Oxygen plasma treatment of hematite

To study the effects of highly oxidizing treatments an oxygen plasma was used. The treat-ment was performed at room temperature for 15 min with a current of 40 mA in the atom mode of the plasma source. In this mode sputter effects due to high energy particle from the plasma are minimized.

The results from two samples shall be discussed here. First, a presumably amorphous sample that has been deposited at room temperature and second, a presumably crystalline sample that was deposited at 400^◦C. The crystallinity of the two samples can only be as-sumed from the XP spectra as no dedicated structural analysis have been performed. The spectra, however, resemble perfectly the spectra from amorphous and crystalline samples that have been discussed in part 4 of this work.

Please note, that prior to the oxygen 800

700 600 500 400

288 286 284

ng ener

Figure 7.1:C1s-spectra of adventitious car-bon before and after the plasma treat-ment.

plasma treatment the crystalline sample had already undergone other treatments like water exposure and heating. In immedi-ate advance to the measurement presented here, the sample had been stored under UHV-conditions for five days after it was heated at 400^◦C in vacuum before. From this storage some adventitious carbon could be found in the survey spectrum as shown in Figure 7.1. After the plasma treatment no carbon could be found on the surface anymore. This carbon, however, should not affect the effect of the oxygen plasma treatment tremendously and is, therefore, in the discussion of the treatment neglected in the following.

150 7 Fermi level manipulation of the surface

Binding energy / eV

Intensity / arb. units

a) Fe2p b) O1s c) VB XPS

720 716 712 708 531 527 8 4 0

A

A

Figure 7.2:XP spectra for an (presumably) amorphous and a (presumably) crystalline sam-ple before and after a treatment in an oxygen plasma.

The XP spectra from the two samples before and after the oxygen plasma treatment are being presented in Figure 7.2. Besides the crystallinity and the resulting spectral changes the sample differ in their Fermi level position before the treatment. While the amorphous sample shows a low Fermi level position of 0.87 eV, the crystalline sample exhibits a Fermi level position of 1.6 eV which is rather high. It could actually result from the prior heating in vacuum which shifts the Fermi level position to higher values.

One minor difference is a small shoulder towards higher binding energies in the O1s-spectrum of the crystalline sample. It is assigned to originate from carbon hydrates or OH^–-adsorbates due to the storage in the system.

After the treatment a shift towards lower binding energies is observed in the core-levels of both samples as well as in the valence band. The shift is larger for the crystalline sam-ple which can be explained by the initial higher Fermi level. After the oxygen plasma treatment the Fermi level in the two samples are rather similar. Positions of 0.66 eV and 7.1 Surface treatment by oxygen plasma and exposure to water 151

0.6 eV are deduced for the amorphous and crystalline sample, respectively. Within the precision of XPS these values are the same.

In order to interpret this observation three important aspects need to be considered.

First, the treatment is the same for both samples. Second, the Fermi level positions before the treatment are very different. This results in a different amount of charges that have to be transferred in order to shift the Fermi level in the two samples to the similar position by the oxygen plasma treatment. Third, magnesium doped films showed a lower Fermi level position than the one that was achieved here.

From the first two statements it can be deduced that the Fermi level is pinned at around 0.65 eV. Otherwise the different Fermi level positions before the treatment should result in different Fermi level positions after the treatment (assuming similarity of the two treat-ments). Including the third aspect clearly shows that the pinning cannot originate from within the hematite thin films but that it has to result from the charge neutrality level associated with the surface conditions induced by an adsorbed species.

The presence of this adsorbed species can be proven from the O1s-spectra in Fig-ure 7.2b). Here, an additional signal emerges at higher binding energies for both sample after the treatment which was labeled with A. This signal is positioned about 2.75 eV above the main line in both samples indicating that it is the same species. In literature the presence of a similar signal after a treatment in air or oxygen plasma with slightly smaller difference to the main line has been observed before. Zhu et al. explained the signal by the presence of oxygen vacancies that were created during their air plasma treatment.[282] This assignment is rather unreasonable as oxygen vacancies should first of all not emit any photoelectrons and secondly it is hard to believe that oxygen vacancies are formed under very oxidizing conditions. This possibility is, therefore, omitted.

Another assignment can be found from Hu et al. who explained the signal to originate from OH^– group attached to the surface after an oxygen plasma treatment. They also include the possibility that this signal arises from nonstoichiometric oxygen like O₂^{2 –} or O^– but only omit the latter in their discussion¹. The presence of peroxides, however, is not further discussed.[57]

In the present case the signal can be reasonably explained by adsorbed peroxide (O₂^{2 –}) species as for this species a difference to the main line of O1s of about 2-3 eV was shown.[241] In a further step on the crystalline sample it was shown that this species can be removed by a mild heating at 200^◦C in vacuum. Afterwards, the Fermi level is also higher again.

1 O₂^– was not included in their discussion

152 7 Fermi level manipulation of the surface

An additional indication for an adsorption on the surface is the observed shift that had to be used in the difference spectra in Figure A12. For the Fe2p-, O1s and valence band slightly different shifts of −0.1 eV, −0.18 eV, and −0.21 eV had to be applied. Even though, the difference is marginal it might indicate that there is band bending present on the surface after the plasma treatment. This conclusion is based on the different in-formation depth of the three region with Fe2p being the most surface sensitive and the valence band having the highest information depth. It should be noted, however, that the expected difference are rather small.[116, 137]

Despite the obvious changes in the O1s-spectra there are no further changes that are as obvious. Difference spectra that are shown in the appendix in Figure A12 revealed that the valence bands do not change at all due to the plasma treatment. Small changes were observed in the difference spectra of the Fe2p-spectra. Here, an adjustment of the inten-sity on the Fe³⁺-satellite resulted in a slightly larger intensity of the main line after the plasma treatment. This was especially the case for the amorphous sample. Assuming that an adjustment of satellite intensity can be interpreted as a similar amount of signal from Fe³⁺in the difference spectra, the additional intensity might be explained by the presence of another oxidation state of iron. Namely, Fe⁴⁺could be considered.Braun et al. found evidence for the formation of Fe⁴⁺for low Fermi level positions in hematite.[240]

The lower Fermi levels of Mg-doped samples, however, seem to be reason enough to disregard this possibility. It could, however, be shown in Figure 6.3 that with higher Mg-contents (associated with lower Fermi level) almost similar changes in the Fe2p-spectra were observed. A tentative assignment of the difference to the formation of Fe⁴⁺ can, therefore, not completely be omitted. However, further investigation on this topic by more suited techniques, e.g. surface sensitive Mößbauer spectroscopy (achieved in a grazing incidence geometry), are necessary.