In
Figure 1 the silicon and oxygen concentration profiles in XPS and the percentage of silicon and oxygen in the single-layer films of SRO
25 and SRO
100 are shown. It can see that the X
Si of the SRO films decreased from 9.9% to 5.3% and from 10.0% to 5.0% for SRO
25 and SRO
100 films with T-A, respectively. It can also be seen that the refractive index (η) was 2.46 and 2.04 to SRO
25 and SRO
100 films as-G, respectively, and with T-A, the η decreases to 1.3 and 1.02, respectively, where we had an oxygen deficiency of −9.7% and −9.5% for SRO
25 and SRO
100 as-G, while the oxygen deficiencies for the T-A SRO films were −6.2% and −5.0%, respectively.
To explain the behaviors that occur with respect to the silicon and oxygen contents and the refractive indices, we have that in a previous investigation [
23], SRO films deposited under the same conditions and using the same HFCVD equipment as the used in this research were analyzed, in that previous research, the same behavior is observed. In this case, it is was explained that the films deposited at a lower hydrogen flux level (25 sccm) without annealing present a higher content of silicon, just as in the results of this investigation, in the HRTEM images of the quoted research, it is shown that the films tend to present agglomerations of amorphous silicon of varied sizes but in general more significant than 15 nm, in the case of higher hydrogen fluxes level (>100 sccm), the silicon content decreases, such case yields that the HRTEM images show a decrease in the size of the silicon agglomerates, when the films are thermally annealed, a diffusion of silicon and oxygen occurs where it is formed a SiO
2–SiO
x matrix where silicon agglomerates with sizes smaller than 2 nm are immersed, these agglomerates show in some cases crystalline orientations, but their conformation also contains silicon in the amorphous state and oxidized terminals, this restructuring in the material causes the change in the refractive indices and silicon content.
Unusual behavior is the presence of a refractive index lower than the value of SiO
2, which the SRO
100 film presents after annealing, this behavior has been reported in [
35], and the explanation given to this phenomenon is that SiO
x film has a more amorphous structure than that of the SiO
2 film, this correlates with the fact that the SiO
x film is less dense and therefore has a lower refractive index [
21]. This phenomenon is also discussed arguing where it is established that the crystalline regions are separated by O-rich regions, these denoting Si and O dominated areas which are clearly separated, where the decrease X
si is related with O-rich regions.
The different behaviors between S-L and D-L could be explained as the S-L SRO films have a thinner thickness with a deposit time of 3 min, this deposit parameter permitted to SRO films to have more amorphous silicon and Xsi bigger, besides less Xo was clearly observed. Therefore, non-stoichiometric silicon oxide was much less stable, when is applied the annealing the Xsi decreases and oxygen increases due to O-rich regions, therefore refractive index decreased. The D-L SRO films had a thicker thickness with a deposit time of 5 min, and other conditions of deposited were realized; between layer and layer deposited, there was some like annealing. Therefore, a behavior difference was obtained, and the refractive index increased in a similar manner with our other works.
The pristine I-V curves obtained from both the as-G and T-A S1 and S2 structures with S-L and D-L are shown in
Figure 3 and
Figure 4, respectively. All were measured with the same voltage sweep from 0 V to 35 V after 35 V to 0 V, followed by 0 V to −35 V and closing the cycle from −35 V to 0 V, with the protection of circuit short of 100 mA. At first sight, the I-V curves corresponding to S1
as-G and S2
as-G structures exhibited current peaks with ups and downs at low voltages in both positive and negative polarizations, while those corresponding to S1
T-A and S2
T-A showed the typical characteristics of the I-V curves for MOS structures. On the other hand, it is also observed that the I-V curves illustrated higher currents for S1
25T-A to major voltages. The behavior of the hysteresis is shown clearly in the I-V curves of S1 and S2.
3.1. Curves I-V Pristine of Structures
In the pristine I-V characteristic curves of
Figure 3 and
Figure 4, different behaviors were identified which we will describe shortly as follows: Number 1 in these pristine I-V curves, it is observed that the S-L structures reached higher amounts of current in the first measurement as shown in
Figure 3a,b than the D-L structures of
Figure 3c,d. Another trend that structures presented was that films without annealing showed higher and lower peaks in current measurements at low voltages in both positive and negative polarizations. It was also observed in this behavior a sudden increase in current at a specific voltage. This phenomenon is linked with the nanostructure and its crystallinity which yields that the agglomeration of many electrons trapped in the Si-nps prevent the trapped charge’s movement and block electrical conduction [
36,
37,
38]. Therefore, this can lead to the creation and annihilation of preferential conductive pathways generated by adjacent stable Si-nps and defects such as unstable silicon nanoclusters (Si-ncls) and others through structural changes and the possible creation of defects due to Si–O, and Si–Si [
38]. Furthermore, on the return of the curve, it showed an increase in the current regarding the first measurement, where a charge trapping and state of less resistance to that of the first curve was observed. This behavior is due to the formation of conductive paths in the material; therefore, the return path was not the same, and the charge trapping was formed. This behavior is known as hysteresis [
35,
36,
37,
38]. This behavior was observed in all the I-V curves of the structures, and this effect occurred in both direct (DP) and reverse (RP) polarizations. But it is most noticeable in structures with heating treatment as well as in S-L SRO films with both polarizations, while in D-L SRO films the hysteresis in positive polarization was observed better. Number 2 in this case, the charge transport phenomenon was identified as a Coulombic Blockade, and its behavior was observed when the current increased sharply at a specific voltage, and it remained in this current as the voltage increased, this was so that since there was electrical conduction due to formation of a trapped electron configuration blocks [
5], also in the resistive switching memory structures with SiO
x or SRO, this behavior was attributed to the presence of a point charge which induced throughout the space the appearance of a force field which broke down in the process [
6,
7,
8,
9,
10,
11,
12,
13] moving the current to a state of low resistance. Number 3 current curve was identified as a region of negative differential resistance (NDR), almost always observed after the Coulombic Block. The form it presented was a series of tiny current jumped close to each other, called resistive switching, according to Yao et al., [
38,
39]. This means that, for a range of values of the applied voltage, an increase in voltage caused current to decrease rather than increase and takes place when electrons traveled at the same average speed; the space charge domain no longer grew, but the electrons continued their journey, and since the electric field was not large enough to form additional domains [
39,
40,
41], then a negative differential resistance region was created, this phenomenon was best observed in the as-G structures.
The curves of the T-A D-L S1,
Figure 3c,d and the as-G S-L S2
Figure 4a,b present more significant hysteresis or charge trapping [
34] than the T-A S2
Figure 4c,d and S1 as-G
Figure 3a,b which means that in the T-A S1 and S-L as-G S2 samples, respond to the creation-annihilation of conductive paths due not only to the Si-ncs but also to defects in the oxide found in the heterojunctions with SRO films according to Kalnitsky et al. [
41]. This negative-differential-resistance behavior in SRO/Si structures was observed through current-voltage (I-V) measurements, and the application of the electric field caused the electrical potentials to be distorted, favoring the quantum tunneling of electrons between the silicon-nanocrystal and the traps of oxide.
The graphs of current versus voltage for the S1 MIS and S2 MIM structures in
Figure 5 and
Figure 6 are the best of five measurements of each structure and are graphed in semi-logarithmic form. In the monolayers, we observed a current regime in the order of milliamperes when applying voltages between −25 to 25 volts, showing current variations in the as-G structures. However, the T-A structures showed a linear relationship between current and voltage until the current was maintained, and occasionally it dropped and suddenly increased, this happens again at that voltage, and current for which was possible to observe bright dots (electroluminescence) this phenomenon was observed only in the S1 MIS structures. We point out that in the S2 structures, no bright dots were observed. On the other hand, the as-G D-L structures showed abrupt increases and drops in current when increasing voltage, it was attributed to the creation and annihilation of conductive paths in the material [
38,
39,
40,
41,
42,
43,
44,
45,
46]. Further, we observed the Coulombic Blockade in these T-A structures. The sweeps with forward and reverse polarization yielded a current behavior similar to that of the S-L structures but with a current regime in the order of microamperes, where at this current and with voltages greater than 30 volts it was possible to observe greater bright points, suggesting the release of charge trapping in the SRO film, generating the conductive paths at currents and voltages greater to microampere and 30 volts, respectively [
41].
3.2. Conduction Mechanisms
The fact that the S1 and S2 structures presented similar results in forward and reverse polarization suggested that the carrier transport in this type of material was carried out through similar mechanisms [
44,
45] for both structures. To understand the transport mechanisms of the S-L and D-L SRO films, current density (J) measurements as a function of the electric field (E), in reverse polarization of the I-V curves plotted in
Figure 5 and
Figure 6 were analyzed due to these structures showed good electroluminescence. The reverse polarization occurred when the gate contact (ITO) was polarized with a negative voltage regarding substrate.
In general, four conductions mechanisms contributed to the carrier transport in these SRO films, namely: Ohmic (O), Hopping (H), Poole–Frenkel (PF), and Fowler–Nordheim (FN) [
34]. The current density-electric field (J-E) analysis depended on the dielectric thickness and the electric field applied to the MOS structure.
Figure 7 and
Figure 8 show the devices analyzed here in the semi-logarithmic J-E curves in reverse polarization for S1
T-A and S2
T-A, structures respectively. Furthermore, in the inserts of each figure are shown, the specific J-E graphs according to the conduction mechanisms in each section of the J-E curve are highlighted with linear regions that correspond to the Ohmic (O), Hopping (H), Poole–Frenkel (PF) and Fowler–Nordheim (FN) conduction mechanisms [
34,
46].
As can be seen, there were several conduction mechanisms in the J-E curve of these structures. That is, for low electric fields (≤1.6 MV/cm), the carriers reached enough energy to overcome the energy barrier at the Si/SRO interface and dominate the Ohmic conduction mechanism, as is shown in the inserts of
Figure 7a and
Figure 8a,c. Another predominant conduction mechanism was observed in intermediate conduction regime at low electric fields, in both S-L and D-L SRO films, this is called Hopping conduction (H) [
34,
47,
48,
49,
50], as seen in J-E curves in
Figure 7b–d and
Figure 8b,d, and it was originated by the trapped electrons jumping from one trap to another within the film SRO. Also, the energy of the trapped electrons may be less than the maximum energy of the potential barrier between the two traps; in this case, the trapped electrons may have continued traveling using the tunneling mechanism.
On the other hand, the Poole–Frenkel conduction mechanism was reported in SRO films where some electrons were found in traps and by thermal excitation were released so they could be conducted within the conduction band of the SRO films by applying an electric field through the dielectric SRO, then electrons crossed the Coulomb barrier which was reduced by the electric field and then increasing the probability of that an electron would be thermally excited from the trap becoming free to travel in the conduction band of the dielectric. The Poole–Frenkel (P-F) conduction mechanism depended strongly on the electric field, and it was independent of temperature, and the electric field was limited to low values (2 MV/cm) [
34,
47]; one can see this conduction mechanism in the inserts of
Figure 7a and
Figure 8a–d.
Additionally, the tunneling Fowler–Nordheim conduction mechanism has been proposed for the SRO films containing Si-ncs or silicon islands where electrons can tunnel by the effect of the electric field existing between the Si-ncs or silicon islands and generated by the potential barrier with a triangular shape (or another one) [
34,
51]. This generally occurs in not very thin dielectric films (>3.5 nm) and at high electric fields (>2 MV/cm), allowing carriers to overcome or tunnel barrier heights from one trap to another. The FN mechanism is the one that dominates [
34,
42,
43,
44,
45,
46,
47,
48,
49,
50,
51,
52].
EL emission occurs under reverse bias and is originated due to charge injection through conductive pathways and radiative recombination processes between energetic states of traps or defects [
53]. It has been reported that when the electroluminescent emission is presented in the form of points, it originates from the efficiently excited emission of defects in the oxide and/or of a few Silicon nanoparticles (Si-nps) [
54]. The Si-nps within the SRO films are randomly distributed so that various conductive paths are created; under this assumption, the current does not flow uniformly through the entire area of the capacitor but passes through conductive discreet paths within the oxide. An increase in the total current will result discreetly in a rise in the current density in each conductive path, which results in a more significant number of radiative recombination events and, therefore, a greater electroluminescent intensity [
47]. As the current increases, more charges flow through the Si-nps and can break off some of the Si–Si bonds (creating E centers). Consequently, the conductive paths are annihilated, resulting in current drops [
48].
Figure 9(a1) just shows this behavior where current increases and drops and the emission of points with different colors can be observed, it possibly indicates the intervention of different defects involved or as it is more commonly reported Si-nps of different sizes are participating [
50]. In [
50], it is said that electrons and holes are injected into Si-QD (quantum dots) by F-N (Fowler–Nordheim) tunneling through the SiO
x matrix. The existence of immersed Si-QDs leads to a decrease in the activation voltage of the F-N tunneling and creates a path for the carriers from the Si substrate to the ITO contact.
According with what it was obtained in this research, the FN conduction mechanism was presented in all structures when EL occurred, before this, the Poole–Frenkel (PF) conduction mechanism occurred at lower electric fields, which is related to electrons trapped in traps or defects which were excited towards the conduction band of the oxide. The full area EL emission obtained was due to the optimization of the injection of the carriers through the material by the cancellation of preferential conductive paths [
49].
As the density of Si-nps increased, a uniform network of conductive paths became possible, allowing for uniform charge flows across the entire area. Meanwhile, as the density of Si-nps decreased, the distance between them increased, reducing the number of available paths, with a resulting set of discrete and preferential conductive pathways within the oxide. Bright spots appeared when structures were operating within a region of high conduction. These jumps and drops in luminescence were due, respectively, to the appearance and disappearance of luminescent dots on the surface of the devices. After the current drop, the EL points disappeared completely, and EL was obtained in the entire area [
55].
3.3. Electroluminescent Structures
Figure 9(a1,b1,c1,d1) shows the reverse polarization (R.P.) I-V curves,
Figure 9(a2,b2,c2,d2) shows the electroluminescent spectra, and from
Figure 9(a3,b3,c3,d3), we can observe the bright dots and full electroluminescent bright area in each one of the respective photographs, belonging to the (Au/Si/SRO/ITO) S1 MIS-structures made up with S-L and D-L SRO films taking into account different deposit parameters.
From
Figure 9(a1), the T-A S1
25 MIS-structure in R.P. and X
Si = 5.3%, the current curve exhibited that when having −20 volts an abrupt current drop from 10 mA to 130 uA happened, at the same time brightly colored dots appeared. As the voltage varied to more negative values, the number of brightly colored dots also increased; however, due to the current drop in which it remained low, such fact provoked that the T-A S1
25 structure did not present a uniform EL emission over the whole area. The phenomenon that we report with this structure was similar to that published in [
48] when the carriers did not flow uniformly through the whole structure area, but they passed through discrete conductive paths within the SRO film, as shown in
Figure 10. Consequently, the structure showed a spectrum with two outstanding peaks, one emission peak centered at around 450 nm and the other one at around 580 nm; these emission peaks remained practically at the same wavelength, but their intensities were increased as the voltages were more negative, as shown in
Figure 9(a2). These EL spectra emission bands at 450 nm were associated with defects in neutral oxygen vacancies (NOVs), while the other emission of 580 nm was attributed to positively charged oxygen vacancies [
10,
24]. It has been reported that the EL emission peak placed in the blue band region increased its excitation voltage due to the contribution of small silicon nanoparticles (Si-nps) [
44].
The EL spectra of the T-A S1
100 MIS-structure and X
Si = 5.0% are shown in
Figure 9(b2); we can see that when applying −15 V with a current of 63 uA, bright white dots started to appear, at once when voltage increase bright white dots were more intense, the EL spectrum intensity of the bright dots increased to the maximum emission with −25 V and 80 uA. This event was caused by holes that were attracted to the silicon surface, creating an accumulation layer, and the holes from this layer were injected toward the ITO/SRO and electrons from the ITO gate to the SRO/Si substrate interfaces. However, since the major contribution of current came from the tunneling of electrons instead of holes for MOS on p-type Si, the meeting point was closer to the SRO/Si interface, and then the recombination happened, both in the SRO film and the Si substrate surface, as reported in other works [
34,
44,
45,
46,
47,
48]. The latter gave rise to the EL emission spectra of the T-A S1
100 structure showing three prominent emission peaks at around 450, 530, and 640 nm being the more intense one at 450 nm. Besides, the peaks at 530 nm of the four spectra showed a slight blue-shift. However, the peaks at 640 nm lay in this band. The two bands gave rise to intense white EL at high injection currents. The image inserted in
Figure 9(b3) depicts dispersed bright white dots at −15 V with a low current of 63 uA, and such dots maintained a greater intensity when current is increased at 600 uA, which are presented as a white light spectrum, contributing to the blue, green and orange bands, such colored emissions are attributed to a competition between defects and Si-ncs, as should be expected due to the mixed material; however, as has been previously reported, the emission was mainly due to defects, especially NOV and NBOHCS ones [
10,
24].
Regarding the best T-A S1
25/100 MIS-structure shown in
Figure 9(c1,c2,c3), it presented a blue color full-area emission when applying −55 V at 108 uA, showing an EL emission located at the 460 nm band blue whose intensity was greater than 30,000 a.u.
Figure 11 shows the progress of how the blue spectrum was emitting, starting with dots and then filling the entire area of the structure. Such an emission originated by the radiative emissions from the weak oxygen bonds (WOBs) and neutral oxygen vacancies centers (NOVs). The key factors which contributed to this emission were thermal annealing, presence of Si-related defects, D-L structure, high voltage bias with low currents, and the radiative recombination in localized states related to Si-O bonds. On the other hand, in accordance with [
38,
39,
40,
41,
42,
43], blue light emission was attributed to defects associated with excess silicon, which was correlated with the increase of the refractive index of 1.93 of this SRO
25/100 films, in the same sense [
47] reports that full area emission obtained is due to the optimization of carrier injection through the material by the annulation of preferential conductive paths. That is to say, [
34,
47] it could be related to the Si-nps density that when going to a uniform network of conductive paths, this charge uniformly flowed through the whole structure area.
Finally, the T-A S1
100/25 MIS-structure presents an outstanding lateral emission focused mainly on its borders structure, as observed in
Figure 9(d3). This phenomenon was provoked due to the establishment of conductive pathways that the Si nanocrystals generate and that allow conduction through the dielectric matrix, producing EL emission in spatial regions belonging to the thinnest layer of the film (ends). Regarding EL intensity spectra,
Figure 9(d2), each broad peak was attributed to defects and Si-nps. It has also been reported that amorphous Si-nps required lower voltages but higher currents to achieve the same EL intensity as their crystalline form [
56].
In
Figure 9(d2), the wide-band emission of the structure is shown, spanning from 450 nm to 1000 nm. Evidently, multiple carrier recombination channels contributed to this emission spectrum. Therefore, to be sure of the emission mechanisms of the device, a deconvolution is plotted in
Figure 12 to fit the peaks of the EL spectra, which according to [
23] were attributed to (NBOHC) E’ ≡Si-O-O≡Si+ centers and non-bonded oxygen hole centers at the wavelengths 617 nm and 685 nm while localized luminescent centers (LLC) at the interface of nc-Si with SiO
2 were the emission mechanisms observed in the peaks fitted for the wavelengths of 825 and 890 nm.
In this device T-A S1
100/25, the emission was on the edge of the electrode due to the electrode having a high resistance for conduction; this did not permit the emission on the surface of the electrode as it is shown in
Figure 9(d3). The conduction was easier by the electrode edge, which produced the radiative recombination and emission. The mechanism responsible for the surface-electroluminescence at the edge was related to the recombination of electron-hole pairs injected through enhanced current paths within the silicon-rich oxide film [
51].
A photograph of the only T-A S2
25 MIM-structure presented some needles-and-points of EL emission. This was reverse polarized at two different voltages, as shown in
Figure 13, along with its EL response. As can be seen in
Figure 13c, the central area of the structure had some bright lines and dots with increasing voltage. This can be attributed to the formation of a small number of preferential conductive pathways within the SRO film, which connected the upper electrode to the lower one. In this structure, the conduction through the active layer was not uniform but rather through discrete pathways, causing light emission to be observed only at the points corresponding to the places where conduction occurred. Additionally, there is a report [
52] that oxygen-related defects rather than silicon nanocrystals are present in their SRO films with low excess silicon, the reason for which it is not possible to conform the EL emission. On the other hand, the EL spectra of the SRO film-based MIM-structure were very low intensity remaining at the wavelength between 600–700 nm as the voltage increased, shown in
Figure 13b. According to [
23], these are attributed to (NBOHC) E’ ≡Si-O-O≡Si+ centers and non-bonded oxygen hole centers. This behavior was similar for all values of the T-A SRO films.
Therefore, this MIM S2 structure showed no high EL activity, exhibiting only a few bright dots’ flashes, compared to the amount of more continuous dots in MIS S1 structures.
On the other hand, to obtain the emission efficiencies in the best structures, it was necessary to know the current density (J
d) and the current density in emission (J
e). J
d was obtained through the I-V curves shown in
Figure 5a,c, for structures S1
25 and S1
25/100, respectively, while J
e was obtained from the I-V curves shown in
Figure 9(a1,c1), for structures S1
25 and S1
25/100, which present electroluminescence in points and complete area, respectively. The samples S1
25 and S1
25/100 have the current emission densities of 317 mA/cm
2 y 19.8 A/cm
2, respectively. The efficiencies were obtained with the equation
[
57]. Therefore, at −50 V, their corresponding efficiencies were 3.2% and 19.7%, so the S
25 and S1
25/100 respectively. So, the S
25/100 structure had the highest emission current density and efficiency among these structures. This result may be related to the thinnest thicknesses of the SRO films in the S1
25/100 structure, which caused the free electrons in the SRO film to suffer less scattering during transport [
57]. However, the SRO film should also not be too thin; otherwise, the film could be easily broken.