Table of contents

The emission and absorption spectra of quantum wells containing electron or hole gas are reviewed. We show that trions, also known as charged excitons, play a dominant role in determining these spectra. We discuss issues related to their behaviour at zero and high magnetic fields, their far-field and near-field spectra, and their role as a probe for delicate correlations of the surrounding electron gas.

An investigation of hardening the buried oxides (BOX) in separation by implanted oxygen (SIMOX) silicon-on-insulator (SOI) wafers to total-dose irradiation has been made by implanting nitrogen into the BOX layers with a constant dose at different implantation energies. The total-dose radiation hardness of the BOX layers is characterized by the high frequency capacitance–voltage (C–V) technique. The experimental results show that the implantation of nitrogen into the BOX layers can increase the BOX hardness to total-dose irradiation. Particularly, the implantation energy of nitrogen ions plays an important role in improving the radiation hardness of the BOX layers. The optimized implantation energy being used for a nitrogen dose, the hardness of BOX can be considerably improved. In addition, the C–V results show that there are differences between the BOX capacitances due to the different nitrogen implantation energies.

GaN metal–semiconductor–metal (MSM) ultraviolet photodetectors with transparent tungsten (W) electrodes were fabricated and characterized. It was found that the 10 nm thick W film deposited with a 250 W RF power could provide a reasonably high transmittance of 68.3% at 360 nm, a low resistivity of 1.5 × 10⁻³ Ω cm and an effective Schottky barrier height of 0.777 eV on u-GaN. We also achieved a peak responsivity of 0.15 A W⁻¹ and a quantum efficiency of 51.8% at 360 nm from the GaN MSM UV photodetector with W electrodes. With a 2 V applied bias, it was found that the minimum noise equivalent power (NEP) and the maximum D* of our detector were 1.745 × 10⁻¹⁰ W and 7.245 × 10⁹ cm Hz^0.5 W⁻¹, respectively.

The direct way to investigate the effect of an electric field F on the optical properties of a quantum well (QW) is to change the well width (Miller et al 1986 Phys. Rev. B 33 6976). Photoreflectance studies of a 100 Å and a 250 Å Ge/GeSi modulation doped QWs possessing 10⁴ V cm⁻¹ perpendicular fields are presented. This technique modulates the QW electro-absorption, which we compute using wavefunctions obtained from a 8 × 8 k ⋅ p calculation. They are shown to have a pronounced tunnelling into the bandgap in the 250 Å QW, in contrast to the 100 Å QW. Kramers–Kronig analysis is used to determine the corresponding changes in the QW refractive index, and overall reflection from the entire sample (including barriers) is obtained within a 2 × 2 matrix formalism. In a QW of width L_z, the Franz–Keldysh effect is expected to be important in an energy interval extending from near the bandgap E_o to E_o + eFL_z. Its role in altering QW optical properties as a function of L_z is tested by comparing calculated and experimental spectra in this interval.

Thin films of AgSbSe₂ have been prepared by heating a Sb₂S₃–Ag stack, with the Ag side in contact with a chemically deposited Se thin film, which served as a planar source of Se vapour. A two-stage process, consisting of the formation of an Ag₂Se film through the reaction of a vacuum deposited Ag film in the Se vapour at about 80 °C, and its reaction at 300 °C with a chemically deposited Sb₂S₃ film, results in the formation of the AgSbSe₂ film. X-ray diffraction studies illustrate the structural evolution in the formation of this film. The material possesses an indirect optical band gap of about 0.9 eV. Thermoelectric measurements on the films showed a Seebeck coefficient of 500 µV K⁻¹ (p-type), and thus a hole concentration of ∼10²² m⁻³. The feasibility of application of these films as a photovoltaic absorber material is illustrated for the structure SnO₂–CdS–(i)Sb₂S₃–(p)AgSbSe₂, in which an open circuit voltage of 530 mV has been observed under an intensity of illumination of 2 kW m⁻² using a tungsten–halogen lamp.

Based on the finite-difference method, the multi-transverse-mode behaviours of vertical-cavity surface-emitting lasers (VCSELs) are investigated numerically. For two types of injection current with different contact radii, the transverse-mode competition, the spatial distribution of the carriers inside the active region of VCSELs and the carrier diffusion effects are investigated thoroughly. Results show that the spatial dependences of the optical modes and the current injection profile play important roles in determining the transverse-mode behaviours of VCSELs. It is found that a narrow disc contact or a centre-vicinity ring contact is favourable for the excitation of the fundamental mode, while the higher-order modes are benefited for a wide disc contact or a ring contact with larger contact radius; the carrier distribution shows nonuniformities in both the radial and the azimuthal directions when the spatial nonuniformities of the optical mode are considered. It further shows that the carrier diffusion also affects the laser modal competitions and the spatial hole burning (SHB) effects greatly by influencing the carrier spatial distribution.

Capacitance– and reverse current–voltage (C–V) measurements have been performed on a Schottky barrier structure incorporating InAs quantum dots (QDs) embedded in n-type InAlAs matrix. The electron confined energy level positions in the QDs have been deduced from a capacitance–voltage analysis. Three electron levels have been found at about 240 meV, 180 meV and 70 meV via the InAlAs conduction band edge, and were attributed to the s ground state, the first p excited state and the d state, respectively. The variation of the leakage current density as a function of the depletion layer width has revealed an additional reverse current in the plane containing the dots. It was attributed to the discharging of the InAs QDs through the InAlAs barrier. The effective generation lifetime depth profile was derived from the C–V and the reverse current–voltage measurements. The effective generation lifetime was found to be controlled by the confined states in the QDs. Electrical field effect studies lead us to suggest a mechanism of electrons injection and emission into/from the confined states in the QDs.

As-grown and oxygen-annealed titanium oxide films grown on silicon substrate were studied. The titanium oxide films were prepared by metalorganic chemical vapour deposition in the temperature range of 400–650 °C. Compared with as-grown titanium oxide films, the oxygen-annealed titanium oxide films have better electrical properties and the stoichiometry of the titanium oxide film dominates the electrical properties. The leakage current can reach 2 × 10⁻⁷ A cm⁻² under an applied field of 5 MV. The hysteresis loop shift voltage and interface state density are 6 mV and 2.6 × 10¹¹ cm⁻² eV⁻¹, respectively.

A novel physical model for the process of silicon anisotropic wet chemical etching is proposed. Based on the actual process of chemical etching reaction, a series of microscopic parameters describing the process have been put forward and non-linear equations with a removal probability function have been listed. For the transfer probability function in the equations, not only have the removal probabilities corresponding to the different microscopic states been included, but also the influence of etching temperature and etchant concentration. Moreover, the influences of the first neighbourhood atom and second neighbourhood atoms on etch rate are all included in the transfer probability. Having compared the calculated results against the experimental data, the feasibility of the model has been demonstrated.

An equation that includes the effects of non-parabolicity on the conduction band of a semiconductor is applied to different nanosystems. This allows us to obtain analytical expressions in systems with simple geometries and reveals significant differences in relation to those obtained in parabolic cases. For quantum wells, we propose an empirical energy–wave vector dispersion relationship that can be used in electron transport simulation. In the case of quantum dots, by choosing suitable parameter values, the results obtained using the non-parabolic model are adjusted to those produced by tight binding. The introduction of non-parabolicity corrections is shown to greatly modify the energy spectrum. To illustrate this fact, we use as an example a system useful in a quantum dot laser.

Infrared spectroscopic ellipsometry (IRSE) in the wavelength range of 9.0–12.6 µm was used to study three GaN film samples: the unintentionally doped, the n-type and the p-type. The parameters of the lattice vibration oscillators and of the plasmon were obtained by fitting with the experimental data. The carrier concentration and the electron mobility of each sample were derived from the plasmon frequency, as well as the damping constant. The results were compared with those from the Hall measurement. The fitted values of the carrier concentration and those from the Hall measurement agree well. The mobilities derived from IRSE were only about half of those from the Hall measurement. The anisotropic refractive index and extinction coefficient dispersion relationships of every film were also obtained in the measured wavelength band.

The design and basic characteristics of a strained InGaAsP–InP multiple-quantum-well (MQW) DFB laser monolithically integrated with an electroabsorption modulator (EAM) by ultra-low-pressure (22 mbar) selective-area-growth (SAG) MOCVD are presented. A fundamental study of the controllability and the applicability of band-gap energy by using the SAG method is performed. A large band-gap photoluminescence wavelength shift of 88 nm was obtained with a small mask width variation (0–30 µm). The technique is then applied to fabricate a high performance strained MQW EAM integrated with a DFB laser. The threshold current of 26 mA at CW operation of the device with DFB laser length of 300 µm and EAM length of 150 µm has been realized at a modulator bias of 0 V. The devices also exhibit 15 dB on/off ratio at an applied bias voltage of 5 V.

Fabrication, characterization and modelling of heterojunction phototransistors (HPTs) are reported. The common-emitter (with current-biased base) and Gummel-plot (with voltage-biased base) modes are employed to characterize and fully comprehend what differences exist between the current- and voltage-biased HPT's performance. The results of further studies include the case when a series of different optical-power injection levels is illuminated at our HPTs. The performance of the current- and voltage-biased HPTs was also compared to that from a newly proposed HPT model and related equivalent circuit with good agreement found. Although an independent voltage source can be used to tune the operating point of a heterojunction bipolar transistor to a higher current level where the dc current gain is larger, the photocurrent generated within the base–collector (B–C) region offers few contributions to the final collector photocurrent. The optical gain obtained from the HPT biased using a high voltage is even smaller than that of the HPT with a floating base. It is concluded that (i) a current-biased HPT's dc base current and photocurrent generated within the B–C region entirely flow into the base–emitter (B–E) junction so that the device's optical gain is enhanced; (ii) however, no enhancement of optical gain for a HPT will be obtained using dc base bias, since the dc current gain is independent of collector (or base) current; (iii) a voltage-biased HPT behaves like a p–i–n photodiode and (iv) electrical base bias using a high external voltage source with a large series resistance is a possible way to enhance the optical gain of a HPT.

A (1 0 0)-oriented chemical vapour deposited (CVD) diamond film was grown by using a hot-filament chemical vapour deposition technique and then fabricated as an alpha particle detector. The current characteristics indicate a fine Ohmic contact for bias voltage up to 150 V. The dark current and the net current induced by α particles of ∼16.0 nA and ∼6.9 nA, respectively, are achieved at 100 V. The net current initially increases linearly with voltage and irradiation time, and then levels off, due to the 'polarization' effect. After irradiation, the dark current increases due to the release of captured carriers in shallow energy levels. The response of the CVD diamond detector to 5.5 MeV α particles demonstrates clearly that the 'polarization' and 'priming' effects intensively affect the detector performance. By pre-irradiating the detector with β particles, the charge collection efficiency is dramatically improved from 19.38% to 36.91% due to the 'priming' effect of the deep traps.

The experimental realization of a strained-SiGe p-type field-effect transistor (δFET), utilizing a δ-doped layer as the conducting channel, grown on a silicon-on-insulator (SOI) substrate, is fabricated and investigated. The proposed structure exhibits not only higher hole mobility but also higher breakdown voltage with reduced leakage current than those of the same structure fabricated on bulk-Si. High device linearity is still sustained and is attributed to good carrier confinement by not only the V-shaped potential well formed by the heavy δ-doping but also the reduced valence-band bending at the Si buffer by using SOI.

The thermal characteristics of antimonide ridge waveguide MQW lasers have been investigated numerically and experimentally, and compared with phosphide lasers in detail. Using the finite-element method, the heat accumulation and dissipation process of the lasers under CW and pulse driving conditions have been simulated, and quantitative thermal time constants were introduced to describe the cooling efficiency of the lasers. The non-uniform temperature distribution inside the active core of the laser and its effects on lasing spectra have been discussed, and confirmed by the measurement of the broadening of the lasing spectrum towards the blue side. A way to improve the thermal property of antimonide lasers have also been proposed and discussed.

Buried oxide, which separates the device area from the substrate in silicon-on-insulator (SOI) wafers, forms a diffusion barrier for transition metals in silicon. The impact of this barrier on the efficiency of traditional gettering techniques for iron and copper is evaluated using computer modelling. Several parameters essential for the modelling, such as the diffusivity of iron in SiO₂ and the segregation coefficient of iron and copper in SiO₂, are verified experimentally. It is found that all available data for the diffusivity of iron in SiO₂ (including data points from the literature and our own value of 1.4 × 10⁻¹³ cm s⁻² at 1100 °C) could be fitted by the equation D(Fe in SiO₂) = 2.2 × 10⁻² × exp(−3.05 eV/k_BT)(cm² s⁻¹). The solubility of iron in silicon dioxide was found to be 4.5 times to 5.5 times less than that in silicon at temperatures from 1020 °C to 1100 °C, which indicates that iron does not segregate in SiO₂. The solubility of Cu in silicon dioxide was determined to be half of that in silicon at 1150 °C and 3.3 times higher than in silicon at 690 °C. Modelling of gettering using these parameters revealed that buried oxide prevents iron from diffusing to gettering sites in the substrate at typical processing temperatures, thus rendering the substrate gettering techniques inefficient. On the other hand, the diffusion barrier protects the device area from contamination from the backside of the wafer. Copper has sufficiently high diffusivity in SiO₂ to diffuse through the buried oxide within a short time at 1000 °C; however, it may be difficult to remove copper from the device area because heavily doped areas of the devices could provide competitive gettering sites for copper. Possible gettering strategies for the SOI wafers are discussed.

Ag₂S thin films of 90–300 nm thickness were deposited in 3–8 h at 60–70 °C on glass substrates from a chemical bath containing silver nitrate, sodium thiosulfate and dimethylthiourea (DMTU) or tetramethylthiourea (TMTU). The glass substrates used for this purpose were previously immersed in a chemical deposition bath to deposit a ZnS thin film of 40 nm thickness in the case of the bath with DMTU or subjected to a pretreatment in a methanol solution of 3-(trimethoxysilyl)propyl methacrylate for deposition from the TMTU bath. Thin film yield is >40% for the TMTU bath, when the deposition produces 60 nm films at a substrate separation of 0.4 mm, and is about 30% for film thicknesses of 90–300 nm for both baths at a substrate separation of 1–5 mm. X-ray diffraction patterns similar to that of the mineral acanthite (Ag₂S) are observed in the case of the films heated in nitrogen at temperatures 100–200 °C; heating at 300–400 °C leads to freezing-in of the high temperature phase, argentite. Films annealed at 300 °C possess a direct (forbidden) optical band gap of about 1.0 eV. Typical electrical conductivity of the films is 10⁻³ (Ω cm)⁻¹. A photosensitivity of approximately 100 is observed in films annealed at 200 °C, under an intensity of illumination 2 kW m⁻² from a tungsten halogen lamp. Mobility–lifetime product is of the order of 10⁻⁵ cm² V⁻¹, which suggests a free carrier lifetime of about 0.1 µs.

A new direct extraction method for the determination of the parasitic capacitances of PHEMTs is presented in this paper. This method is based on a general scalable small signal equivalent circuit model under pinch-off bias condition. The main advantage of this approach is that all parasitic capacitances including C_pg, C_pd and C_pgd can be extracted simultaneously by using PHEMTs of different sizes but with the same pad structure. Good agreement is obtained between modelled and measured results for 2 × 20 µm, 2 × 40 µm, 2 × 60 µm and 2 × 100 µm gate width (number of gate fingers × unit gate width) double heterojunction δ-doped PHEMTs.

Within the framework of the dielectric continuum approximation and Loudon's uniaxial crystal model, the interface optical (IO) phonon modes and the corresponding Fröhlich electron–phonon interaction Hamiltonian in a wurtzite AlN/GaN/AlN quantum well wire (QWW) are derived and studied. Numerical calculations are mainly focused on the frequency dispersion of the IO phonons and electron–phonon interaction coupling function. Results reveal that, in general, there are four branches of IO phonon modes in the systems. The dispersions of the four branches of IO phonon modes are obvious only when the axial direction wave number k_z or the azimuthal quantum number m is small. The degenerating behaviour of the IO phonon modes in wurtzite QWW has also been observed for small k_z or m. When k_z or m are relatively large, with the increasing of them, the frequencies of these IO phonon modes converge to the two definite limiting frequencies in wurtzite single planar heterostructure, and this feature has been explained reasonably from the mathematical and physical viewpoints. The calculations of the electron–phonon coupling function show that, though some branches of IO phonon modes exchange their localized positions with each other at a large m, there always exist two branches of IO phonon modes localized on each interface. The high-frequency IO phonon modes compared with the low-frequency ones play a more important role in the electron–phonon interaction. Detailed comparison of the dispersion behaviours of the IO phonons and electron–IO phonon couplings properties in wurtzite QWWs with those in zinc-blende QWWs has also been made.

Continuous-wave (CW) mode operation InGaAsN/GaAsN double-quantum-well lasers with a laser wavelength of 1.295 µm are demonstrated by metal-organic chemical vapour deposition (MOCVD). With the use of a high-bandgap GaAs_0.9P_0.1 into the active region before the growth of p-type layers, a room temperature (RT) threshold current of 99 mA and the characteristic temperature (T₀) values of 155 K in a temperature range of 25–95 °C and 179 K in a temperature range of 25–85 °C are obtained from a 4 × 1000 µm² ridge waveguide uncoated laser diode. The T₀ value of the conventional structure without the high-bandgap GaAs_0.9P_0.1 is 118 K in a temperature range of 25–95 °C. High-temperature performance is improved and the results of numerical analysis suggest that it may be attributed to the reduced electronic leakage current.

A new measurement method for GaN films and their Schottky contacts is reported in this paper. Instead of the fabrication of Ohmic contacts, this measurement is based on a special back-to-back Schottky diode that has a rectifying character. A mathematical model indicates that the electronic parameters of the materials can be deduced from the device's I–V data. In the experiment of an unintentionally doped n-type GaN layer with a residual carrier density 7 × 10¹⁶ cm⁻³, the analysis by the new method gives the layer's sheet resistance ρ_s = 497 Ω, the electron mobility μ_n = 613 cm² V⁻¹ s⁻¹ and the ideality factor of the Ni/Au–GaN Schottky contacts n = 2.5, which are close to the data obtained by the traditional measurements: ρ_s = 505 Ω, μ_n = 585 cm² V⁻¹ s⁻¹ and n = 3.0. The method reported can be adopted not only for GaN films but also for other semiconductor materials, especially in the cases where Ohmic contacts of high quality are hard to make or their fabricating process affects the film's character.

We discuss experimental properties and possibilities of the maximum barrier height assessment for low Schottky barrier heights which have apparently Ohmic I–V curves. It is shown that the usage of numerical methods for low barrier height parameters extraction is very useful. We discuss the properties of the Schottky contacts with the same barrier heights but different semiconductor doping concentrations. For such contacts there can be an accumulation or depletion layer at the semiconductor surface for the same barrier height depending on the doping concentration. I–V curves of such structures have similar character, but a rather large difference can be expected in C–V curves.

InAsP/InGaAsP strain-compensated multiple quantum wells (SC-MQWs) were grown using gas source molecular beam epitaxy. The luminescent properties of the SC-MQWs after thermal annealing and direct wafer-bonding onto a GaAs substrate were studied. It is shown that photoluminescence (PL) intensities of the samples are improved by a factor of 4.1 upon annealing at 620 °C. The luminescence intensities of the samples bonded at 580 and 650 °C under ∼5 MPa pressure and 35 min annealing process are comparable with that of the as-grown ones. A proposed mechanism of defects annihilation relating to the evolution of nanoscale As-rich and P-rich clusters is then used to account for the dependence of PL intensities and the full width at half-maximum of InAsP/InGaAsP SC-MQWs on annealing temperatures.

We present a detailed design and experimental study of diode laser structures emitting at 808 nm based on the combination of a GaAsP quantum well with well-established AlGaAs waveguide structures. By increasing the thickness of the confinement layers of the laser structure, its vertical far field divergence is reduced down to 15° with only a small increase of the threshold current and small loss of efficiency. 200 µm aperture 'broad area' devices achieve at a heat sink temperature of 25 °C a continuous wave (CW) output power of more than 15 W with a wall-plug efficiency of 50% with a vertical far field divergence of 18°. This output power illustrates the excellent high-power performance by using super-large optical-cavity structures with improved beam characteristics in comparison to the conventional broad waveguide lasers.

The temperature dependence of current–voltage (I–V) and capacitance–voltage (C–V) characteristics of the Au/n-InP Schottky barrier diodes has been measured in the temperature range of 80–320 K. The forward I–V characteristics are analysed on the basis of standard thermionic emission (TE) theory and the assumption of a Gaussian distribution of the barrier heights (BHs). It has been shown that the ideality factor decreases while the barrier height increases with increasing temperatures, on the basis of TE theory. Furthermore, the homogeneous BH value of approximately 0.524 eV for the device has been obtained from the linear relationship between the temperature-dependent experimentally effective BHs and ideality factors. The modified Richardson plot, according to inhomogeneity of the BHs, has a good linearity over the temperature range. The value of Richardson constant A* has been found to be 5.97 A cm⁻² K⁻², which is close to the theoretical value of 9.4 A cm⁻² K⁻² for n-InP. Moreover, the temperature coefficient of the BH is found to be −3.16 × 10⁻⁴ eV K⁻¹ for Au/n-InP.

A new mechanism of thermo-emf generation in p–n structure based on separation of charge carriers by energy in space is suggested. This sorting is carried out by electron–phonon drag. Equations for thermoelectric current are obtained within the framework of a two-temperature model.

Low temperature photoluminescence (PL) studies have been carried out on ion-implanted silicon in order to elucidate upon the structure evolution of the self-interstitial (I) clusters as a function of implantation dose, energy, species and post-implantation annealing conditions. PL measurements on as-implanted and low temperature annealed (up to 450 °C) Si show a sharp X band and a W band at 1200 nm and 1218 nm, respectively. The W band shows gradual quenching of PL above ∼60 K with a characteristic activation energy of 59 meV. We argue that the W band originates from a compact di-interstitial cluster in Si. Short duration annealing at 600 °C results in multiple sharp peaks in the range of 1228–1400 nm for the high energy (MeV) and high dose (⩾1 × 10¹² cm⁻²) implantation, while low energy (keV) or low dose ions induce two broad peaks in the same wavelength range. Prolonged annealing at 600 °C induces primarily two broad peaks centred at 1322 nm and 1392 nm. These broad, but distinct, PL signatures are attributed to a chain of I-clusters, while the multiple sharp peaks possibly result from multiple configurations/excited states of the compact but bigger I-clusters. For annealing at and above 680 °C and dose of ⩾1 × 10¹³ cm⁻², the sharp PL peak observed at 1376 is attributed to {3 1 1} rod-like defects. We argue that the changing line shape and energy of the PL spectra with processing temperature is a possible indicator of the shape evolution of the clusters from compact to extended structures as predicted recently from simulation.

The structural defects near the film/substrate interface of a homoepitaxially grown film by the vapour–liquid–solid method and heavily doped 4H–SiC film were studied by transmission electron microscopy. The structural defects which were observed consisted of low density threading edge dislocations, high density basal plane screw semi-loops due to misorientation by low angle twisting around the c axis as well as rare stacking faults after the dissociation of basal plane perfect dislocations into partial ones. These defects, localized within a thin layer close to the film/substrate interface, accommodate the two lattices, leaving the rest of the overgrown film free of defects.

We investigate the extrinsic parasitics in symmetric ultra-thin body double-gate (SUTBDG) devices with raised source/drain (S/D). An analytical model for the fringing capacitance is derived by using dual conformal mapping and proven accurate by two-dimensional (2D) device simulation. From the fringing capacitance, we project an aggressive scaling in gate height as gate length shrinks, in order to maintain a roughly constant fraction of the fringing capacitance to the total gate capacitance in sub-20 nm transistors. Convenient analytical models for the extrinsic series resistances are also derived and validated by simulation. For a raised S/D structure with a wrapped contact and contact resistivity as good as 1 × 10⁻⁸ Ω cm², we find that the heavily doped thin-body region contributes the greatest part of the extrinsic series resistance, followed by the resistance of the contacted region and the spreading resistance. For a non-raised S/D with a wrapped contact, we find that the resistance of the contacted region is dominant. Our device simulation results suggest that optimal spacer thicknesses depend on the trade-off between fringing capacitance and series resistance.