3.1. Composition of as-Implanted SiO2/Si and Al2O3/Si Samples
The study of the chemical composition of the implanted samples before and after annealing was carried out by the XPS method, which makes it possible to implement elemental analysis by taking into account the shifts of the photoelectron lines that are caused by the chemical bonds of atoms with their environment. In our previous work [
22], the distribution profiles of implanted gallium and oxygen atoms in SiO
2/Si samples that were not subjected to post-implantation annealing were investigated. It was established that, even in the absence of annealing, all samples contain not only gallium in the elemental state, but also oxidized gallium in the stoichiometric Ga
2O
3 and the oxygen-deficient Ga
2O states. The distribution profiles of gallium (
Figure 3) indicated the presence of two maxima. The first one predominantly contained gallium in the elemental state and was closer to the surface. The second maximum was located at depths that were close to the maximum of the implanted impurity distribution and deeper. Moreover, almost all implanted gallium at this depth was in the oxidized state. Additionally, of interest is the fact that the lines associated with the presence of Ga-O bonds were observed even in the absence of additional implantation by oxygen ions. This indicated the participation of oxygen from the SiO
2 matrix in the oxidation process.
As can be seen in
Figure 3, the order of implantation of Ga
+ and O
+ ions had a significant effect on the concentration and profile of gallium with different oxidation states. Indeed, the implantation of oxygen prior to the implantation of gallium led to a higher total concentration of the latter, as well as to a higher degree of its oxidation; meanwhile, the implantation of O
+ after Ga
+ does not contribute to the oxidation of previously implanted gallium, and it also causes its significant desorption.
This can be explained as follows. It is known that ion implantation is characterized by the appearance of so-called “thermal spikes” during the introduction of ions [
24]. This phenomenon is more pronounced for heavy ions, which form dense displacement cascades. Gallium is one such ion. Apparently, during its implantation in SiO
2, local heating in the region of ion tracks promotes the reaction of Ga with matrix oxygen and the formation of gallium oxides. Local heating is most pronounced in the region close to the average projected range
Rp(Ga), where the energy density released in elastic collisions is the highest, hence the local temperature is higher; on the other hand, the Ga located closer to the surface is partially preserved in the elemental state and forms the near-surface maximum observed in
Figure 3a. If the matrix is additionally enriched with oxygen by irradiation with O
+ before Ga
+ implantation, the degree of Ga oxidation becomes even higher. Moreover, since the unbound gallium can undergo radiation-stimulated desorption, the pre-implantation of O
+ weakens it due to a more complete binding of gallium to oxygen in the SiO
2 matrix. Oxygen ions, being lighter, create less dense cascades compared to Ga
+ and, for them, the effect of “thermal spikes” is expressed to a much lesser extent; therefore, O
+ implantation after Ga
+ implantation does not contribute to the oxidation of Ga, and it even leads to an additional radiation-accelerated desorption (apparently due to the knocking of Ga out of the lattice sites and its transfer to a more mobile interstitial state).
For the implanted Al
2O
3/Si samples, the situation is different (
Figure 4). The main difference from the case of implantation into the SiO
2 matrix is the lower concentration of gallium at the same implantation dose. Another important peculiarity is the almost complete absence of oxidized gallium in the case of only Ga
+ implantation without additional irradiation with oxygen. This indicates that the oxygen from the matrix hardly participates in the oxidation of implanted Ga.
In the case of oxygen implantation prior to gallium, two maxima appear on the depth distribution profiles of gallium: the first one mainly consists of gallium in the elemental state, while the second one is due to the presence of oxidized gallium in the Ga
2O and Ga
2O
3 states. A similar situation was also observed in the SiO
2/Si matrix; however, there was a noticeable shift in the elemental gallium profile that was closer to the surface. In addition, a common feature of the distribution profiles in the Al
2O
3/Si matrix is a significantly greater impurity depth compared to that calculated using the SRIM program (
Figure 1). In the case of oxygen implantation after gallium ions, the total gallium profile is close to the case of O
+ → Ga
+ implantation, but a bimodal peak structure was not observed in this case. In this case, the gallium atoms were predominantly in the elemental state, and only a small part of the implanted atoms (~1–2 at.%) was in the oxidized state with different degrees of oxidation. This is additional evidence that, in the absence of annealing, the participation of oxygen from the matrix for oxidation is hindered. This process also requires the matrix, into which gallium atoms are implanted, to be initially oversaturated with oxygen.
A separate consideration of the peculiarity of the Ga depth distribution in the Al
2O
3/Si matrix, which is associated with a significantly greater impurity depth compared to the calculated one, shows that this effect is observed for all variants of implantation. However, it has its peculiarities in each case. In the case when only gallium atoms are implanted, the shift of the maximum is ~10 nm. Upon implantation of Ga
+ → O
+, this shift increases. In the case of oxygen implantation before gallium ions, the first maximum approximately coincides in position with the previous ones, and the second maximum relates mainly to gallium bound with oxygen, which is located at much greater depths. The nature of this phenomenon is associated with the effect of radiation-accelerated diffusion. The fact that the increase in the depth of gallium distribution is more pronounced in the case of additional oxygen implantation can be explained as follows. If it is assumed that gallium diffuses by the interstitial mechanism, then the presence of excess oxygen vacancies in the irradiated film leads to their capture of diffusing gallium atoms, which slows down diffusion. In the case of preliminary irradiation with oxygen ions, some of the oxygen vacancies are filled; as a result, the diffusion flux of gallium atoms increases. The preferential capture of gallium atoms by oxygen vacancies in the absence of preliminary irradiation with oxygen ions is also confirmed by the low concentration of Ga-O bonds (
Figure 4a) since the gallium atom that has fallen into the vacancy position is surrounded by other oxygen atoms and does not form other bonds with oxygen. In the case of the reverse order of implantation, the effect of the shift toward greater depths is less pronounced.
3.2. Composition of Implanted SiO2/Si and Al2O3/Si Samples after Annealing
The distribution profiles of implanted gallium atoms in the samples after annealing at 900 °C in a nitrogen atmosphere are presented on
Figure 3d–f. A common feature for the SiO
2/Si samples implanted in different regimes is a significant decrease in the concentration of gallium atoms in the elemental state. At the same time, no noticeable decrease in the total concentration of gallium was observed. For a sample implanted only with Ga ions, the near-surface peak related to elemental gallium disappears after annealing. Together with the process of oxidation of gallium, its diffusion toward greater depths is observed. In this case, oxidized gallium is in an oxygen-deficient Ga
2O state and its concentration in the stoichiometric composition practically does not change after annealing. This confirms the hypothesis that there is a critical concentration of oxygen atoms in the matrix, which can participate in the formation of Ga-O bonds. After this formation is complete, the process of oxidation of the implanted gallium significantly slows down. In the case of Ga
+ → O
+ implantation, the redistribution of gallium and its insignificant diffusion toward the surface are also observed. Annealing of these samples leads to the additional oxidation of both metallic gallium and gallium in the oxygen-deficient state, which contributes to an increase in the fraction of gallium in the stoichiometric Ga
2O
3 state.
Finally, for the O+ → Ga+ implantation order, the total distribution profile of gallium atoms remains almost unchanged after annealing. The oxidation of gallium, which is in the elemental state before annealing, was observed. For this order of implantation, there was a slight decrease in the concentration of gallium in the stoichiometric Ga2O3 state with a simultaneous increase in the concentration of Ga-O bonds with a lack of oxygen. Based on the presented data, it can be concluded that the final distribution of gallium atoms in various chemical states is determined not only by the diffusion of gallium (both radiation-stimulated and thermal diffusion during annealing), but also by the possible diffusion of oxygen atoms from this region. This assumption is supported by the following facts. First, according to the results obtained for the case of implantation of Ga ions only, it can be noted that there is a limited number of oxygen atoms from the matrix that can participate in the oxidation process. Second, it can be expected that the oxidation of gallium in the elemental state as a result of annealing can be determined not only by the processes of chemical interaction at elevated temperatures, but also by the influx of oxygen atoms into the near-surface region from the depth.
Since a significant decrease in the concentration of oxidized gallium in the stoichiometric state of Ga
2O
3 was observed for this variant of ion synthesis, an attempt to increase it was made by annealing the samples in an oxygen atmosphere, while all other annealing conditions (temperature and duration) remained the same. The study of the chemical composition revealed an interesting effect—namely, the almost complete oxidation of implanted gallium to the stoichiometric Ga
2O
3 state (
Figure 3g). This fact additionally supports the conclusion that, for a more efficient oxidation of gallium, it is necessary to provide such conditions that ensure the sufficient oxygen content in the implanted region for obtaining the stoichiometric composition of Ga
2O
3. In our case, during the implantation of Ga
+ and O
+, the concentrations of oxygen and gallium atoms were equal, and the additional participation of oxygen from the matrix was insufficient for obtaining a stoichiometric composition. This fact requires additional verification and will be investigated in the framework of further research.
The bimodal distribution of gallium, which is observed in almost all distribution profiles, especially for samples after annealing, is noteworthy. It can be assumed that the formation of an additional peak, which is closer to the surface compared to the calculated maximum of the impurity distribution, is associated with the formation of radiation defects, and the distribution maximum of which is usually located closer to the surface. The sink of gallium atoms into the defect-enriched region can occur due to radiation-stimulated diffusion, which leads to the appearance of an additional peak in the gallium concentration profile.
The distribution profiles of Ga atoms in the implanted Al
2O
3/Si samples after annealing are presented on
Figure 4d–f. A common feature of all the samples was the loss of implanted gallium as a result of annealing. This effect was the most pronounced for the samples implanted only with Ga
+, in which an almost uniform distribution of gallium with a concentration of ~2 at.% was observed. For a sample implanted in the O
+ → Ga
+ order, a decrease in the total concentration of implanted gallium was also observed with its simultaneous diffusion to the surface. It should be noted that, despite the decrease in the total concentration, gallium in the samples after annealing predominantly remained in the fully oxidized Ga
2O
3 state. Finally, for the Ga
+ → O
+ implantation order, the impurity loss was less pronounced. This may be due to the formation of stable complexes containing gallium and oxygen atoms since, compared to the reverse order of ion implantation, this profile does not split into elemental Ga and gallium in the oxidized state. In any case, after annealing, the total concentration of gallium in the oxidized state for all the used variants of ion synthesis was too low for the efficient formation of an array of Ga
2O
3 nanoinclusions in the Al
2O
3/Si matrix.
3.3. Structural Properties of Implanted SiO2/Si Samples after Annealing in the Oxygen Atmosphere
According to the XPS data, the most effective in terms of obtaining the maximum concentration of oxidized gallium in the state of stoichiometric Ga
2O
3 was the SiO
2/Si sample that was irradiated first with O ions and then with Ga ions after annealing at a 900 °C in an oxygen atmosphere. In this case, the efficiency of the formation of chemical bonds corresponding to the stoichiometric Ga
2O
3 composition exceeded 90%. To study the structural properties of this sample, the method of transmission electron microscopy, including a high resolution, was used.
Figure 5 shows the cross-sectional patterns of the SiO
2/Si samples: the O
+ + Ga
+ structure after annealing in an oxygen atmosphere at 900 °C. In the overview of a bright-field TEM image (
Figure 5a), the formation of two layers of dark contrast was observed. This correlates with the XPS data, according to which the gallium distribution profile exhibited a weak peak at depths of ~20 nm, and a broad intense peak at a depth of ~60 nm.
The high-resolution image obtained in the region of darker contrast (
Figure 5b) shows the formation of a large number of spherical nanoinclusions with sizes of ~4–9 nm, some of which are indicated in the figure. For the inclusions with atomic planes observed in the TEM images, processing was carried out by measuring the interplanar distances and interpreting the diffraction patterns obtained by the Fourier transform method. An analysis of the obtained data shows that Ga
2O
3 nanoinclusions are formed in two different crystalline phases, β-Ga
2O
3 and γ-Ga
2O
3. This partially agrees with the previously obtained data on the ion synthesis of the nc-Ga
2O
3 obtained by annealing in a nitrogen atmosphere, for which the formation of β-phase nanocrystals was found [
22]. One of the possible explanations for the observed difference can be the following. As was assumed earlier [
22], during the ion synthesis of Ga
2O
3 nanocrystals, chemical bonds between gallium and oxygen (Ga
xO
y complexes) were formed after irradiation. These bonds serve as the centers for the nucleation of a new phase. During annealing, the process of nanocrystal formation goes through several stages. First, small non-phase inclusions containing Ga and O atoms are formed as nuclei, which can later be enlarged due to the addition of new atoms. With an increase in annealing temperature, the formation of nanoinclusions in the metastable γ-Ga
2O
3 phase that contain a large number of structural defects is possible. A further increase in annealing temperature and time leads to the preferential transition of nanoinclusions to the stable β-phase. However, in the annealing conditions used in this work (oxygen-containing atmosphere), this transition did not complete; therefore, the presence of the nanocrystals of both phases was observed in the samples. Comparing these data with the previously obtained results for the case of step-by-step annealing in a nitrogen atmosphere [
22], it can be assumed that, along with the influence of the annealing atmosphere, the “history” of heat treatments can also play a certain role. The mechanism of phase transitions in nc-Ga
2O
3 under various heat treatment conditions requires separate investigations.