Doping Effect Of PtSi / Si Interface And Theoretical Study On Si Surface Electron Interaction In Oxide Growth

Since silicon has a wealth of resources and perfect properties, scientists predict that it will continue to play a dominate role in electronics industry in the next decades. Therefore, the development of silicon based electronic devices has two key directions: one is using new materials, such as metal-semiconductor junctions, to replace the conventional semiconductor PN junctions, which makes devices smaller. The other is integrating new materials on silicon, such as crystalline complex oxides, which makes devices more multifunctional. In this work, we use first principles calculations to study the problems in these two aspects to explain the relative experiments.The discussion in this thesis includes two parts. After introducing the motivation and the theoretical framework for our study, we investigate the doping effects on modulating Schottky Barrier Height (SBH) for the PtSi/Si interface in chapter3. The inner relationship between doping and SBH modulation is explained in detail. We focus on the problems in growth crystalline oxides on silicon surface in chapter4. We demonstrate the role of electrons on surface dangling bonds in closing oxidation channels in special conditions.Metal-semiconductor junctions usually have large SBH values, which can be modulated by some methods to meet the application requirements. In these junctions, PtSi(010)/Si(001) is widely studied because of its small p-SBH value. Experiments found interfacial heavily doping is an effective way to modulate its SBH. However, the doping range in the experiment is very wide (-20nm), which is difficult for first principle calculations to simulate. Moreover, previous theoretical studies carried out the calculation regardless of the interface structure. It is in contrary to the bond-polarization theory, which points out that the structure affects the SBH value. Therefore, a more detailed ab inito investigation that takes the structural effect into account in the doping effect in PtSi/Si interface is required. The bulk property of PtSi and the change of it under strain on the Si substrate are studied first. We found the weak metallic properties essentially remain unchanged after the surface structureal deformation. Based on the analysis of PtSi bulk structure, two reasonable atomic alignments for Si and PtSi layers at the interface are simulated. The calculation results show the stable structure is characterized by the Pt-Si interface bonding instead of Si-Si bonding. In this interface, due to the presence of Pt-Si polar bond, the Si atoms at interface layer of both materials are charged:Bader analysis shows each Si atom in PtSi side gets～0.5electrons, while each Si atom in the Si side loses-0.4electrons. Overall charge neutrality is satisfied by contributions from other atoms. Thus, a dipole presents, which enhance the p-SBH value. When different kinds of donors and acceptors are doped at the interface, it is equivalent to introduce new dipoles to modify this original interface dipole, which further changes the barrier.The key to determine SBH in first principles calculations is to determine the line-up of the reference levels (averaging electrostatic potential) at the interface. In order to reveal individual doping effects on the alignment of the reference levels and further on the SBH of PtSi/Si(001), we study individual doping of elements (Mg, S, B, Ga, P, and As) in different layers and different positions. This simulates the real case doping because of the continuity of electrostatic potential and the principle of potential superposition. The results show that the donators and acceptors in the Si-side enhance and lower the SBH, respectively. And the results mainly are opposite while doping in the PtSi-side. These laws in line with the expected modulation effect caused by the dipole:donors transfer electrons to the other side through interface bonds, while acceptors get electrons. Therefore, the charge transfer introduces new interface dipoles to modulate the SBH.In addition to the qualitative relationship discussed above, we find a good linear relation between the changes of SBH values and introduced dipoles by36kinds of dopings. Furthermore, we find this coefficient does not depend on the doping situations but rather like a nature property of the interface. We use a parallel plate capacitor model to estimate this physical value, called effective dielectric constant, which is4.92with a small standard error of0.21.Finally, we check whether the overall doping effect can be considered as a simple summation over the contributions of all the individual dopants when the dopants are far from each other. We first sum the SBH modulation of two individual dopants, and compare the result to the co-doped case. For different kinds of elements, the difference is below0.01eV. Since the dipole in the metal is small, these results explain why the total modulation effect always has the same direction with the Si-side doping effect, while the doping reign covers both metal-side and Si-side.In chapter4, we focus on the problems in the growth of crystalline oxides on silicon surface in special conditions. The mechanism of this growth is important for producing multi-functional field effect transistors. Previous experiments found that amorphous silica present at the interface reduces the good properties of the interface, when crystalline oxides (such as SrTiO3) are grown on silicon. Recently, it is found that in order to avoid the oxidation of silicon substrate, a monolayer Sr should be pre-grown on the clean Si(001) surface with precisely controlled experimental conditions. However, in such a special condition (low oxygen concentration, low energy and low temperature), its physics and the role of Sr prior to the growth of SrTiO3remain to be explored.In such a special growth condition, the oxidation process can be considered as no barrier. To investigate the mechanism at the atomic lever, we use ab inito calculations to simulate a single oxygen atom barrierless adsorbing on Si(001), and explore the role of electrons on the surface dangling bonds in closing these oxidization channels. Our results show that for the barrierless adsorption of a single O atom on Si(001), there are4kinds of stable(metastable) configurations, depending on the start-sites of the O incoming on the surface. Among these4configurations, the BB (oxygen on the back-bond of the surface dimer) is the only one to form the Si-O-Si bond with breaking the Si-Si bond, and its configuration is very close to the corresponding structure for vitreous silica measured in experiments. Therefore, the BB configuration should be considered as the oxidized structure in the initial stage.Calculations show that in contrast to the conventional silicon oxidation, only a few oxidation channels exist. The initial positions of these oxidization channels are far away form the dangling bond of Si atoms in the surface dimer. And for the left initial points close to these atoms, oxygen atoms mainly prefer to stay on the dangling bonds. This is due to the electrons on Si(001) surface can transfer to the oxygen atom, which reduces the ability of O to get more electrons to break the backbond. In each dimer, the up atom of the Si dimer is p3-like hybridized and its dangling bond consisting of a remaining s orbital is fully filled by two electrons, which can transfer to O, and attract it to the dangling bonds. The down-atom is sp2-like hybridized, leaving one pz orbital not in bonding as an empty dangling bond, which can get electrons from nearby dangling bonds of up atoms of Si-dimer. In this process, the p3hybridization on up-atom is reduced and form sp2hybridization, while the sp2hybridization on down-atom is reduced and form p3hybridization, and the dimer flips in order to adapt to the electron redistribution. In this way, the down-atom can also attract O to its dangling bond. The calculation results of projected density of states (PDOS) on the up-atom of the absorbed dimer in different absorption processes support our argument.In the conventional silicon oxidation, oxygen gas in the air takes away the electrons on the dangling bonds of Si(001) surface immediately. The oxidation process happens quickly after these electrons are exhausted. However, in special conditions in the experiment, the quantity of oxygen in the air is limited. If enough surface electrons are provided to the dangling bonds, the oxidation channels may be narrowed or even closed. In order to further support the argument, we performed numerical experiments by doping Ga (hole) and As (electron) on Si(001). The results show that, when holes are introduced (one per4×4unit cell), the area of breaking the BB bond in the4×2unit cell around it is increased by about45.5%.When the electrons are introduced, the area is reduced by about18.2%.Therefore, when a monolayer of Sr atoms covers the Si(001) surface, Sr atoms can provide electrons to the dangling bonds of the surface Si atoms to close the oxidation channel. We further calculate the potential energy surface (PES) of single oxygen atoms in the the1monolayer Sr-covered Si (001) surface. The results show that the Sr atoms are located at the positions of oxidation channel, and the potential is increased by Sr in these areas to block the oxygen to get in. These results can further illustrate the role of surface electrons in closing oxidation channels.At last, the significance of these two works and further research are summarized.