Direct CoSi 2 Thin Film Formation with Uniform Nanograin-sized Distribution by Oxide-mediated Silicidation
3.3 Results and Discussion
Fig 3-1 shows the bright field TEM cross sectional image of the as-deposited sample using Si (002) edge on imaging condition. From the image, apparently, both Shiraki oxide and Co layer are pretty uniform with sharp interface and the thickness is calibrated to be about 2 nm and 4 nm, respectively.
Fig. 3-2 (a) and (b) are the TEM bright field plan-view images and diffraction patterns of the reactive silicide upon annealing with all the other layers removed, where the annealing conditions are (a) 600 °C 90 sec (b) 460 °C 120 sec followed by 600 °C 120 sec. Both diffraction patterns show that the silicide is CoSi2. Although Co is deposited by sputtering, the resulting phase agrees with that of Tung et al. [8-10]
bypassing CoSi, Co2Si. Because the SiOx acts as a diffusion barrier, the Co diffusion rate toward the Si substrate is reduced. According to Vantomme [16] and Pretorius [17], if the Co effective concentration at the Cobalt silicide growth interface is low enough, this will lead to the biggest negative change in the free energy for the CoSi2
formation. In addition, the direct CoSi2 formation can effectively reduce the formation temperature because the reaction path is Co + 2 Si Æ CoSi2 rather than CoSi + Si Æ
CoSi2, in other words, it need not to break the CoSi bonding [18].
Comparing Fig.3-2(a) and (b), the most striking feature is the average grain size and the grain size distribution as shown in Fig.3-3, which are 24±7.0 nm and 5±1.8 nm corresponding to the one-step and two-step annealing, respectively. Both diffraction patterns in Fig.3-2 show strong 111 preferred orientation, while the diffraction rings in Fig.3-2(b) appear to be much more diffuse than Fig.3-2(a) confirming to the much smaller grain size in Fig.3-2(b). Apparently, the two-step annealing produces smaller and more homogeneous grain size distribution than the one-step annealing although the sample upon the two-step annealing also experiences the higher temperature annealing (600 °C) for even longer time of 120 sec. The difference on grain size and its distribution must be due to the number of nucleation sites, which is closely related to the temperature. Due to its kinetic limitation, the nucleation site should be largely determined by the diffusion process through the SiOx
layer. Therefore, we speculate that the number of nucleation sites is closely related to the microstructure of the SiOx, which could be a function of the annealing temperature.
From the study of Fitch et. al [19], they show that the SiOx would become denser toward more stoichiometric SiO2 upon annealing and eventually turn into SiO2 at 900
°C for 30 sec. Baten and Fedorovich [20,21] have found that cobalt diffuses through SiO2 without any chemical interaction with the SiO2 networks but only occupies
interstices of the very open SiO2 structure and migrates along the interstices as diffusion channels without affecting the regular lattices. The SiOx microstructure has more interstices than SiO2, thus more interstices channels exist in the SiOx. Therefore, it is supposed that higher annealing temperatures (600 °C) will reduce the number of interstices channels because of more stoichiometric SiO2 network and hence the CoSi2
nucleation sites. Fig.3-4 shows two FTIR spectra from the SiOx layers of two samples upon annealing at 460 °C for 5 min and at 600 °C for 5 min, respectively with all the other layers on top of the SiOx layers removed. From Fig.3-4, the frequency spectrum from the 460 °C annealing sample exhibits a peak at about 1050 cm-1, while that from the 600 °C annealing sample reveals another peak at about 1075 cm-1. According to Chao et al. [22], the absorbance frequency of SiOx is from 1025 to 1060 cm-1 and that of stoichiometric SiO2 is 1075 cm-1. Thereby, the FTIR results support our supposition in that, upon 600 °C annealing, the SiOx layer indeed gradually turns to SiO2. However, the two-step annealing also experiences 600 °C annealing at the second step, why the nucleation sites are not reduced? The reason is that once the diffusion channels have been formed and completely stuffed with cobalt atoms at the lower temperature (460 °C), the channels will impede oxygen diffusion and cause the SiOx
microstructure remains open for cobalt diffusion continuously at higher temperatures.
To further justify the assertion made in the mechanism responsible for the
homogeneous nanograin-size distribution formed in Fig.3-2b, another two-step annealing experiment is performed with the annealing time half of that used in Fig.
3-2b to study the nucleation and growth in the earlier stage. Fig.3-5a is the TEM bright-field plan view image of this sample, where the annealing condition is 460 °C 60 sec followed by 600 °C 60 sec. The image shows that the sample is still in the process of grain growth, and exhibits bimodal size distribution with the peaks centered at the grain size of 6 nm and 15 nm as shown in Fig.3-5b. Following the above discussion, we presume that the large-sized grains have undergone nucleation at 460 °C and growth further at 600 °C, while the small-sized grains only undergo nucleation at 460 °C and growth is stopped at 600 °C. The annealing time of 60 sec at 460 °C apparently is not enough to let enough Co diffuse through the SiOx layer and form enough Co-stuffed channels. Therefore, the subsequent 600 °C annealing close up some uncompleted Co diffusion channels by forming more stoichiometric SiO2
network, which results in the bimodal size distribution. Fedorovich [21] measured the diffusion coefficient of Co diffusion in SiO2 film as D = 10 -7exp[(-1.7±0.05)/kT]
cm2/s. Based on this equation, this requires 3000 and 40 minutes for Co to diffuse 2nm thick SiO2 at 460°C and 600°C, respectively. However, the SiOx microstructure is looser, the activation energy should be smaller and the required time is expected to be much less.
In addition, Detavernier et. al., [23] found that at even higher annealing temperature of 850 °C, high resistive CoSi phase forms by a lateral growth phenomena where Si from the substrate diffuses through the CoSi2 and reacts with the remaining Co to form CoSi following CoSi2 nucleation directly underneath the weak regions of SiO2. Therefore, the higher temperature annealing tends to obtain bigger grain size, however CoSi, which can be eliminated by the two-step annealing.
3.4 Conclusions
CoSi2 thin film with homogeneous nano grain size of 5 nm can be obtained by oxide mediated silicidation in which cobalt is deposited by DC magnetron sputtering on SiOx/Si with the SiOx as a mediated layer followed by ex-situ two-step annealing (460 °C 120 sec and 600 °C 120 sec). The microstructure of the CoSi2 film can be altered by controlling nucleation and growth conditions. It is found that enough annealing time at the lower temperature of 460°C facilitates Co diffusion through the SiOx layer and forms the diffusion channels, which result in more nucleation sites and homogeneous nano grain size distribution. Once the diffusion channels have been formed at 460°C, they will remain open for the subsequent annealing at higher temperatures.