Recent efforts towards a Si-based nanophotonics will be reviewed. It will be shown that Si nanostructures embedded in silica represent an extremely promising route. In particular a detailed correlation between structural and optical properties will be presented. We will demonstrate that, in contrast to what generally believed, properties of films grown by different methods are indeed very different as a result of the agglomeration properties. In particular, while Si-rich oxides deposited by sputtering present an almost full agglomeration of the Si excess atoms and strong luminescence at temperatures of about 1100 °C, films grown by plasma enhanced chemical vapor deposition (PECVD) present agglomeration of only a small fraction of the Si excess and require very high annealing temperatures to maximize luminescence. Amorphous Si nanoclusters may represent an interesting alternative for the monolithic integration of optical functions in Si technology. In particular, amorphous nanoclusters exhibit an intense room temperature electroluminescence (EL) with the advantage to be formed at a temperature remarkably lower than the temperature needed for the formation of Si nanocrystals. In addition, low operating voltages are required. The doping of these structures with rare earth ions will also be discussed. The introduction of Er ions on these structures produces a preferential transfer of the energy from an exciton in the nanocrystal to the rare earth. We will show that the different agglomeration of the excess Si in a SiOx matrix prepared by PECVD or sputtering can strongly affect the Er luminescence. Er-doped Si-nanocluster based light emitting devices operating at 1.54 microns will be demonstrated. We will show that the electrical and optical properties of the devices strongly depend on the technique used to grow the active layer. In particular, efficient devices based on sputtered films operate at much lower current densities with respect to PECVD ones. The waveguiding, confining and emission properties of active silicon-on-insulator (SOI) slot waveguides and photonic crystal (PhC) slabs containing Er and silicon nanocrystals are also studied in details. It is shown that the vertical emission properties of the slot waveguides are changed dramatically by patterning the waveguide core with a two-dimensional PhC lattice with an enhancement of the near-normal emission by more than two orders of magnitude with respect to the unpatterned slab. Finally, the advantages of using erbium silicate compounds as silicon compatible efficient emitters will be demonstrated. It is shown that carrier mediated excitation is achievable in these systems when a silicate/silicon multilayer is fabricated. Moreover electroluminescent devices based on Er silicate compounds will be demonstrated.These data are presented and the route towards electrically-driven amplifiers and lasers will be discussed.

Si-laser development is critically dependant on the optimization of optical gain in silicon structures. We present measurements of optical gain for samples containing silicon nanocrystals (Si-nc), produced by ion implantation, with different sizes and depth distributions providing a wide spread of optical emission and wave guiding characteristics. The effect of the ion fluence, thermal annealing temperature, and hydrogen passivation on the optical properties of the samples has been investigated. Measurement of optical gain in Si-nc structures is very challenging, requiring a very careful approach. Special attention must be given to pumping beam uniformity, diffraction effects, detector optical coupling and sample degradation at high pump laser power, factors that are critical to the accuracy of VSL measurements. Detailed measurements of the amplified spontaneous emission (ASE), for wavelengths in the 600 to 950nm range, have been performed using a high-resolution variable stripe length (VSL) approach with CW excitation. The dependence of optical gain on pumping power has been investigated. Gain measurements using the pump and probe (PnP) method were performed in support of the VSL measurements. The net modal gain has been evaluated for different spectral wavelengths and compared to the spectral dependence of the photoluminescent emission (PL). Generally, highest optical gain is at the peak wavelength PL emission.The photoluminescence spectra recorded during VSL measurements do not exhibit significant spectral distortion or energy shifts. This strongly suggests that the stimulated emission process is associated with an inhomogeneous optical amplifier medium. The parametric dependence of optical gain on ion implantation, thermal annealing and passivation will be discussed.

Various hybrid approaches to build-up laser diode for silicon photonics has been recently successfully realized, as e.g. various III-V materials integrated within the silicon chip or near-infrared emitting Er-coupled silicon nanocrystals (Si-ncs) in SiO2. Lasing in a purely silicon-based material, however, has not yet been reported and represents an important keystone for the basic semiconductor research and further possible application as a low cost silicon laser source. One of the promising materials for this purpose appears to be a material based on oxidized silicon nanocrystals (Si-ncs) embedded at high densities in SiO2 based matrix. The high volume fraction, small diameter and blue-shifted emission of Si-ncs are theoretically required for the observation of the stimulated emission. The high density requirement, however, does not match within (the present technology state-of-the art) the high optical quality standard necessary for the optical feedback application. In our previous publications we have made a step closer towards the lasing in silicon nanocrystals [1-4], however, the optical quality of the samples was rather poor. In this contribution we wish to report on our recent study of the time resolved optical gain in a new samples generation, prepared by electrochemical etching and intense H2O2 post-etching of Si wafers. Such samples contain small Si-ncs (2-3 nm in diameter) embedded at very high densities (~10^19 Si-ncs/cm3) in an SiO2 based sol-gel matrix. They exhibit higher optical quality and emit below 600 nm. We investigate behavior of the optical gain between 500 and 650 nm. We brought out standard VSL (variable stripe length) and SES (shifting excitation spot) measurement using two different excitation laser systems to confirm our measurements. Namely we used femtosecond amplified Ti:sapphire laser (100 fs pulse length, rep. rate 200 kHz) and Q-switched Nd:YAG laser with nanosecond pulses (7 ns pulse length, rep. rate 10 Hz). We observed in both cases optical gain in the spectral region 500-600 nm, blue-shifted compared to PL spectrum. Material with such wide gain spectra can be interesting for its potential using in optoelectronics as a tunable laser light source.[1] Appl. Phys. Lett. 84 (2004), 3280[2] J. Appl. Phys. 99 (2006), 116108[3] Appl. Phys. Lett. 88 (2006), 251105[4] New Journal of Physics 10 (2008), 063014

We will review in this contribution our recent developments on silicon based light emitting diodes (Si-LEDs). Si-LEDs of silicon nanoclusters (Si-nc) emitting in the red will be first introduced and different solutions, like pulsed polarization and/or nitride-oxide double stack will be discussed to increase efficiency (up to 0.1%) and reliability. A simulator for the electrical injection and conduction in these materials has been developed, and its results will be compared with experiment and theory. Doping Si-nc in SiO2 with other impurities is shown to shift the emission wavelength, and results will be presented on high power white tunable Si-LEDs (co-doping with C) and 1.5 μm emitting Si-LEDs (co-doping with Er). Finally, other device architectures will be shown to have unique properties, like Si-LED MOS transistor, which shows fast built in modulation capabilities of the current (and emission) of the gate, due to the wide frequency response of the transistor transconductance and the fast quenching of the emission of the Si-nc by Auger non-radiative recombination.

Conventional optical components are too large for LSIs when compared with electronic components such as transistors. The fundamental difference in size comes from the fact that the wavelength of light (~1 μm) is longer than the de Broglie wavelength of electron (~10 nm). One of the approaches to fill the gap of this size is to use near-field optics which size is not constrained by the diffraction limit. Strong near-field can be created by the use of surface-plasmons, and they offer an approach to nano-scale photonic device. The size of the confined near-field region is comparable with the featured size of the materials. We have developed several types of surface-plasmon antennas for photodetectors. A small semiconductor structure is located near the antenna to absorb near-field light. The size of the structure can be made as small as that of the Schottky depletion layer, and hence the separation between electrodes can be reduced almost to the size of the near-field region.Silicon has not been a popular material for high-speed photodetectors in spite that it has been the best material for LSIs. One of the main reasons for this is that the absorption length of silicon is as large as 10 μm or more, resulting in a long carrier drift time and slow response. We have demonstrated a plasmon photodiode or a “Si nano-photodiode” that uses a small volume of near-field light to reduce the size of the silicon. The size of Schottky region in silicon can be reduced almost to the size of the near-field region (10-100 nm long) and the transit time of the carriers is fairly short accordingly. The full-width at half-maximum of the impulse response was as fast as 20 ps even when the bias voltage was less than 1 V. The surface plasmon antenna acts as a resonant cavity that temporarily reserves optical energy. The cavity function makes sense when the resonance is achieved in a shorter time than the response time of the photodiode.We have also developed a waveguide-integrated photodiode with a surface plasmon antenna for on-chip optical interconnects. A pair of interfacial periodic metal–semiconductor–metal Schottky electrodes was fabricated. It also works as a surface plasmon antenna. A high speed response time of 17 ps was obtained. We have demonstrated the operation of on-chip optical clock system by using this technology.

Quantum confined Stark effect (QCSE) in Si nanocrystals can be utilized for the production of Si based electro-optical devices sucs as optical modulators and waveguides, since light absorption and emission can be controlled by using an external bias. Optical sources based on waveguide devices such as ring resonators can be fabricated by making use of QCSE that can provide tunable light for optically functional circuits. Successful fabrication of such devices may lead to brakethroughs in the fabrication of integrated electo-optical systems. In this work, Quantum confined Stark effect (QCSE) on excitons confined in Si nanocrystals embedded in SiO2 matrix is demonstrated by photoluminescence (PL) spectroscopy at room and cryogenic temperatures. PL peak shifts to lower energies with increasing electric field are recorded as expected from carrier polarization within the quantum dot. Reducing the measurement temperature further enhances the QCSE due to improved localization at the lowest energy states of the quantum dot. The variation of the PL intensity with applied voltage and temperature are also studied to understand the effect of other mecahisms such as carrier escape from the nanocrystals and Auger recombination. It is shown that the emission intensity remains constant for a wide range of applied voltage indicating that the observed energy shift is related to QCSE rather than carrier population of the nanocrystal. We have also found that the effect depends on the polarity of the applied voltage. This dependence is studied through C-V measurements in which the charge injection into the oxide from the substrate can be observable.We have shown that experimental results agree well with the simpified theoretical approaches which commonly suggests the quadratic dependence of the energy shift on the applied electric field. Moreover, we have performed rigorous theoretical calculations based on an atomistic pseudopotential Hamiltonian solved using the linear combination of Bloch bands technique. We identify that the nanocrystals should have a diameter of about 5.6 nm according to the peak emission wavelength when matched to the calculated effective optical gap. Our theoretical results reproduce the peak emission shift as a function of the applied electric field without any fitting parameters. Furthermore, the theoretical analysis reveals that most of the Stark shift is due to valence states. The nonmonotonic behavior of the PL peak intensity is also reproduced by the theoretical calculations at 30 K and 300 K. In summary, this is the first demonstration of QCSE in silicon nanocrystals which is also supported by rigorous theoretical analysis.

Silicon nanocrystals (Si-nc) obtained by silicon implantation in silicon oxide (SiO₂) are known to exhibit either photoluminescence (PL) or electroluminescence (EL) under proper excitation. However, EL intensity is typically 100 to 1000 times weaker than PL intensity. Resistivity of the surrounding SiO₂ matrix and the effect of the high electric field applied are suspected to be responsible for the weak EL. These hypotheses are being investigated by applying selected bias voltages (i.e. electric fields) to MOS-like devices containing Si-nc during the PL process. In this experiment, an amorphous SiO₂ layer with a thickness of ~60 nm has been thermally grown on an n-type silicon substrate, then two successive Si⁺ ion implantations have been made to ensure good uniformity of Si-nc within the oxide. Si⁺ ions were implanted with an energy of 25 keV to a fluence of 2.5 x 10¹⁶ cm^-2, for the first implantation, and with an energy of 12 keV to a fluence of 1.0 x 10¹⁶ cm^-2 for the second implantation. A 15 nm semitransparent gold electrode has been deposited on top of the oxide to allow for optical measurements while biasing the device. Photoexcitation is generated by a 405 nm laser diode directed at 45° on the top electrode of the samples, while the PL signal is collected normally to the electrode. Variable DC voltages have been applied between the top electrode and the back silver electrode, the latter forming an ohmic contact with the doped silicon substrate.The measurements show that in all cases, the PL intensity is significantly affected by the applied electric field, while no clear energy shift is observed. In forward bias, increasing the voltage enhances the PL intensity until a specific field strength is reached (~0.4 MV/cm), then gradually quenches the PL signal. In reverse bias, however, increasing the voltage always reduces the PL intensity. Also, the electric current, I, flowing through the device has been measured as a function of the applied voltage V, yielding I-V curves with the expected rectification behavior typical of a diode. However, in reverse bias, the laser photoexcitation induces a strong enhancement of the current along with the PL quenching described above, while no noticeable change occurs in forward bias. Finally, for all biases, the rate of change of both the PL intensity and the current, I, with respect to voltage, V, is high for electric field strengths under ~0.4 MV/cm and becomes lower at higher fields. These results show that an applied electric field has indeed important effects on the luminescence processes in Si-nc. These effects are discussed in terms of photogeneration, injection and recombination of charge carriers, and pave the way to a stronger EL signal in Si-nc embedded in a SiO₂ matrix.

Hetero-epitaxy provides a means to combine lattice-mismatched semiconductor materials without introduction of defects. This approach makes it possible to engineer the electronic and opto-electronic properties of layered materials through the control of the composition and strain. Si_1-xGe_x on Si is among the best studied hetero-epitaxial strained systems, and conventional transmission electron microscopy (TEM) played an important role in the observation of layer morphology and misfit dislocation in relation to the strain relaxation in this system. Contrary to conventional high resolution TEM lattice images where Si and Ge atoms can hardly be distinguished, annular dark field (ADF) imaging obtained in a scanning electron microscope (STEM) provides a means to probe atomic distribution due to the Z (atomic number) contrast nature of this technique. The ADF-STEM image contrast is known to be dependent on the average atomic number, Z, in a simple Zⁿ power-law relationship. For most microscope geometries, n is in the range of 1.6 to 1.9. Contradictory to this compositional contrast prediction, ADF-STEM imaging of dilute GaN_yAs_1-y strained films showed that the lower average atomic number strained GaN_yAs_1-y layers were brighter than the higher average atomic number neighboring GaAs ,This phenomenon was explained by the local atomic displacement around substitutional N [Wu et al, J. Phys: Condens. Matter, 20, 075215 (2008)]. Here the ADF-STEM technique is applied to the study of compression strained Si_0.8Ge_0.2 layers and tensile strained dilute Si_1-yC_y (y ≤ 0.012) epitaxial layers grown on (100) Si substrates by molecular beam epitaxy (MBE). A series of ADF images were obtained with detector inner semi-angle ranging from 29 – 92 mad, and sample thickness ranging from 50 – 300 nm. The observed contrast of the ADF-STEM images is discussed in relation to the different strain status in Si_1-xGe_xand Si_1-yC_y epitaxial layers. The intensity line profile of the ADF-STEM images is analyzed to obtain the composition profile in the film. The results obtained for different sample thicknesses and ADF detector angles agree very well with composition profiles obtained with analytical TEM techniques: energy dispersive X-ray spectroscopy (EDX) and electron energy loss spectroscopy (EELS), as well as secondary ion mass spectrometry (SIMS) and Auger electron spectroscopy (AES).

There have been numerous attempts toward efficient light emission and lasing in the various types of Si nanostructures. Dynamics of high-density carriers in Si-based nanostructures play an essential role in light emission and optical gain processes. Under the high-density excitation, the nonradiative Auger recombination causes the saturation of the photoluminescence (PL) intensity in Si-based nanostructures. The understanding of the mechanism of the Auger recombination in the Si nanostructures is one of the most essential and crucial issues for the physics of highly dense carriers in nanostructures and the design of Si-based laser. In this paper, we report the PL dynamics in the Si_1-xGe_x/Si superlattice (SL) and single quantum wells (SQW) under high-density excitation at low temperatures. The Si_1-xGe_x/Si heterostructures are a material system compatible with the standard Si processing technology and provide us with a unique opportunity to investigate dynamics of high-density carriers in the artificially-controlled nanostructures. The samples were the 99 period strained Si_1-xGe_x/Si SL with x=0.15 and the SQW. The time-integrated PL spectra and the PL dynamics under 3.0 eV excitation were measured at 10 K by a InGaAs array detector and a gated-photon counting system with a photomultiplier. In the SL, three PL peaks, assigned to the TO, TA, and no phonon-assisted transition, are observed. With increasing the excitation intensity, the PL lines show the broadening to the higher energy side, and the blueshift of the PL peak energy appears. These phenomena are explained by the saturation of the localized exciton and the redistribution of excitons to the delocalized states. Moreover, the saturation behavior of the PL intensity and the rapid PL decay are observed under high-density excitation. These behaviors are due to Auger recombination. To extract the characteristics in the SL, we compare the PL intensities in the SQW with the SL under the same excitation conditions. While the PL intensities in the SL are proportional to the excitation intensity up to 10⁶ W/cm², those in the SQW show the saturation behavior, indicating the appearance of the Auger loss, even under the weak excitation of 10² W/cm². Although the carrier sheet density is about 10² times smaller than that in the SQW by considering the well periods in the SL (99 pairs), the significant difference indicates that the Auger recombination in the SL is suppressed compared to the SQW. This suppression is believed to be related to the delocalized states in the SL. Due to the electronic coupling between the wells, the delocalized state in the SL extends over the Si barrier layer. The Auger rate in the bulk Si was calculated to be smaller than that in the Si_1-xGe_x well layer. Therefore, the high-density carriers in the SL show the small Auger loss, compared to the SQWs. The superlattice structures have many advantages over the bulk crystals and SQWs, because of their small Auger rate and strong PL.

Si_1-xGe_x alloy nanowires (SiGeNWs) synthesized by the vapor-liquid-solid(VLS) growth process have been attracting increasing interests because the electronic properties can be tailored in an extended range compared to those of SiNWs, and they can be used as transistors, chemical sensors, and light emitting devices. For these applications, precise control of the composition and the impurity profiles, and the development of the technique to characterize the profiles are indispensable. In this work, we employ micro Raman spectroscopy to characterize concentration and distribution of Ge and electrically-active impurities in in-situ boron (B)-doped SiGeNWs synthesized by gold (Au)-catalyzed chemical vapor deposition (CVD).In Raman spectra of SiGeNWs, three major peaks are observed at 500-520, 400-410, and 280-290 cm^-1 due to the vibrations of adjacent Si-Si, Si-Ge, and Ge-Ge pairs, respectively. The Ge composition can be estimated from the intensity ratio. Furthermore, the active B concentration can be estimated from the spectral shape. We show that the Si-Si mode of B-doped SiGeNWs has a long tail towards a high-wavenumber side. This spectral shape is attributed to Fano resonance between discrete phonon Raman scattering and continuous electric Raman scattering caused by the excitation of holes in the valence band. To evaluate the concentration of active B atoms from the asymmetric spectral shape, the spectra are fitted by Fano resonance formula and asymmetry parameters are extracted. From the comparison of the asymmetry parameter with those obtained for reference samples, i.e., B-implanted Si, the concentration of active B atoms is roughly estimated. We show that there are distributions of the Ge and the active B concentrations within individual B-doped SiGeNWs. The comparison of these distributions in SiGeNWs reveals that there is strong correlation between the concentrations, i.e., high Ge concentration region is always more heavily B doped. The correlation strongly suggests that supply of B₂H₆ during VLS growth enhances conformal deposition of high Ge and B concentration layers and the conformal deposition results in the distribution of the Ge and the B concentration within SiGeNWs.

Si nanocrystals (NCs) embedded in a SiO₂ matrix are widely investigated in view of their potential photonic, optoelectronic and photovoltaic applications. In spite of that, many important unresolved issues remain. A particularly challenging one concerns relaxation and recombination processes of hot carriers in Si-NCs.In this contribution, we present optical investigations of electron/hole relaxation in Si-NCs. The study ha