In order to prevent the chemical reaction between Fe deposition and Si (111)-7×7, the surface was adsorbed by CH3OH according to the procedure above. As shown in Fig. 1a, the sticking probability for CH3OH adsorption would be stable after the exposure situation about 10− 6 Pa, 30 s. By calculating a series of STM images, CH3OH is saturated at the adatom coverage of 1/2. Every triangular half unit cell contains three Si-OCH3 (shown as relatively bright dots) and three Si adatoms (shown as relatively dark dots) in the STM image. H atoms saturate the deeper rest site, which cannot be observed clearly. With the help of mass spectrometer, the dissociation process was proved as:
(1)
so as to establish its adsorption model (Fig. 1b). The dissociation process was deducted in Fig. 1c, suggesting the existence of a transformation state between methanol molecule and methanol/hydrogen ion. At the moment of bond breaking, the exact location of CH3OH had been specifically discussed. If we regard the whole CH3OH adsorption as a process of “destruction” and “reconstruction” to the surface potential, two dissociation conjectures were proposed in Fig. 1c. Greater destruction implies the deeper Si atomic layer, while the relatively weak reconstruction is presented in the form of surface quasi-potential. The interesting phenomenon in CH3(CH2)n−1OH dissociation, confirmed that the key to transformation state is O-H bond rather than C-H bond. That is, the formation of rest Si-H bond is earlier than that of adatom Si-O bond. Combined with the bigger attractiveness of electron-rich Si rest sites, it can be inferred that the conjecture (2) is much closer to reality than conjecture (1). In our transformation model, the rest Si atomic layer not only determines the location of dissociation process, but also implies the deeper/lower influence range of surface quasi-potential (Fig. 1c). Besides, the specific mechanism of CH3OH adsorption provides an important reference to the later adjustment of the nitriding process.
After the steaming condition of 10− 6 Pa × 30 s, a linear structure of Fe clusters could be formed on the surface of Si(111)-7×7-CH3OH. On the horizontal direction (Fig. 2a), the distance between each linear structures was controlled within 10 nm. The height was measured about 1.15 nm, which indicates multi-layer atomic stack. Metal atoms (like Au, Sn and Zn) [27] also deposited on the same surface of Si(111)-7×7-CH3OH, while the obvious linear structure has not been found. This interesting phenomenon was assumed to be the result of ferromagnetic effect. With the help of MPMS-7T, more specific magnetic performance was measured as shown in the Fig. 2b. Although values are relatively small, the intensity of residual magnetization is better than that of the magnetic field, which is manifested in a linear structure on the horizontal direction. After that, a nitriding process was carried out at room temperature. Like CH3OH adsorption, NH3 also undergo a dissociation process on group sites of Fe adatoms. Fe atoms in direct contact with Si(111)-7×7-CH3OH have different performance from other Fe atoms. Specifically, under the influence of surface quasi-potential, some quasi Fe atoms are distinguished from linear Fe clusters, inducing the nitriding process greatly. Considering the influence range of surface quasi-potential, the number of quasi Fe atoms is very limited, mainly belong to the deep Fe atomic layer on Si(111)-7×7-CH3OH surface. Therefore, the bottom layer could be defined as the quasi-potential layer. Previously, CH3OH was used as an intermediate layer between Fe and Si atoms. Now, top Fe atomic layer itself becomes a “barrier” between NH3 and surface quasi-potential. Figure 2c shows the scanning result of initial nitriding experiments, no obvious linear structure was found. As an important intuitive parameter, linearity is not only a necessary condition for future application, but also represents the magnetic performance of storage units. Just as the magnetic result shown (Fig. 2d), we can only find a worse atomic distribution on the surface of Fig. 2c. With the introduction of NH3, the magnetic field strength was only slightly improved, but the intensity of residual magnetization was greatly weakened. Moreover, in the absence of enough N ions on the surface, it is difficult to find an obvious nitriding peak in XPS (Fig. 2e). Low nitriding efficiency suggested the low dissociation efficiency, that is, only by improving the induced effect of surface quasi-potential can the formation probability of iron-nitride be increased. Refer to the transformation state of CH3OH adsorption, the number of Fe atomic layers become the key to the whole nitriding process as well as the magnetic enhancement. In this way, we tried to control the Fe deposition more precisely. As shown in Fig. 2a, Fe atoms were mainly stacked on Si(111)-7×7-CH3OH surface in the form of multi-layer. In the process from the pure metal wire to the substrate surface, Fe atoms were steamed with Ar introduction (Fig. 2f). Figure 3a showed a single Fe atomic layer, whose cluster structure and linear structure had not been fully formed. After that, the sample was nitrided under the same conditions of Fig. 2c. As can be seen from the new results (Figs. 3b and 3c), both atomic distribution and magnetic performance have been significantly improved. Forming a linear structure, our iron-nitride clusters have better magnetic performance than Fe clusters, especially the intensity of residual magnetization. Moreover, higher nitriding efficiency directly leads to the obvious nitriding peak in Fig. 3d. The XPS result was clearly in favor of the existence of Fe-N bonds with a N1s binding energy (398.1 eV) and agrees fairly well with the N1s values of surface iron-nitride, which occur in the 396.2-398.3 eV range. As a result, the better induced effect of surface quasi-potential on nitriding process was verified.
With the help of mass spectrometer (Figs. 4a and 4b),the process of NH3 dissociation can be detected gradually. The key to the nitriding efficiency is deducted as sufficient N ions. When Fe atomic layers are stack to a certain extent (like Fig. 2c), we can only find a small number of N ions. Specifically, the dissociation products of NH3 were mainly NH2− or NH2−. After adjusting Fe atomic layers, as the proportion of NH2−/NH2− decrease obviously, the intensity of N ions and H ions increase. One of the most significant phenomena is that there was almost no sign of the NH2−. It can be proved that a transformation state exactly divides the dissociation process of NH3 into two parts. That is, before and after the breaking of N-H bonds (Fig. 4c). NH3 does have a similar transformation state like CH3OH, that is, NH3 first adsorbed at lower sites of Fe deposition. Under the induced effect of quasi Fe atomic layer, NH3 is dissociated efficiently. After breaking the bond, N atom would rise to the top surface and be fixed with Fe-adatoms, forming iron-nitride. At last, the application of these magnetic clusters was discussed. Some dimension measurements were carried out around the new iron-nitride sample. It can be found that height values of each clusters in the linear structure are basically maintained at a stable value. Besides, the width of linear structures was still controlled bellow 10 nm. After that, air was introduced into the observation chamber. Without the adverse effects of oxidation, the stability of iron-nitride clusters is surely better than that of pure iron clusters. After repeated scanning of XPS after air introduction, no peak represents N-O or Fe-O bond was found. From the viewpoint of Si peak (Fig. 4e), in addition to Si-Si bonds, some Si-O bonds were newly found after exposed in the air environment. It can be proved that O2 gas would react with Si atoms rather than N or Fe atoms. These interesting phenomenon are worthy of our deep investigation and utilization, as well as the foundation for high density magnetic storage application.