and resulted in a shallower acceptor level and consequential higher conductivity.17

4.3.3 The effect of oxygen flow rate on the properties of Cu-Al-O films

In this section the change of properties with the oxygen flow rate will be discussed The growth conditions were: O2 flow rate at 4 to 20sccm, Ar flow rate at 30sccm, working pressure at 150mTorr, plasma power at 50W and the substrate temperature at 700°C The film thickness was measured to be from 120 to 180nm, equivalent to a growth rate of 2-2.5nm/min

XRD was employed with a fixed incidence angle of 1 degree The XRD results are shown in Figure 4-16 As indicated in the previous section, the hump at 21.20° is from the quartz substrate There are mainly two peaks around 43.46° (2.083Å) and 50.57°

(1.806Å) in all the films, except the film grown at the lowest oxygen flow rate that showed one more small peak at 53.60° (1.710Å) Here the XRD spectra were similar

to that being analyzed before, thus the XRD peaks might be from β-CuAlO2

Scattering Angle 2θ (deg.)

Figure 4-16 XRD spectra of as-deposited films grown at different oxygen flow rates

Because of charging, the morphologies from SEM of films grown at 12sccm and 20sccm were not shown Figure 4-17 shows the SEM pictures of the annealed films grown at oxygen flows of 4−8sccm No big difference in the morphology was observed in the films grown at different oxygen flow rates except for the particle size The particle size decreased from about 45 to 15nm when the oxygen flow rate increased

Figure 4-18 shows the transmittance of the as-deposited and annealed films in the range of 300-1100nm The inset shows the transmittance at 650nm The films were annealed in the RHF1400 Carbolite furnace at 350°C for 5 minutes in air

300 500 700 900 1100 0

10 20 30 40 50 60 70 80

4 8 12 16 20 0

20 40 60 80 100

E

(a)

0 15 30 45 60 75 90

0 20 40 60 80 100

CEA

(b)

Figure 4-18 Transmittances of (a) as-deposited and (b) 350°C annealed Cu-Al-O films grown

at different oxygen flow rates, A: 4sccm, B: 6sccm, C: 8sccm, D: 12sccm and E: 20sccm

Before annealing, with the increase of oxygen flow rate, there was no much difference

in the transmittance for the films grown at different oxygen flow rates except the one

at 4sccm, which had lowest transmittance in the visible range After annealing, the transmittance of every film was greatly improved to be above 70% in the near-infrared

range Above about 500nm, the film at 6sccm showed the lowest transmittance, while other films showed similar transmittance One obvious change due to annealing was observed that one absorption peak near 600nm disappeared after annealing To explain this phenomenon, the absorbance of as-deposited films as a function of photon energy

is shown in Figure 4-19

0.1 0.2 0.3 0.4 0.5

Figure 4-19 Absorbances (plot against photon energy) of as-deposited films grown from acac precursors at different oxygen flow rates, A: 4sccm, B: 6sccm, C: 8sccm, D: 12sccm and E: 20sccm Eg is the absorption edge

All spectra showed an absorption edge Eg around 2.2eV It cannot be determined if this is direct bandgap or indirect bandgap from this figure.26 From the plot (αhν)1/2

against hν, an indirect bandgap of the film grown at 4sccm was estimated to be 2.25eV, which was very close to this value The absorption peak below Eg can be explained by an excitonic effect.26 An exciton is a bound electron-hole pair, usually free to move together through the crystal It is formed because of strong Coulomb attraction between the electron and the hole All excitons are unstable with respect to

the ultimate recombination process in which the electron drops into the hole.27 This is why the absorption peak disappeared after annealing Theoretically, the optical absorption edge is perfectly abrupt; however, there are several effects that make the model change The excitonic effect mentioned above is one reason The existence of

an impurity bandgap and the emission of a phonon with absorption of a photon are also possible reasons of this band tail absorption.26

3.5 3.7 3.9 4.1 4.3

after annealing

before annealing

Oxygen Flow Rate (sccm)

Figure 4-20 Optical bandgap versus oxygen flow rate for as-deposited () and 350°C annealed ( ) films grown from acac precursors

The direct optical bandgaps of the films deduced from absorption spectra versus oxygen flow rate are plotted in Figure 4-20 For as-deposited films, the bandgap decreased first, and then increased A minimum point at 8sccm was observed After annealing, the bandgap still had similar trend with the lowest point at 8sccm It can be concluded that the relationship between the bandgap and the oxygen flow rate is not linear and the film grown at 8sccm has the smallest bandgap The change of the

optical bandgap cannot be explained by the change of carrier concentration due to the immeasurability of the carrier concentration

The conductivity of all the films was not very good and the Hall effect measurement was not applicable for these films Therefore, the two-probe method was employed to measure the resistance The films grown at oxygen flow rates of 4sccm and 20sccm showed very poor conductivity while their resistances were beyond the range of measurement The resistances of other films were in the range of 5 to 25MΩ Roughly, the films with the lower resistivity had smaller bandgap After annealing, the resistance was increased by about 50% As discussed in last section, annealing resulted in a decrease of defects concentration (mainly interstitial oxygen) and a consequential increase of resistivity

XPS spectra Cu2p of the annealed films (Figure 4-21) were investigated after sputtering-clean and all spectra were plotted after the calibration by using the C1s peak as 284.8eV The peak at a high binding energy (952.5eV) is Cu2p1/2 whose appearance is due to the spin doublet separation resulted from ionization There is a

satellite peak between peak 2p3/2 and peak 2p1/2, which is called a shake-up line This shake-up line came out because some ions were left in an excited state a few electron volts above the ground state leading to an increase in the binding energy of the emitted photoelectron.25 The existence of this peak is an evidence of the existence of

Cu2+

The Cu2p3/2 peaks of all films appeared around 932.6eV, showing the dominance of

Cu+ in the films However, there is some difference in the shake-up peak at 943.8eV For sample A (oxygen flow rate was 4sccm), there was no such satellite peak at all

This satellite peak was seen clearly for samples C (8sccm) and E (20sccm), meaning that more Cu2+ was formed at higher oxygen flow rates

ECBA

Figure 4-21 XPS Cu2p spectra of 350°C annealed films grown at different oxygen flow rates,

A: 4sccm, B: 6sccm, C: 8sccm and E: 20sccm, D: 12sccm is not included because of too low counts

The binding energy of Cu2p3/2 for every film was not exactly the same after calibration This was attributed to a chemical shift, which was related to the chemical environment of the element A higher binding energy means a stronger bond to the nucleus or looser bonds with adjacent atoms.25 Therefore, it is known from the figure that when oxygen flow rate was at an intermediate value, the bonds of copper atom with the adjacent oxygen atoms were less strong It is noticed that the samples with less strong Cu-O bond had better conductivity This is possibly because the positive holes move mainly in the CuO2 layer whose structure is much more open than AlO6

layer Less strong Cu-O bond means longer bond length, which gives holes more space to move

4.4 Further Discussion on Film Properties

The rhombohedral structure was normally found in delafossite compounds To lay stress on the layer structure, the hexagonal description was often used, which means that the structure diagram was drawn referring to the hexagonal axis28 (Figure 4-22)

c axis

Figure 4-22 Rhombohedral ABO2 in hexagonal description, the vertical direction is c axis (adapted from R N Attili, M Uhrmacher, K P Lieb, and L Ziegeler, Phys Rev B53, 600 (1996))

Figure 4-23 shows a rhombohedral lattice in which the primitive cell is defined by the

rhombohedral axes a 1 , a 2 , a 3; but a non-primitive hexagonal unit cell can be chosen by

adopting the axes A 1 , A 2 and C.29 The latter has lattice points at 000,

3

23

13

1 and

Figure 4-23 A rhombohedral lattice (a1, a2, a3) referring to hexagonal axes (A1, A2, C) (Adapted from R W James, X-ray crystallography, Wiley, New York (1953))

The structure of laser-ablation prepared CuAlO2 film was confirmed as R3 m

(rhombohedral) detected by XRD.8 In the XRD spectra observed in the present work, only a few weak peaks were obtained, which were very difficult to identify XRD data only implied β-CuAlO2 and Cu phases Careful analysis of TEM diffraction rings and high-resolution images excluded the possibility of the existence of metal copper and CuAl2O4, suggesting that the major phase was CuAlO2 (β-CuAlO2 and rhombohedral CuAlO2) and minor phases were Cu2O and Al2O3

However, β-CuAlO2 is a structure with a large volume that was unstable, which was stated by Gessner7 Its instability could be proved by its disappearance under the bombardment of high-voltage electrons in the TEM chamber or upon heating This meta-stable phase only appeared in as-deposited films and decomposed to Cu2O and

Al2O3 upon heating The phase of rhombohedral CuAlO2, which was not observed by XRD but was clearly shown by TEM, emerged in the as-deposited films probably as small crystallites with preferred orientation

Polycrystalline even amorphous growth is a characteristic of the CVD process, and identifying the structure is difficult by using traditional tools such as XRD It is the same characteristic that the films benefit from because the mixture of polycrystalline and amorphous states allow the existence of a large amount of defects, which may be

the origins of p-type conduction and therefore, the reason of the good conductivity

CuAlO2 is regarded as a p-type semiconductor similar to Cu2O Hall effect is the most common technique to determine the type of conductivity However, in the present work, Hall effect could not be employed sometimes It is believed that the Seebeck effect is more reliable to determine the conductivity type of wide bandgap semiconductors.19 The difference between the Seebeck effect and the Hall effect is that the Seebeck effect uses a thermo effect (the thermovoltage of the films) and the Hall effect uses a magnetic effect From the Seebeck effect measurements in the

present work, all conductive films showed stable p-type conductivity Regrettably, the

Seebeck measurement does not give the value of carrier concentration and mobility

Yoshida et al proposed a co-doping theory that suggested one way to improve the type conduction by doping n-type co-dopants This theory was based on theoretical

p-calculation In CuAlO2, the origins of holes could be oxygen interstitials or/and copper vacancies so Cu2+ ions could act as n-type co-dopants, whose existence was confirmed

by XPS Thus the present work may provide an experimental support to the co-doping

theory The fact that the co-existence of p-type and n-type carriers can explain the

immeasurability of Hall effect for some samples, which will be discussed in the next chapter

The conductivity was found to increase with the increase of the growth temperature However, in section 4.3.2, the best conductivity noted was 0.27S·cm-1, which was worse than the film presented in section 4.3.1 (2.0S·cm-1) even though it was prepared

at higher growth temperature This might be due to the different plasma powers In section 4.3.1, the plasma power was 200W while in section 4.3.2, the power was only 100W In section 4.3.3, the conductivity increased first and then decreased while the oxygen flow rate was increased Other researchers also observed the non-linearity of conductivity versus oxygen partial pressure.30, 31 This was associated with the non-stoichiometry of the films.22

4.4.3 Optical properties

Some information of bandgap can be obtained from the optical absorption Figure 4-19 shows an obvious absorption peak below the absorption edge (2.2eV) This peak was attributed to an excitonic effect In wide-band semiconductors, it should be a Wannier exciton, which has weak attraction between the electron and the hole.27 The classic example of this is Cu2O, where the absorption spectrum measured at low temperatures showed a progression of exciton levels just below the bandgap at 2.1eV.32 The Wannier excitons are strongly perturbed by lattice vibrations and are not

so stable Hence, in the annealed films, exciton levels did not appear

In section 4.3.2, the direct optical bandgap of the as-deposited films was from 3.93 to 3.77eV when they were deposited at 630-800°C The bandgap decreased with the growth temperature This result is completely different from the data of De and Ray33who studied the effect of substrate temperature on the bandgap of magnetron sputtered tin oxide films At a substrate temperature of 150°C, the tin oxide films had a bandgap

of 3.83eV, which increased to 4.13eV when the substrate temperature was increased

to 450°C These authors suggested that a higher growth temperature enhanced the growth of the SnO2 phase in the films and improved the crystallinity significantly, thereby increasing the bandgap The increase of bandgap with growth temperature could also be explained by the Burstein-Moss shift34, 35 (refer to section 5.4.2), which indicates that the increase in the number of carriers shifts the band edge towards shorter wavelengths

The phenomenon that the magnitude of the bandgap shrinkage increased with doping concentration was also observed by other researchers.36 , 37 The shrinkage was explained by the change in the nature and strength of the interaction potentials between donors (acceptors) and the host materials, leading to the increase of the tailing of the absorption edge and hence the reduced bandgap (refer to section 5.4.2) The tailing of absorption edge was apparently observed in the current work (Figure 4-19) thus the decrease of bandgap with the growth temperature might be due to the bandgap shrinkage effect

In section 4.3.3, the bandgaps obtained were from 4.14 to 3.77eV for as-deposited films and 4.33 to 3.58eV for annealed films, which decreased first then increased with the increase of oxygen flow rate Both Burstein-Moss and bandgap shrinkage theories are related to the carrier concentrations However, the data for carrier concentrations were not available for these films It is noticed that the film with lower bandgap had higher conductivity and the increased tailing of the absorption edge was observed, similar to the results in section 4.3.2, the change of bandgap might also be due to the bandgap shrinkage effect

More discussions for the bandgap are in Chapter 5

A high conductivity of 2.0S·cm-1 was achieved, with the activation energy of 0.12eV and the carrier concentration of 2.6×1019cm-3 The high conductivity and high carrier concentration were explained by the small activation energy and co-doping theory The conductivity increased with the growth temperature, which could also be illustrated by the co-doping theory

The transmittance in the visible range was up to 65% for the as-deposited films and up

to 70% for the 350°C annealed films The bandgaps were estimated from the measurements of optical absorption and all films had a wide bandgap ranging from 3.5

to 4.14eV The change of bandgap with growth temperature and oxygen flow rate was explained by the bandgap shrinkage effect

The valence of copper in Cu-Al-O compound was determined by XPS and a quantitative analysis showed that the percentage of Cu2+ in the film was much smaller than Cu+ The existence of Cu2+ provided an experimental proof for co-doping theory

References:

1 J C Kenvin, M G White and M B Mitchel, Langmuir 7, 1198 (1991)

2 A E Turgambaeva, A F Bykov and J K Igumenov, J De Physique IV 5, C5-221 (1995)

3 O B A Ajayi, M S Akanni, J N Lambi, C Jeynes and J F Watts, Thin Solid Films 185, 123 (1990)

4 T Maruyama and S Arai, Appl Phys Lett 60, 322 (1992)

5 P Hirsch, A Howie, R B Nicholson, D W Pashley and M J Whelan, Electron microscopy of thin crystals, Butterworths, London (1977)

6 JCPDS, International Centre for Diffraction Data (1998)

7 W Gessner, Z Anorg Allg Chem 352, 145 (1967)

8 H Kawazoe, M Yasukawa, H Hyodo, M Kurita, H Yanagi and H Hosono, Nature

13 H.Yanagi, S I Inoue, H Ueda and H Kawazoe, J Appl Phys 88, 4159 (2000).

14 A Kudo, H Yanagi, H Hosono and H Kawazoe, Appl Phys Lett 73, 220 (1998)

15 O Porat and I Riess, Solid State Ionics 81, 29 (1995)

16 H K Yoshida, T Nishimatsu, T Yamamoto and N Orita, Phys Stat Sol (b) 210,

429 (1998)

17 T Yamamoto and H K Yoshida, Japan J Appl Phys 38, 166 (1999)

18 M L Hitchman & K F Jensen, Chemical vapor deposition: principles and

Applications, Academic Press, London and San Diego (1993)

19 P A Cox, Transition metal oxides:an introduction to their electronic structure and properties, Clarendon Press, Oxford (1992)

20 T Muranoi and M Furukoshi, Thin Solid Films 48, 309 (1978)

21 N Croitoru and E Bannett, Thin Solid Films 82, 235 (1981)

22 H L Hartnagel, A L Dawar, A K Jain and C Jagadish, Semiconducting

transparent thin films, Institute of Physics, Bristol and Philadelphia (1995)

23 S M Jarzebski and J P Marton, J Electrochem Soc 123, 299C (1976)

24 A G Sabnis, J Vac Sci Technol 15, 1565 (1978)

25 J F Moulder, W F Stickle, P E Sobol and K D Bomben, Handbook of X-ray photoelectron spectroscopy, Perkin-Elmer Corporation, Eden Prairie (1995)

26 P Bhattacharya, Semiconductor optoelectronic devices, Prentice Hall,Englewood Cliffs (1993)

27 C Kittel, Introduction to solid state physics, John Wiley & Sons, New York (1996)

28 R N Attili, M Uhrmacher, K P Lieb and L Ziegeler, Phys Rev B53, 600 (1996)

29 R W James, X-ray crystallography, Wiley, New York (1953)

30 J R Bellingham, A P Mackenzie and W.A Phillips, Appl Phys Lett 58, 2506 (1991)

31 S Schiller, G Beister, E Buedke, H J Becker and H Schicht, Thin Solid Films 96,

113 (1982)

32 P W Baumeister, Phys Rev 121, 359 (1961)