1. Introduction

Despite the great advances provided by silicon science to the micro-electronics industry, it is clear that future progress in the field relies on the development of (new) materials and/or methods with improved performance. This is particularly true in the current “Digital World” that requires increasingly fast and efficient modes of generation-processing-transport-storage of information. A very convenient approach to comply with these needs is the so-called Si photonics, which consists in attributing new optical functionalities to the well-established Si science and technology [1]. Within this context, the Si−rare-earth (RE) system occupies a privileged position since it is expected to combine the superior electronic characteristics of Si-based materials with the unique optical properties of RE ions.

In most of the cases (involving either solid or liquid hosts, crystalline or amorphous atomic structures), RE ions are present under the trivalent RE³⁺ form [2]. Depending on the preparation method and/or the host details, divalent RE²⁺ ions can also take place − Eu²⁺, Yb²⁺, and Sm²⁺ being the most representative ones. In fact, the Sm²⁺-doped CaF₂ system [3] provided one of the first solid-state lasers (at that time called optical masers) − a few months after the seminal work made by T. H. Maiman involving Cr³⁺-doped corundum crystal, or the ruby laser [4]. Albeit the fact that, at that time, the Sm²⁺-related light emission was achieved only at 20 K, the work influenced many other studies rendering room-temperature Sm³⁺-related visible laser radiation [5], as well as interesting hole-burning experiments [6,7], for example.

Motivated by some of these facts, this work contains a preliminary study of the SiO_x−Sm system in which both Sm³⁺ and Sm²⁺ ions coexist. As far as the present author is aware this is the first report on the subject that, additionally, discusses the possible reasons behind the development of Sm³⁺/Sm²⁺ ions into the SiO_x matrix, as well as their most likely photon excitation-recombination mechanisms. According to the study, that was based on simple production and characterization methods, the SiO_x−Sm system represents a very convenient alternative of visible light-emitting material to the ever increasing Si photonics industry.

2. Experimental details

Much of the experimental work involving the preparation and characterization of the present Sm-doped SiO_x film can be found elsewhere [8]. For completeness, however, the most important figures are reproduced here. A 500 nm thick Sm-doped SiO_x film was prepared by sputtering a 5 inch solid Si target in an atmosphere of pure argon, and the insertion of Sm and oxygen was achieved by partially covering the Si target with an area-defined (706 mm²) thin layer of Sm₂O₃ powder. The film was investigated by optical transmission, Raman scattering, and photoluminescence measurements - both in the as-deposited form and after thermal treatments at 250, 500, 750, and 1000 °C. The treatments were cumulative (30 min each) and were carried out under a continuous flow of oxygen. According to the compositional analysis (as provided by energy-dispersive x-ray spectrometry) the as-deposited film presented 0.6 at.% of samarium and 13.8 at.% of oxygen. After thermal anneal at 1000 °C, the oxygen content increased to 50.1 at.%, whereas the Sm concentration remained almost unchanged [8].

3. Results and discussion

Figure 1(a) shows the effect of thermal annealing treatments on the optical transmission spectra of the present Sm-doped SiO_x film. According to the spectra, the most important change occurred after treatment at 1000 °C, in which case the optical bandgap reached ~2 eV. This value was obtained following the model proposed by Tauc et. al [9], by means of a (αE)^1/2 versus E plot [Fig. 1(b)], where α and E stand for the absorption coefficient and the photon energy, respectively. Allied to the increased oxygen concentration (from 13.8 at.% in the as-deposited film to 50.1 at.%), with the replacement of Si−Si by Si−O bonds and the consequent recession of the top of the valence band states [10], part of the changes verified in the optical properties of the SiO_x film is also related to certain structural transformation.

 

Fig. 1 (a) Optical transmission spectra of a Sm-doped SiO_x film, as-deposited, and after thermal annealing at 250, 500, 750, and 1000 °C (30 min each) under a flow of oxygen. The spectra were obtained from films deposited on fused silica, which spectrum is also shown. The fringes in the spectra appear due to photon interference effects at the film-substrate interfaces. (b) Tauc's optical bandgap as determined from the (αE)^1/2 versus E representation of some Sm-doped SiO_x films: as-deposited, and after annealing at 750 and 1000 °C.

Download Full Size | PDF

In fact, it is interesting to observe a small bandgap shrinkage (Fig. 1) after treatment at 750 °C − right before the crystallization of the SiO_x film at 1000 °C. The aspects leading to this effect are present in Fig. 2 that shows the Raman results of the SiO_x film treated at increasing temperatures. According to the spectra of Fig. 2(a), after treatment at 1000 °C the crystallization was not complete and the film can be described by a collection of silicon crystallites embedded into an amorphous Si + SiO_x medium. Moreover, a closer inspection of the Raman spectra indicate important details concerning the atomic structure of the film as the thermal treatment advanced. A good measure of the structural disorder present in amorphous Si-based films can be achieved from the scattering intensities due to the transverse-acoustic TA and transverse-optical TO phonon modes at ~140 and 470 cm⁻¹, respectively [11].

 

Fig. 2 (a) Raman scattering spectra (632.8 nm photon excitation under a power density of ~350 μW μm⁻²) of Sm-doped a-SiO_x films, as-deposited, and after thermal annealing at increasing temperatures. The spectrum of a crystalline Si sample (commercial Si wafer) is also shown for comparison. All spectra were normalized for comparison purposes. (b) Raman scattering intensity ratio involving the TA and TO phonon modes, characteristic of amorphous silicon, as a function of the annealing temperature. The I_TA/I_TO ratio refers to Sm-doped and undoped SiO_x films (see inset) and is proportional to the atom disorder present in the samples.

Download Full Size | PDF

Indeed, the I_TA/I_TO ratio had been applied to the study of different materials [12] and processes [13] with great success along the years. Since the I_TA/I_TO ratio scales with the atomic disorder, the experimental data of Fig. 2(b) indicate that some structural improvement took place in the film treated up to 750 °C, whereas further anneal at 1000 °C had the opposite effect. In the first case, the energy supplied by the thermal treatments induced the redistribution and rearrangement of atoms and chemical bonds. In the latter, additional energy increased the number of Si−O bonds and promoted the formation of Si crystallites.

As a result, the stress provoked by the Si crystallites embedded into the SiO_x matrix led to the observed augment of the structural disorder. A similar behavior was verified in one SiO_x film containing no samarium, prepared and thermally annealed following the same conditions [Fig. 2(b) and inset]. Comparing the results of these two films, and consistent with the presence of samarium species, the I_TA/I_TO ratios exhibited by the Sm-doped SiO_x film are slightly higher.

In part because of its small optical bandgap and/or due to a high density of defects, no photoluminescence PL was observed in the Sm-doped SiO_x film as-deposited or annealed up to 750 °C. After treatment at 1000 °C, on the contrary, the film starts exhibiting PL signal as indicated in Fig. 3. The PL spectra were obtained, at 10 and 300 K, by exciting the film with different photon wavelengths provided either by argon ion (514.5, 496.5, 488.0, and 476.5 nm) or diode lasers (532 ± 1, 450 ± 5, and 405 ± 5 nm). All measurements were made in the same experimental setup (involving a Si photon avalanche detector coupled to a 0.5 m monochromator) by exciting the film with 15 mW laser light (~300 μm spot diameter). According to the spectra of Fig. 3 it is clear that both the photon excitation wavelength and the temperature of measurement influenced the PL signal of the Sm-doped SiO_x film. Moreover, the spectra are constituted by a series of rather sharp PL lines due to Sm³⁺ (transitions A, B, C, and D) and Sm²⁺ (transitions E and F) ions. Considering the absence of spectroscopic information regarding the SiO_x−Sm system, the identification of transitions A−F was made according to the classical work by Dieke [14], amongst others [15], and is shown in Table 1.

 

Fig. 3 (a) PL spectra of a Sm-doped SiO_x film (after annealing at 1000 °C) as obtained from different photon excitation wavelengths (λ_exc). The spectra were achieved at 10 K, normalized (see the multiplying factors), and vertically shifted for comparison reasons. (b) Same as in (a), except for the temperature of measurement (300 K). Labels A, B, C, and D (E and F) denote transitions due to Sm³⁺ (Sm²⁺) ions. The stars indicate features either due to the Raman signal of crystalline silicon [first and second order of transverse-optical modes at ~547 nm (or 520 cm⁻¹) and at ~561 nm (or 980 cm⁻¹), respectively], or associated with a laser artifact (at ~592 nm).

Download Full Size | PDF

Table 1. Main visible and near-infrared optical transitions exhibted by the present Sm-doped SiO_x film. The transitions are due to Sm³⁺ or Sm²⁺ ions and, as indicated in Figs. 3 and 4, they were identified by letters A−F. The most intense PL signals appear in bold and those involving Sm²⁺ ions were denoted in Italics. PL features evident only at 300 K were indicated between brackets. In addition to A−F, other very weak transitions could be detected and were associated with the Sm²⁺ ions: at ~766 nm (⁵D₀ → ⁷F₃) and ~814 nm (⁵D₀ → ⁷F₄).

View Table

In addition to the Sm-related transitions, the ~530−650 nm range also presented light emission due to the SiO_x matrix. According to the literature, such a broad signal is attributed to oxygen-related defects and/or band(tail)-to-band(tail) transitions [16], which is consistent with SiO_x films containing ~50 at.% of oxygen and optical bandgaps around 2 eV [17]. Also, both Raman and PL results are incompatible with the presence of Si crystallites in the SiO_x film with the sizes typically required to observe quantum confinement effects [18,19].

The PL spectra of a Sm-free SiO_x film prepared (without Sm₂O₃ powder), annealed, and measured under exactly the same conditions are shown in Fig. 4. The measurements were carried out at 10 and 300 K and the spectra of the Sm-doped SiO_x film were also presented for comparison. Except for minor differences in the PL intensity I_PL, it is worth noting that the PL spectra of the Sm-doped film is a mixture of contributions due to the SiO_x matrix and the Sm³⁺/Sm²⁺ ions − their relative intensities being susceptible to the energy transfer mechanisms. It is also clear from Figs. 3 and 4 the effect of the temperature of measurement in decreasing the overall PL intensity at 300 K, as a result of non-radiative recombination processes [22]. At this point it is opportune to mention that, despite the unintentional influence of the experimental setup (i.e., longpass color filter + diffraction grating + optical detector) onto the real shape of the spectra associated with the SiO_x matrix, it had no consequences on the present discussion and conclusions: neither on the PL spectral assignment (and PL intensities I_PL), nor on the energy transfer mechanisms. In fact, PL measurements (following excitation with 405 nm photons) were performed by using different longpass color filters (cut-off at 455 ± 6 and at 550 ± 6 nm) rendering practically the same spectra.

 

Fig. 4 Photoluminescence spectra of a Sm-free SiO_x film (after annealing at 1000 °C), as obtained with 488.0 nm photon excitation at: (a) 10 K, and (b) 300 K. The spectra of the Sm-doped SiO_x film (annealed at 1000 °C), acquired following exactly the same experimental conditions, are also shown for comparison. The labels denote optical transitions due to Sm³⁺ (A, B, C, and D) and Sm²⁺ (E and F) ions. Despite the use of arbitrary units, the PL intensity of all spectra can be compared.

Download Full Size | PDF

Amongst all the PL results the most remarkable one refers to the coexistence of Sm³⁺ and Sm²⁺ ions into the SiO_x matrix. Based on the literature, Sm²⁺-related light emission had never been observed in SiO_x [23–27], although it is commonly verified in many other hosts. In this last case, however, the Sm²⁺-related emission only appears after exposing the samples to different activation or reduction methods induced by energetic sources such as, for example: γ-rays [28], x-rays [29], β irradiation [30], and fs-laser radiation [31].

This is not what happened with the present SiO_x film, in which the Sm²⁺ ions were permanent and already existent without the need of activation or reduction methods. On the other hand, the origin of the Sm²⁺ ions seems to be similar, i.e., dependent on the structural-electronic characteristics of the host, and is tentatively ascribed to the presence of charged defects in the SiO_x matrix.

One of the most common defects present in amorphous silicon samples and at the surface of silicon crystallites surrounded by amorphous SiO_x (or SiO₂) is the silicon dangling bond DB. Silicon dangling bonds are coordination defects giving rise to electronic states around midgap that, additionally, exhibit an amphoteric behavior. More specifically, Si DBs can exist under three different states: positively charged DB⁺ (when unoccupied), neutral DB⁰ (when occupied by one electron), and negatively charged DB⁻ (when occupied by two electrons). Whereas thermal annealing treatments and oxygen incorporation can drastically reduce the number of these defects, their energy and charge characteristics depend very much upon several experimental details like preparation method, sample thickness, annealing temperature and duration, etc [32]. Considering the off-stoichiometric and amorphous nature of the SiO_x film, oxygen-related defects (particularly the E' center, the non-bridging oxygen hole center NBOHC, and the peroxy radical POR [33]) are present and should influence the valence of the samarium ions. According to this reasoning, silicon- and/or oxygen-related charged defects are at the origin of the electronic charges needed to the development of Sm³⁺ and Sm²⁺ ions into the SiO_x matrix. Moreover, it is clear that the Sm³⁺/Sm²⁺ ions presented luminescence after the suppression of the non-radiative transitions − represented not only by coordination defects but by tail states as well.

Allied to the influence of charged defects it is also possible that the chemical environment provided by the oxygen species present in the SiO_x matrix, per se, favored the coexistence of Sm³⁺/Sm²⁺ ions. Such statement is based on the studies of several Sm-doped amorphous SiN films prepared and characterized following an experimental approach similar to those being reported here [34,22]. In addition to the comparatively low (or no) oxygen content [35], the lack of any clear Sm²⁺-related light emission in the SiN films can also be originated from: (a) the SiN optical bandgap value and disposition of electronic states, that are incompatible with the efficient excitation-recombination of the Sm²⁺ ions [22], (b) the absence of Si crystallites and related (charged) defects, and (c) different chemical bonding and atomic structure [36].

Still related to the photoluminescence results, it is evident from the multiplying factors of each spectrum of Fig. 3 that the intensity of the PL transitions was affected not only by the temperature of measurement, but by the wavelength of the exciting photons too. A quantitative analysis of this behavior is presented in Fig. 5, at 10 and 300 K, for transitions involving the SiO_x matrix (at ~580 nm), and the Sm³⁺ (at ~649 nm) and Sm²⁺ (at ~685 nm) ions. Based on the data of Fig. 5 it is possible to state that: (a) considering the PL intensity of all transitions, and the rather small amount (~0.6 at.%) of Sm species into the SiO_x matrix, the experimental results corroborate the idea that RE ions are highly efficient radiative recombination centers [37]; (b) the maximum emission from the SiO_x matrix took place under excitation with photon energies compatible with its band edges − in which case both the non-radiative losses and energy transfer to the Sm ions are reduced [22]; and (c) the PL intensities of the Sm³⁺- and Sm²⁺-related transitions are very particular, and seem to be complementary in a clear indication to the existence of quasi-resonant energy transfer mechanisms [38,39].

 

Fig. 5 Photoluminescence intensity due to Sm³⁺ (⁴G_5/2 → ⁶H_9/2, transition C at 649 nm) and Sm²⁺ (⁵D₀ → ⁷F₀,⁷F₁ transition E at 685 nm) ions as a function of the excitation wavelength, at: (a) 10 K, and (b) 300 K. The PL intensity associated with the SiO_x matrix (broad signal at ~580 nm) is also shown. Despite the use of arbitrary units, all PL intensity values can be compared. The lines joining the data points are just guides to the eye.

Download Full Size | PDF

Due to the lack of PL-excitation measurements of the present Sm-doped films, as well as in the absence of any spectroscopic investigation of the SiO_x−Sm system, the data of other Sm-containing materials had been considered to analyze the I_PL results of Fig. 5. Therefore, Fig. 6 contains the I_PL values of Fig. 5 (at 300 K) along with: the energy levels of Sm³⁺ ions in LaCl₃ [14], and the bands due to Sm²⁺ ions when embedded in CaF₂ [40]. The figure also displays the PL-excitation spectra of SrB₄O₇ samples containing either Sm³⁺ or Sm²⁺ ions [41]. Notwithstanding the obvious compositional-structural differences involving the SiO_x and LaCl₃, CaF₂, and SrB₄O₇ hosts, the correspondence between all data of Fig. 6 is remarkable.

 

Fig. 6 Photoluminescence intensity due to (a) Sm³⁺ (at 649 nm), and (b) Sm²⁺ (at 685 nm) ions. The data were obtained from a Sm-doped SiO_x film (annealed at 1000 °C) following different photon excitation wavelengths. The photoluminescence excitation spectra of a Sm-doped SrB₄O₇ sample [41], and the energy levels (or bands) associated with Sm³⁺ (or Sm²⁺) ions in LaCl₃ (or CaF₂) are also shown for comparison [14] ([40]). All experimental data were obtained at 300 K.

Download Full Size | PDF

The whole set of experimental data − PL-excitation spectra, energy levels and bands, and present I_PL results − shows a rather good association, indicating that most of the I_PL features of Figs. 5 and 6 are, indeed, related to the quasi-resonant photon excitation of Sm³⁺ and Sm²⁺ ions. According to this reasoning, the most intense Sm³⁺-related PL signals were achieved following photon excitation at: (a) ~405 nm (that is coincident with the ⁴F_9/2, ⁴L_13/2, and ⁴P_3/2 energy levels), (b) ~476 nm (⁴I_11/2 level), and (c) ~488 nm (⁴I_9/2 level). Analogously, for the Sm²⁺ ions, the strongest PL emissions were observed after photon excitation with ~510−535 nm, in agreement with the energy region in which the density of states of the 4f⁵ 5d¹ bands is higher [40].

The efficient PL emission of the Sm³⁺/Sm²⁺ ions by quasi-resonant photon excitation is not the only possible one, and a few comments are in order: (1) the I_PL values shown in Figs. 5 and 6 are representative of the optical transitions due to Sm³⁺/Sm²⁺ ions and SiO_x matrix, and were chosen to minimize any cross interference between the data; (2) in spite of the presence of Si crystallites in the Sm-doped SiO_x film (Fig. 2), they exerted no clear influence on the photoluminescence of the Sm³⁺/Sm²⁺ ions; (3) in fact, as far as the thermal treatment is concerned, its most important effect was the development of Si−O bonds that considerably improved the PL signal − either reducing the number of non-radiative transitions or producing a more appropriate chemical environment to the Sm³⁺/Sm²⁺ ions; and (4) the present phenomenological explanation to the PL excitation-recombination processes is not free from amendments but, in its present form, matches very well the experimental data.

A diagram illustrating the main energy levels and bands of the Sm³⁺ and Sm²⁺ ions and the SiO_x matrix is presented in Fig. 7. Following the available literature in the field, the diagram was made on scale to provide a realistic picture of the excitation−recombination processes taking place in the Sm-doped SiO_x film.

 

Fig. 7 Diagram of the main PL transitions and energy levels (or bands) associated with the Sm³⁺ and Sm²⁺ ions (and SiO_x matrix). The A−F labels refer to the PL transitions identified in Table 1. The photon excitation used in the present study, as provided by different laser sources, is also depicted.

Download Full Size | PDF

4. Concluding remarks

Along the years, Si-based materials and methods were essential to the development of the modern micro-electronics industry. Combined to the unique properties of some rare-earth elements, they are expected to provide further technological advancements that will impact the field of telecommunications (in the NIR range) or flat panel displays (in the VIS range), for example. The production and spectroscopic study of the SiO_x−Sm system, as the one being reported here, aim to contribute with useful ideas-information to the subject. Therefore, a Sm-doped SiO_x was prepared and investigated following standard methods and conditions like sputter deposition, thermal annealing, and optical spectroscopy techniques.

The ensemble of experimental work indicated that: (a) argon ion sputtering a Si + Sm₂O₃ target is a suitable approach to achieve Sm-doped SiO_x films [8]; (b) whereas the content of samarium and oxygen depend on the target details, additional oxygen was incorporated into the film by means of thermal treatments under a flow of oxygen; (c) along with changes in the chemical composition of the film, thermal annealing at higher temperatures (> 750 °C) induced the development of silicon crystallites; (d) both oxygen enhancement and structural modification widened the optical bandgap of the SiO_x film and suppressed the number of non-radiative transitions; and (e) rather intense photoluminescence had been observed in the visible and near-infrared ranges, only after thermal annealing the film at 1000 °C.

Based on the analysis of the photoluminescence spectra it was possible to conclude that: (1) they are a combination of contributions due to the SiO_x matrix (broad signal in the ~530−650 nm range), Sm³⁺ (at ~562, 602, 649, and 709 nm), and Sm²⁺ (at ~685 and 730 nm) ions; (2) the overall PL intensity was improved at low temperature (10 K) because of the suppression of non-radiative recombinations; and (3) the Sm³⁺ and Sm²⁺ ions were photon excited, predominantly, by (quasi-)resonant processes. The coexistence of (spontaneous and permanent) Sm³⁺ and Sm²⁺ ions seems to be completely original, and was tentatively ascribed to the presence of charged defects into the SiO_x matrix.

Whereas it is common practice to estimate the number of Sm³⁺/Sm²⁺ ions based on their relative photoluminescence intensity [42], the present study indicates that such procedure is dependent on the temperature and photon excitation conditions. Indeed, to determine the precise atomic environment and the relative amount of the Sm³⁺/Sm²⁺ ions still represent considerable challenges and should be the subject of future research. Therefore, within the next steps to continue the present study one can cite the elucidation of the mechanisms behind the presence of Sm³⁺ and Sm²⁺ ions into the SiO_x matrix, the real number of Sm³⁺/Sm²⁺ ions, as well as their chemical environment. In view of them: the development of strategies to achieve SiO_x films with only Sm³⁺ or Sm²⁺ ions, and the search for hole burning effects in the SiO_x−Sm system.