2 – 5. INFORMACIÓN COMPLEMENTARIA

stoichiometry of the crystalline phase. This is indicated by SEM EDX measurements (Fig. 5.71) which show that the ITO matrix actually incorporates oxygen from the atmosphere, since the In/Sn atomic ratio remains stable while (In+Sn)/O decreases. However, the decline is not particularly significant if the error bars are taken into account showing that the actual Oxygen intake by the bulk structure is rather small. Even at low P(O2),

the inner part of ITO remained oxidised since the EDX SEM atomic (In+Sn)/O ratios obtained were close to the nominal value. This in agreement with what was observed in section 5.2 where even at Ts values as high as 400◦C (which favours oxygen bulk

diffusion), the interaction of the oxygen atmosphere with the bulk material was relatively moderate (Fig. 5.35).

XPS gave further insight into the intake of oxygen and related phenomena at the film/air interface. The surface was slightly oxygen deficient at higher (>5 mT) than at lower

5.3. Influence of Oxygen pressure 153

P(O2) (∼0.2 mT). This might indicate some degree of oxygen adsorption on the top

surface at very low P(O2). This process may be favoured via the larger amount of oxygen

vacancies and holes populating the top surface at such pressure. This oxygen deficient environment was responsible for Sn 3d and O 1s peak asymmetries (Fig. 5.73). The formation at the surface of a new phase incorporating Sn and O atoms is not ruled out either. This possible phase would disappear gradually as P(O2) increases, this being

attributed to the restored stoichiometry (as regards to the bulk) of the structure.

Based solely on the XPS results presented in this study, these possible other phase(s) could not be identified with any certainty. The involvement of Sn and O atoms seems very likely as their respective peak profiles were modified at low P(O2). Nonetheless, the

presence of a suboxide like In2O (a non-transparent and highly resistive material [207])

cannot be totally ruled out since the samples obtained at low P(O2) were darkish. SEM

imaging (Fig. 5.60) did show particles sitting on the top surface. Their true nature could not be quantified owing to SEM EDX spatial resolution constraints, but they are expected to be of metallic and/or suboxidic in nature. Therefore, changes to the shape of the core-levels with oxygen content in the chamber were attributed to the formation of intermediate compound(s), combined with possible surface suboxide/metallic particle formation favoured by a very poor oxygen environment at P(O2)=0.2 mT.

In addition to the O 1s asymmetry, systematic changes with P(O2) of binding energies

towards lower energies were observed for the In 3d, Sn 3d and O 1s spectra. Such a shift in the whole spectrum was interpreted as a change in the surface Fermi level EF. A shift

towards lower binding energy as P(O2) increases is compatible with an increase in the

surface work function, which corresponds to the energy needed to move an electron from the surface Fermi level into vacuum. This observation is in agreement with the increased surface work function observed after oxidative surface treatments [208]. Interest in the ITO surface potentials was generated by the observation that an increase of the ITO work function leads to a desired lower injection barrier for holes at the interface with organic conductors. This increase in work function is typically achieved by oxidative treatments such as oxygen plasma, UV ozone, or chemical treatments. However, in solar cells, the ITO electrodes are used as electron collectors which requires a low work function. The change in EF was attributed to changes in surface coverage by O2.

The film textures were quantitatively evaluated via the texture coefficients TC(400) or

5.3. Influence of Oxygen pressure 154

Hall-effect measurements. Excellent linear agreement (R=0.98) in chamber 1 is found between TC(222)and µ (Fig. 5.55), although the experimental point at 50 mT was not

taken into account as it showed very poor agreement with the rest of the data. This is attributed to oxygen accumulating in the structure and forming complexes/precipitates. In addition, this linear relation means that the maximum in µ observed previously (Fig. 5.66) at 10/20 mT in chamber 1 also corresponds to the film with the strongest (222) preferred orientation.

No particular strong linear regression (R=0.64) was found for chamber 2, especially in the low pressure range (0.2 - 5 mT), where the crystal damage dominates the mobility behaviour. Therefore, µ appears to be more independent of the texture than in chamber 1. No other explanations could be found for the discrepencies between chambers 1 and 2 since films in both chambers showed about the same carrier density, so, any argument based on defect screening by a high carrier density does not hold. The mixed findings in this study reflect the situation in the literature, where orientation is sometimes believed to affect the mobility of free carriers [209] [210] [211], while other authors suggest no influence whatsoever [29][212].

The mobility µ was found to be linearly related to the FWHMs of preferred oriention peaks in both chambers (Fig. 5.54), which gives an indication of the degree of crystallinity of a certain film (of no or negligible crystallite size and strain broadening). Linear relations were identified for both chambers. One is between P(O2)=1 and 20 mT and P(O2)=20

and 50 mT in chamber 1. At P(O2)=20 mT, the proportionality changes corresponding

to the maximum observed mobility (Fig. 5.66). After P(O2)=20 mT, µ degrades much

faster than in the 1 - 20 mT range as witnessed by the steeper slope of µ versus the FWHM of the (222) peaks. As for chamber 2, linear relations between µ and the FWHMs were found for all the pressure ranges investigated (0.2 - 70 mT) except at P(O2)=1, 30

and 40 mT. It is concluded that the higher the FWHM, the lower the mobility.

The particularly marked discrepancy in chamber 2 near P(O2)=30 mT in terms of FWHMs

and µ is explained on the basis of a structure argument. The (222) preferred orientation peak showed a systematic and reproducible asymmetry at larger 2θ (Fig. 5.56): this broad peak (shoulder) was attributed to a combination of SnO (JCPDS 01-072-2324), SnO2 (JCPDS 01-071-5329) and possibly Sn3O4 (JCPDS 00-020-1293) phases. These

were not stable at P (O2) >30 mT. The sharper and more intense peak at 2θ=29.49◦

was still attributed to the cubic In2O3 phase despite a significant 1◦ shift of the main

5.3. Influence of Oxygen pressure 155

ever larger FWHM(222) between P(O2)=5 and 20 mT is suggested to arise from this

particular phase development, which also has a large (positive) impact on µ (Fig. 5.66), while no effect whatsoever was noticed on the carrier concentration N (Fig. 5.69). Hence, these phases appear to be more conductive than In2O3.

GAXRD confirmed that these phases are present across the whole layer since the peak shape is conserved as the incident angle is varied (i.e. as different thicknesses were probed) as observed in Fig. 5.58. Such a behaviour was not observed in chamber 1, at least between P(O2)=1 and 50 mT. At P (O2) >30 mT, the asymmetry disappears. In

addition, when working under a static atmosphere (and also a lower energy density), it also vanishes. This points to the fact that gas dynamics might play a central role in the formation of secondary phases during PLD growth. It is worth mentioning that very few reports of asymmetry in the (222) peak have been published to date [30]. The non-thermal nature of PLD favours the formation of such metastable phases which could explain why such effects are not systematically reported. It is suggested that the high energy density used in chamber 2 might favour such formation since it results in a plume containing a larger amount of reactive species facilitating metastable phase formation.

GAXRD also showed a slight change in film density as P(O2) was increased. A shift

towards higher 2θ was observed as P(O2) was changed from 0.2 to 7.5 mT (Fig. 5.74).

This shift is explained in terms of oxygen vacancy filling and bombardment by energetic particles during growth. As P(O2) increased, more oxygen vacancies are filled and the

molar volume of the film increases. This moderate intake was evidenced by EDX SEM (Fig. 5.71) and probably explains the observed moderate shift of the (222) peaks. Since the film is attached to a rigid substrate (glass), this shift is expected to result in some compressive stress. In addition, bombardment by energetic particles of the film surface could also play a role in this peak position shift as it is also known [5] to additional compressive stress via compaction and distortion of the surface.