Open Access
Add to BibSonomyAbstract
Laterally electrically-pumped Si light-emitting diodes (LEDs) based on truncated nanocrystalline-Si (nc-Si)/SiO2 quantum wells are fabricated with complementary-metal-semiconductor-oxide (CMOS) process. Visible electroluminescence (EL) can be observed under a reverse bias larger than ~6 V. The light emission would probably originate from the spontaneous hot-carrier relaxations within the conduction and the valance bands when the device is sufficiently reverse-biased. The EL spectral profile is found to be modulated by varying structure parameters of the interdigitated finger electrodes. Up to ~20 times EL intensity enhancement is achieved as compared to vertical-current-injection LED prepared using the same material system. Based on the lateral-current-injection scheme, a Si/SiO2 MQW LED with Fabry-Perot (FP) microcavity and an on-chip waveguided LED that emits at 1.55-µm are proposed.
©2011 Optical Society of America
1. Introduction
The successful fabrication of complementary-metal-semiconductor-oxide (CMOS)-compatible high-efficiency light-emitting diodes (LEDs) will surmount the last obstacle for Si photonics. Intensive research effort [1–12] has been devoted to the realization of low-voltage electrically-pumped Si-based LEDs since the first observation of efficient luminescence from porous Si due to the quantum confinement effect of nano-scaled dimension of the material [1–9]. Among all the Si nano-structures, multiple quantum wells (MQWs) consisting of alternating ultra-thin layers of nanocrystalline-Si (nc-Si) and dielectric films has received intensive research attention in aspects of both photoluminescence (PL) and electroluminescence (EL) [5–9], since the initial proposal of Si/SiO2 superlattice as the Si-based quantum well light source [10]. However, it is a disadvantage that efficient carrier injection is difficult to be achieved in the conventional vertical-current-injection nc-Si/SiO2 MQW LEDs, because the current transport is limited by the highly-insulating SiO2 layers. However, MQW lasers based on III-V compounds consist of alternating small and large bandgap materials which are both semi-conductive. Therefore, the current-injection is a unique issue for the Si/dielectric MQW devices in which the large bandgap material is highly-insulating. A further disadvantage of conventional structures employing vertical-current-injection scheme is the large difference of carrier densities in the individual quantum well. In a conventional MQW device, electrons and holes are injected into the wells in the direction vertical to the well surfaces. The mobility of electrons is much higher than that of holes, and thus electrons move faster than holes under a voltage bias. Therefore, the carrier density is higher in quantum wells near the p-electrode than in those near n-electrode under a voltage bias. Accordingly, the well numbers cannot be increased arbitrarily due to large difference of carrier intensities in each individual quantum well. This problem has been well illustrated and studied by Tessler [11]. The inefficient current-injection as well as the limited number of quantum wells prevents the vertical Si MQW light source from achieving a high gain of power. Although there have been many investigations on the LEDs based on vertical-current-injection nc-Si/dielectric superlattice, very few researchers investigated on the nc-Si MQW LED by injecting carriers parallel to the surface of multilayer films. In the lateral-current-injection MQWs, the carriers can be more efficiently injected into all well layers without being affected by two-dimensional potential barriers. In addition, there is no lowering in the injection efficiency even if the thickness of large-bandgap well is increased or the number of multi-quantum well layers is increased.
This paper demonstrates a MQW-based Si LED based on laterally injecting current directly into the nanocrystalline Si (nc-Si) films without carrier tunneling across SiO2. With the lateral current injection structure, the effective current injection efficiency is largely increased as the applied voltage is fully applied across the active region instead of dielectrics which cannot be avoided in vertical-current-injection nc-Si/SiO2 MQW LEDs. This leads to a significant enhancement of EL intensity. Since there is no top electrode on the active region, further enhancement can be achieved by light-extraction enhancement techniques directly applied on the active region. Moreover, by advantageously avoiding the top electrode, this lateral configuration allows for the easy fabrication of a new type of CMOS-compatible, low-loss, waveguide-based light source.
2. Device fabrication
Ten periods of alternating α-Si and SiO2 layers were deposited as the active region on top of a layer of 500-nm-thick thermal SiO2 on Si substrate. Ultra-thin α-Si films were deposited by plasma-enhanced chemical vapor deposition (PECVD) using SiH4 as the source gas diluted by He. The thickness of α-Si is ranging from 2 to 10 nm, while the SiO2 thickness was kept at 3 nm. Afterwards, the films were annealed in N2 ambient for 1 hour at high temperatures to induce crystallization of ultra-thin α-Si films. To choose the annealing temperature, the thermal annealing was carried out for a duration of 1 hour at the temperatures from 600 to 1100 °C. Micro-Raman spectroscopy and Transmission electron microscopy (TEM) and have been employed to study the structure properties of the materials. The micro-Raman measurements were performed at 325 nm. For the samples that were used to perform Raman measurements, we initially deposited a layer of 2-µm SiO2 beneath the Si/SiO2 multilayer to isolate the Raman signal from Si substrate. Figure 1 shows the evolution of the Raman spectra for as-deposited samples and the samples with high temperature annealing for duration of 1 hour at the temperature of up to 1100 °C. It can be seen in Fig. 1 that the Raman speak around 520 cm−1 which is identified to crystalline Si is increased with the annealing temperature. One can conclude from the figure that the crystalline content of Si is increased with the annealing temperature when it is higher than 700 °C, which is in agreement with study on nanocrystallization of α-Si/SiO2 superlattices by high temperature thermal annealing [12]. In this study, we choose 1100-°C as the annealing temperature to induce nanocrystallization, because 1100-°C annealing can induce the most crystalline state concluded from the Raman spectrum evolution with annealing temperature. As shown in Fig. 2 , TEM images also confirm the presence of the crystalline state in Si thin layer even for 3-nm Si layers. However, the fully crystalline state of thin Si layers is difficult in thinner (<5 nm) Si layers due to the extremely higher thermal budget required (i.e., >1200 °C) as indicated in Ref [12]. Therefore, we would claim that the ultra-thin Si layer is in a mix phase of nanocrystalline and amorphous state after annealing at 1100 °C especially for thinner (<5 nm) Si layers, and this is supported by the TEM images shown in Fig. 2. For a clear and simple description, it should be point out that in this paper the thin Si film is denoted as α-Si before annealing and nc-Si after annealing although it is not rigorously correct in terms of the film structural state.
Fig. 1 Raman spectra of as-deposited sample (i.e., without annealing) and the samples annealed for 1 hour at the temperature of 700, 900, 1000, and 1100 °C.
Fig. 2 High resolution transmission electron microscopic (HR-TEM) images of α-Si/SiO2 multilayers (a) 10-nm-Si/ 3-nm-SiO2 and (b) 3-nm-Si/3-nm-SiO2.
Based on the film system fabricated above, two types of LED structures, i.e., line emission (i.e., Type-I) and enhanced surface emission structure with interdigitated finger electrode (i.e., Type-II), were fabricated based on truncated nc-Si/SiO2 MQW with lateral-current-injection scheme. Three-dimensional (3D) schematics of the two types of structures are shown in Fig. 3(a) and 3(b), respectively. For a clear demonstration, EL photos at room temperature under electrical pumping for both of the two types LED devices are included in the respective figures.
Fig. 3 (a) 3D schematic and EL photo of Type-I device with line emission. (b) 3D schematic and EL photo of Type-II device with interdigitated finger electrode. (c) Simplified process flow showing the key steps of fabricating a lateral-current-injection nc-Si/SiO2 MQW LED with interdigitated finger electrodes. (i) Deposition of nc-Si/SiO2 MQWs. (ii) Active region patterning and etching. (iii) Deposition of poly-Si and implantation to form p + and n + electrode. (iv) Poly-Si patterning and etching to form a grating-like structure on the active region.
After the nc-Si/SiO2 multilayers was deposited and annealed, a layer of 100-nm SiO2 was deposited on top of the multilayer to passivate the active region. The active region was defined by deep UV lithography and reactive ion etching through the multilayer to expose the thermal oxide. For Type-I devices, the active region was patterned into a narrow line of width ranging from 0.5 to 3 µm. As regard to Type-II devices, the active region was patterned into interdigitated finger structures to have a large surface area for emission. Afterwards, a layer of 160-nm poly-Si was deposited by low pressure chemical vapor deposition (LPCVD) followed by high-dose implantation of boron and phosphorous for p + and n + electrode formation, respectively. The emission window was formed by etching the poly-Si on top of the active region. Finally, the fabricated devices were annealed in N2 ambient for 1 hour at 1000 °C for dopant activation.
Taking the Type-II device as an example, the fabrication process flow was summarized in Fig. 3(c). In this study, the finger width and spacing are denoted by a and w, as shown in Fig. 3(c). After opening the emission window by etching poly-Si on top of the nc-Si/SiO2 multilayer, a grating-like structure consisting of periodic array of poly-Si and air were formed, as shown in Fig. 1(c).
3. Results and analysis
Visible EL emission can be observed by naked eye in daylight under a reverse bias larger than ~6 V, as shown in Fig. 3(a) for Type-I device and 3(b) for Type-II device. The EL spectra were taken by a PDS-1 photomultiplier tube (PMT) detector together with a monochromator. Figure 4(a) shows the EL spectrum for Type-I device with line emission width of 1 µm under a bias of −10 V. It can be seen that it is a broad EL spectrum covering the entire visible spectrum. A linear relationship is observed between the reverse current and integrated EL intensity, as shown in Fig. 4(b). The current-voltage (I-V) curve shows linear characteristics when the device is under forward bias and reverse-bias if the voltage is larger than ~6 V. More detailed study shows that the linear relationship between EL intensity and current is valid over the entire detected EL spectrum.
Fig. 4 (a) EL spectrum under a reverse bias of 10 V for Type-I device with line emission width of 1 µm. (b) I-V characteristics and the integrated EL intensity as a function of applied voltage. The nc-Si film thickness is 10 nm.
As shown in Fig. 3(a), the structure can be regarded as ten p-i-n junctions in a parallel connection, in which i regions are the nc-Si films. As well known, the electric field in i-region of a revised p-i-n junction is much higher than that of a forward-biased p-i-n junction. When the junction field is large enough, hot carriers can be generated in the i-region. There have been intensive research activities over the past two decades on the hot-carrier induced EL under reverse bias [13–18]. Similar EL spectra have been observed in silicon p-n junction under reverse bias. The origin of the emission in Si p-n junctions has been attributed to different mechanisms such as the Bremsstrahlung effect [15, 16], the intra-conduction recombination [16], and the hot carrier recombination [16]. As regarding to our results, the linear relationship between the EL intensity and the reverse current rules out the contribution from direct electron-hole radiative recombination. The EL measurements were also carried out at various temperatures from room temperature to 300 °C, and no obvious change was found. This result suggests that the phonon-assisted recombination should not play an important role. Bremsstrahlung luminescence is also negligible because the dopant concentration in our device is much lower than 5×1020 cm−3, above which the Bremsstrahlung process can take place [16]. Therefore, we are inclined to conclude that visible EL emission is originated from the spontaneous direct hot-carrier relaxations within the conduction (hot electrons) and the valance bands (hot holes). This paper presents a very initial study on hot-carrier EL from Si quantum wells with lateral-current-injection, thus more comprehensive work is required to quantitatively understand the exact mechanism of the hot-carrier-induced light emission from laterally current injected nc-Si/SiO2 MQWs. Regarding the system working based on hot carriers, the device reliability and life time is really a concern. In this study, we have done a continuous operation of 48 hours on both the Type-I and Type-II devices, and the devices did not show any noticeable degradation of the EL intensity with the working time.
The electrical and light emission properties of Type-II device have been also studied. I-V characteristics of Type-II devices shows a linear relationship similar to the electrical behavior of Type-I devices as shown in Fig. 4(b). However, the EL properties of Type–II devices are found to be significantly modified by the interdigitated finger structures. For example, Fig. 3(a) shows the EL spectrum of Type-II devices with the w of 1 µm and a of 2 µm. The EL spectrum of Type-I device is also included in Fig. 3(a) for comparison. It can be seen that overall EL intensity of Type-II device is significantly increased as compared to the Type-I device. The ~10-times intensity enhancement would be due to the much larger emission area formed by the interdigitated finger structures than that of the single line emission of Type I device. Another reason that would probably contributes to the EL intensity increase is that the textured top surface of Type-II device enhances the escape probability for internally generated photons, ideally by a factor of 2n 2, where n is the effective refractive index of the active region [19, 20]. It is necessary to point out that the finger width w cannot be larger than 3 µm, meanwhile the finger spacing a would better be larger than finger width w. Otherwise, a high voltage larger than 15 V is needed to generate hot carriers, which does not meets the CMOS requirement. As can be seen in the micrograph of Type-II device EL emission (left part of Fig. 3(b)), it is noticeable that the EL intensity is stronger in the area near p + region than in the area near n + region. It is another piece of evidence that the EL emission is originated from the hot-carrier relaxation under reverse-bias, because of the much lower impact ionization rate of holes in Si [21].
Other than the EL intensity enhancement, the EL spectral profile of Type-II device has been found to be modulated as compared to the Type-I device. As can be seen from Fig. 5(a) , unlike the broadband emission over the entire visible spectrum without obvious peaks, the EL spectrum of Type-II device with w = 1 um and a = 2 µm presents a main peak at ~550 nm and a small peak at ~800 nm with much lower intensity. The EL spectra are also found to be modulated by varying the interdigitated finger parameters (i.e., width w and spacing a). Figure 5(b) shows the EL spectra of type-II device with w = 1 um and a = 10 um. As shown in the figure, it presents a main EL peak at ~650 nm and a shoulder peak at ~500 nm. The overall intensity is even higher than that of the Type-II device with w = 1 um and a = 2 um shown in Fig. 5(a). This is probably due to the much higher current density in the nc-Si films for the device with larger a, although the number of emission lines becomes smaller.
Fig. 5 (a) EL spectra under a reverse voltage of 10 V for the Type-II device with w = 1 µm and a = 2 µm. (b) EL spectra under a reverse voltage of 10 V for the Type-II device with w = 1 µm and a = 10 µm. (c) Calculated transmission spectra of Type-II device with w = 1 µm and a = 2 µm. (d) Calculated transmission spectra of Type-II device with w = 1 µm and a = 10 µm. The nc-Si film thickness is 10 nm for the device in this figure.
In order to understand the EL spectral peak shift due to the change of finger parameters, a nonlinear equation method based on plane-wave expansion is used to calculate the transmission spectra of the grating-like structure formed by fabricating the interdigitated finger electrodes. In the simulation, Rigorous Coupled Wave Analysis (RCWA) method was used to solve the Maxwell equations. The RCWA method uses the concept of a unit cell to handle both 2D and 3D periodic structures and is specifically tailored for multilayer structures. As shown in Fig. 5(c), a transverse-electric (TE) mode transmission peak at ~550 nm of can be found, which is in consistent with the EL emission peak shown in Fig. 5(a). The transmission peak at ~800 nm also can be found for the TE mode, which would contribute to the small EL emission peak at ~800 nm. The TE mode transmission spectra are more consistent with the EL emission spectra than the transverse-magnetic (TM) mode, indicating that the EL emission from laterally-current-injected nc-Si/SiO2 MQWs is dominated by TE mode. This phenomenon also can be observed from Fig. 5(b) and 5(d), i.e., the TE mode transmission shows a main peak at ~650 nm and a shoulder peak at ~500 nm which are both presented in the EL spectrum. We found that the EL spectrum almost does not change except for some peak intensity adjustment when the finger spacing a is larger than 10 µm, indicating the modulation effect of the grating-like electrode becomes negligible when the finger spacing is much larger than the emission wavelength by one order.
As regard to a quantum well system, researchers are usually interested in the effect of quantum well thickness on the device properties. In this study, the influence of nc-Si thickness on the EL properties has been investigated. As can be seen in Fig. 6 , the nc-Si thickness has strong effect on the spectral shape as well as the EL intensity. The decreased EL intensity with decreasing nc-Si film thickness is because that the lateral current-injection becomes more difficult in thinner nc-Si films. As can be seen in the figure, the EL intensity of 2-nm Si becomes much smaller than other thicknesses. An important factor that may contribute to this observation is that the continuous α-Si/SiO2 thin films undergo a breaking upon thermal annealing, leading to the formation of Si nanocrystals embedded in SiO2 [22], and thus it is much more difficult for current injection. Therefore, it is not surprising that the EL intensity is higher for the device with thicker nc-Si films at the same reverse bias. The reverse-biased-induced EL cannot be ascribed to the direct-band electron-hole recombination due to quantum confinement effect of nc-Si film, but a strong effect due to the nc-Si thickness on the EL spectral shape is observed as shown in Fig. 6. Therefore, one can expect that the nc-Si thickness can affect the generation of hot carriers and hot-carrier-related luminescence states, although the exact origins are still unclear in this study.
Fig. 6 EL spectra of Type-II devices under the reverse bias of 10 V for the devices with different nc-Si film thicknesses. The measured devices have the finger width of 0.5 µm and spacing of 10 µm.
4. Discussion
In order to have a clear picture on the enhancement of EL emission due to lateral-current-injection, vertical-current-injection LEDs were fabricated using the same material system as that of lateral-current-injection LEDs. Figure 7(a) presents the schematic of the conventional vertical-current-injection nc-Si/SiO2 MQW LED which we have fabricated for comparison. These two kinds of structures are designed to have the same emission window size and fabricated with the same material system. Figure 7(b) shows the EL spectra for both of the lateral- and vertical-current-injection LEDs. Under the same applied voltage of 10 V (please note that the vertical-current-injection LED is forward-biased while lateral-current-injection LED is reverse-biased), the lateral-current-injection LED shows ~20 times stronger EL intensity than that of vertical-current-injection LED. Regarding to the power-conversion efficiency of the structure, it has been estimated to be ~2×10−3 by comparing with a commercial LED with similar wavelength, while the vertical-current-injection LED shows a power-conversion efficiency of 1×10−4. Two factors would contribute to the EL enhancement of the Type-II device. One is the effective current injection increase by avoiding current transport in SiO2 layer, and the other is EL extraction enhancement by removing n + poly-Si top electrode which is absorptive in the visible spectrum. Moreover, one can also observe that the EL spectra are different for the two devices, indicating that the lateral-current-induced EL has different mechanisms from that of vertical-current-induced EL. As well known, the vertical-current-induced EL has two main emission mechanisms comprising quantum-confinement effect of nc-Si and defect-related recombination centers [4–7]. Regarding to our vertical-current-injection device, it is easily to ascribe ~820-nm EL peak to the quantum confinement effect and the ~550-nm peak to the defect-related luminescence centers. However, the EL emission from the lateral-current-injection LED is related to the hot-carrier relaxation. In this study, we cannot conclude that the light emission efficiency of lateral-current-injection structures is higher than optimized vertical ones that have been reported by others [23–25] due to the different EL emission mechanisms between lateral- and vertical-current-injection schemes. The EL from vertical-current-injection LED is due to quantum confinement effect and luminescent defect states, while the EL from lateral-current-injection LED is ascribe to the hot-carrier related luminescent states in ultra-thin Si films.
Fig. 7 (a) Cross-sectional schematic the lateral-current-injection LED with the same emission window size as the lateral-current-injection LED illustrated in Fig. 3(b). (b) EL spectra of the lateral- and vertical-current-injection LEDs under the same voltage of 10 V. The nc-Si film thickness is 10 nm.
Indeed, low-voltage operation of Si-based vertical LED devices with nc-Si embedded in SiO2 has been reported [23–25]. However, low-voltage operation and high efficiency is not enough for real LED application. In the previously reported vertical devices, the active region which is made of dielectric films embedded with nc-Si cannot be too thick due to the current injection will be very difficult in thick SiO2. Such devices with low-voltage operation (<5 V) were fabricated employing a MOS-like structure having the nc-Si embedded gate oxide embedded with the thickness less than tens of nanometers, and total emitting optical power cannot be increased too much due to the limitation of active region thickness. This problem can be solved by employing the proposed lateral-current-injection structure because the number of active region layers can be increased without decreasing the current injection.
This work provides a platform with more flexibility to re-design and integration. First, many light extraction techniques, such as Fabry-Perot (FP) cavity, photonic crystal structures, surface plasmons and so on, can be directly applied onto the bottom and top of the active region, since there is no electrode connected with the bottom and top surface of the active region. Of course, micro-cavity or photonic crystal structure can be fabricated in the vertical LED devices if people want to do them [26,27]. However, the top and bottom electrodes which were usually made of heavily doped poly-Si would make additional optical loss due to the scattering and free carrier absorption by reflecting the light in the cavity or photonic crystal structures. One cannot afford such additional loss of optical power especially for Si-based LED devices due to its already confirmed low emission efficiency. The lateral-current-injection structure is more suitable for micro-cavity and photonics crystal design due to the absence of absorptive top and bottom electrodes. For example, FP cavity formed by bottom and top Bragg reflectors can be easily employed with this lateral-current-injection nc-Si MQW LED, as shown in Fig. 8(a) . FP cavity has been reported and easily fabricated based on nc-Si systems for PL studies [28]. An electrical-driven pumped slot-waveguide LED with FP microcavity has been proposed [29], and it needs to fill Er-doped SiO2 into the vertical slot with the width of tens of nanometers. It is difficult to realize with conventional CMOS tools, except using an atomic layer deposition (ALD) system which allows for high aspect-ratio nanostructure deposition and subsequent accurately controlled chemical mechanical polishing (CMP) process. As can be seen in Fig. 8(a), with the lateral current injection scheme, the LED structure with FP cavity can be easily fabricated with PECVD deposition, ultra-violet (UV) lithography, and plasma dry etching. Optical interconnects used for on-chip and chip-to-chip communication requires an efficient infrared waveguided light source. The infrared Si-based light source that emits at ~1.5 µm have been researched and investigated by doping Er doping into Si nanostructures [30,31]. Because this lateral-current-injection scheme excludes the top electrodes on the active region of MQW structure, one can expect the successful fabrication of an electrical-pumping waveguided light sources. Based on the lateral-current-injection nc-Si/SiO2 MQW LEDs presented in this work, we propose a waveguided light sources that can be fabricated in a standard CMOS line and monolithically integrated with electronic Si chips. The cross-sectional schematic of the proposed structure is shown in Fig. 8(b). The cladding material is SiO2 and waveguide core can be made of silicon nitride. The doped Er in the active region is used as medium for 1.55-µm light emission. The index contrast between oxide and nitride provides extra optical confinement besides the index difference between Si3N4 and the superlattice consisting of nc-Si/Si3N4.
Fig. 8 (a) A proposed structure of lateral-current-injection LED based on truncated nc-Si/SiO2 MQWs with F-P cavity formed by bottom and top Bragg reflectors consisting of alternating SiO2 and Si3N4 films. (b) The cross-sectional schematic of a proposed on-chip waveguided LED with 1.55-µm emission based on the lateral-current -injection scheme reported in this study.
5. Summary
In conclusion, a lateral-current-injection LED based on nc-Si/SiO2 quantum wells has been demonstrated. Strong visible EL can be observed when device is reversed biased. The device exhibits a linear relationship between the EL intensity and the reverse current. The EL can be enhanced and modulated using interdigitated finger structures. The emission modulation is well explained using the plane-wave expansion method. As compared to the vertical-current-injection LED fabricated with the same Si/SiO2 multilayers, the lateral-current-injection LED shows a ~20 times stronger EL intensity. Based on the LED structure with lateral-current-injection scheme, FP cavity can be easily applied and fabricated. Moreover, an nc-Si MQW waveguided LED has been proposed. It may open a way to the efficient on-chip light sources for optical interconnects.
References and links
1. L. T. Canham, “Silicon quantum wire array fabrication by electrochemical and chemical dissolution of wafers,” Appl. Phys. Lett. 57(10), 1046–1048 (1990). [CrossRef]
2. V. Svrcek, D. Mariotti, and M. Kondo, “Ambient-stable blue luminescent silicon nanocrystals prepared by nanosecond-pulsed laser ablation in warter,” Opt. Express 17, 521–527 (2010).
3. X. Sun, J. Liu, L. C. Kimerling, and J. Michel, “Room-temperature direct bandgap electroluminesence from Ge-on-Si light-emitting diodes,” Opt. Lett. 34(8), 1198 (2009). [CrossRef] [PubMed]
4. L. Ding, T. P. Chen, M. Yang, J. I. Wong, Z. Cen, Y. Liu, F. Zhu, and A. A. Tseng, “Relationship between current transport and electroluminescence in Si+-implanted SiO2 thin films,” IEEE Trans. Electron. Dev. 56(11), 2785–2791 (2009). [CrossRef]
5. W. K. Tan, M. B. Yu, Q. Chen, J. D. Ye, G. Q. Lo, and D. L. Kwong, “Re light emission from controlled multilayer stack comprising of thin amorphous silicon and silicon nitride layers,” Appl. Phys. Lett. 90(22), 221103 (2007). [CrossRef]
6. W. K. Tan, M. B. Yu, Q. Chen, W. Y. Loh, J. D. Ye, Z. H. Zhang, G. Q. Lo, and D.-L. Kwong, “Thin amorphous Si/Si3N4-based light emitting device prepared with low thermal budget,” IEEE Electron Device Lett. 29(3), 228–231 (2008). [CrossRef]
7. A. Anopchenko, A. Marconi, E. Moser, S. Prezioso, M. Wang, L. Pavesi, G. Pucker, and P. Bellutti, “Low-voltage onset of electroluminescence in nanocrystalline-Si/SiO2 multilayers,” J. Appl. Phys. 106(3), 033104 (2009). [CrossRef]
8. T. Creazzo, B. Redding, E. Marchena, J. Murakowski, and D. W. Prather, “Pulsed pumping of silicon nanocrystal light emitting devices,” Opt. Express 18(11), 10924–10930 (2010). [CrossRef] [PubMed]
9. M. Wang, A. Anopchenko, A. Marconi, E. Moser, S. Prezioso, L. Pavesi, G. Pucker, P. Bellutti, and L. Vanzetti, “Light emitting devices based on nanocrystalline-silicon multilayer structure,” Physica E 41(6), 912–915 (2009). [CrossRef]
10. R. Tsu, “Silicon based quantum wells,” Nature 364(6432), 19 (1993). [CrossRef]
11. N. Tessler, and G. Eisenstein, “Transient carrier response in multiple quantum well lasers,” in Proceeding of 13th IEEE Semicon. Laser Conf. pp. 44–45 (1992).
12. M. Zacharias, J. Bläsing, P. Veit, L. Tsybeskov, K. Hirschman, and P. M. Fauchet, “Thermal crystallization of amorphous Si/SiO2 superlattice,” Appl. Phys. Lett. 74(18), 2614–2616 (1999). [CrossRef]
13. M. Freitag, V. Perebeinos, J. Chen, A. Stein, J. C. Tsang, J. A. Misewich, R. Martel, and P. Avouris, “Hot carrier electroluminescence from a single carbon nanotube,” Nano Lett. 4(6), 1063–1066 (2004). [CrossRef]
14. C. W. Liu, S. T. Chang, W. T. Liu, M.-J. Chen, and C.-F. Lin, “Hot carrier recombination model of visible electroluminescence from metal-oxide-silicon tunneling diodes,” Appl. Phys. Lett. 77(26), 4347–4349 (2000). [CrossRef]
15. A. L. Lacaita, F. Zappa, S. Bigliardi, and M. Manfredi, “On the Bremsstrahlung origin of hot-carrier-induced photons in Silicon device,” IEEE Tran. Electron Device 40(3), 577–582 (1993). [CrossRef]
16. J. Bude, N. Sano, and A. Yoshii, “Hot-carrier luminescence in Si,” Phys. Rev. B Condens. Matter 45(11), 5848–5856 (1992). [CrossRef] [PubMed]
17. B. K. Ridley, “Hot electrons in low-dimensional structures,” Rep. Prog. Phys. 54(2), 169–256 (1991). [CrossRef]
18. H. Aharoni and M. du Plessis, “Low-operating-voltage integrated silicon light-emitting devices,” IEEE J. Quantum Electron. 40(5), 557–563 (2004). [CrossRef]
19. M. A. Green, J. Zhao, A. Wang, P. J. Reece, and M. Gal, “Efficient silicon light-emitting diodes,” Nature 412(6849), 805–808 (2001). [CrossRef] [PubMed]
20. T. Trupke, J. Zhao, A. Wang, R. Corkish, and M. A. Green, “Very efficient light emission from bulk crystalline silicon,” Appl. Phys. Lett. 82(18), 2996–2998 (2003). [CrossRef]
21. S. Sze, Physics of Semiconductor Devices, (Wiley, New York, 1981), pp 45–47.
22. S. Boninelli, F. Iacona, G. Franzo, C. Bongiorno, C. Spinella, and F. Priolo, “Thermal evolution and photoluminescence properties of nanometric Si layers,” Nanotechnology 16(12), 3012–3016 (2005). [CrossRef]
23. S. Fujita and N. Sugiyama, “Visible light-emitting devices with Schottky contacts on an ultrathin amorphous silicon layer containing silicon nanocrystals,” Appl. Phys. Lett. 74(2), 308–310 (1999). [CrossRef]
24. P. Photopoulos and A. G. Nassiopoulou, “Room- and low-temperature voltage tunable electroluminescence from a single layer of silicon quantum dots in between two thin SiO2 layers,” Appl. Phys. Lett. 77(12), 1816–1818 (2000). [CrossRef]
25. S. Prezioso, A. Anopchenko, Z. Gaburro, L. Pavesi, G. Pucker, L. Vanzetti, and P. Bellutti, “Electrical conduction and electroluminescence in nanocrystalline silicon-based light emitting devices,” J. Appl. Phys. 104(6), 063103 (2008). [CrossRef]
26. A. Muscara, M. E. Castagna, S. Leonardi, S. Coffa, L. Caristia, and S. Lorenti, “Design and electro-optical characterization of Si-based resonant cavity light emitting devices at 850 nm,” J. Lumin. 121(2), 293–297 (2006). [CrossRef]
27. C. D. Presti, A. Irrera, G. Franzò, I. Crupi, F. Priolo, F. Iacona, G. Di Stefano, A. Piana, D. Sanfilippo, and P. G. Fallica, “Photonic-crystal silicon nanoclusters light-emitting device,” Appl. Phys. Lett. 88(3), 033501 (2006). [CrossRef]
28. F. Giorgis, “Optical microcavities based on amorphous silicon-nitride Fabry-Perot structures,” Appl. Phys. Lett. 77(4), 522–524 (2000). [CrossRef]
29. C. A. Barrios and M. Lipson, “Electrically driven silicon resonant light emitting device based on slot-waveguide,” Opt. Express 13(25), 10092–10101 (2005). [CrossRef] [PubMed]
30. F. Iacona, D. Pacifici, A. Irrera, M. Miritello, G. Franzò, F. Priolo, D. Sanfilippo, G. Di Stefano, and P. G. Fallica, “Electroluminescence at 1.54 µm in Er-doped Si nanoclusters-based devices,” Appl. Phys. Lett. 81(17), 3242–3244 (2002). [CrossRef]
31. S. Yerci, R. Li, and L. Dal Negro, “Electroluminescence from Er-doped Si-rich silicon nitride light emitting diodes,” Appl. Phys. Lett. 97(8), 081109 (2010). [CrossRef]
