C h apter 7 : ZnO su rface studies
7.1 Introduction
The w urtzite structure of zinc oxide gives rise to several com plicating features on each o f the dom inant crystal faces which are exhibited in the usual m orphology. The tw o m ajor non-polar faces, (10Î0 ) and (1120), o f the crystal are subject to surface relaxation or rum pling, the extent o f w hich is currently under dispute as different experim ental techniques appear to contradict each other and theoretical calculations furtherm ore appear to predict significantly different surface relaxations com pared to each of the experim ental techniques.
For the (lOTO) surface in particular, there appears to be significant disagreem ent between experim ental and theoretical conclusions. The grazing angle X-Ray diffraction m easurem ents of Jedrecy et a l.[l] suggest that the relaxation o f both zinc and oxygen surface ions are slight and that the surface, w hilst stepped, features a high concentration o f vacancies in the terraces. A study by D uke et a l.[2] had previously found a larger relaxation for the zinc ions o f approxim ately 0.3Â, inducing an angle between the direction of the upperm ost Zn-O bond com pared to the equivalent bonds in the bulk. Scanning Tunnelling M icroscopy (STM ) evidence from Parker et al. [3] describes a surface which features frequent steps in the [10Î0] and [0001] directions. The m ajority of theoretical studies have reported results sim ilar to the findings o f D uke et al., with zinc ions relaxing into the surface further than the oxygen ions, usually by or slightly greater than 0.25Â .[4][5][6][7]. A previous investigation using atomistic potentials, conducted by N yberg et a l.[8] again found that the zinc ions relaxed into the surface m ore than the oxygen ions, by 0.22Â. The apparent differences between theoretical and experim ental studies such as vacancy concentrations, differing surface relaxations and the frequency o f steps is som ew hat puzzling and w ould seem to be a good problem to apply the newly form ulated atom istic potentials to.
The (1120) surface appears to be less of a contentious issue. Experim ental evidence reviewed by Henrich and C ox[9] concludes that the term ination at this surface is
C h apter 7 : ZnO surface studies
bulk-like with little relaxation o f any ions and this finding is repeated by the high quality electronic calculations o f W ander and H arrison[10]
The pair of polar surfaces, (0001) and (00 0Î), zinc and oxygen term inated respectively, are by themselves interesting, being both polar and stable. The nature of the stability o f these faces in regard to their polar character has to date not been successfully explained. The m ere fact that a truncated term ination w ould result in a significant dipole existing across the surface suggests that som e m ethod o f reconstruction w ould be expected and the debate about the nature o f this surface is still open. STM evidence from Parker et al.[11] suggests that the (0 0 0 Î) surface is covered in triangular pits and protrusions w hich m inim ise the num ber o f dangling bonds within the structure. M aki et al.[12] conducted A FM , reflection high-energy electron diffraction (RHEED) and coaxial im pact-collision ion scattering spectroscopy (CAICISS) on the (0001) surface and observed regular broad terraces with small zinc ion relaxation (outwards). Jedrecy et a l.[13] m ade sim ilar findings using grazing incidence X-ray diffraction, reporting that the (0001) surface exhibited small outw ards relaxations o f the zinc ions and that terraces on both faces w ere long broad. The Jedrecy study also found a larger inw ards relaxation o f 0.3Â o f oxygen ions on the oxygen term inated surface and suggested that partial occupancies play a significant role on both surfaces, with an occupancy factor o f 0.75 on the zinc term inated surface. However, the conclusion o f a low -energy alkali ion scattering experim ent by Overbury et al.[14] was that high levels o f vacancies w ere unlikely, although m ore support was given for three surfaces exhibiting no reconstructions. Further support for large displacem ents (0.24) o f oxygen ions on the oxygen surface is given by W ander and H arrison[15] w ho perform ed ab initio calculations on a slab model and report that the surfaces are stabilised by a charge transfer m echanism , rather than by vacancies. Atomistic potential based m odels, such as those perform ed by Nyberg et al.[16] are unlikely to find support fo r unreconstructed polar surfaces due to their inability to transfer charge between the constituent ions. Indeed, Nyberg et al. report that the polar surfaces could only be stabilised in their m odel by means of introducing vacancies or faceting. Overall, the picture o f these surfaces is unclear
C h apter 7 : ZnO surface studies
T his study o f zinc oxide surfaces will m ainly focus on the structure, relaxation and vacancy concentration o f the (00Î0 ) surface, using the new ly constructed interatom ic potentials, as described in C hapter 6. The other m ajor surfaces have also been investigated, and the findings are also detailed in this chapter.