Pulsed laser crystallization of hydrogen-free a-Si thin films for high-mobility poly-Si TFT fabrication

The possibility to fabricate high-mobility polysilicon TFTs by nanosecond pulsed laser crystallization of unhydrogenated amorphous Si thin films has been investigated. Two types of lasers have been used: a large area ( 1 cm²) single ArF excimer laser pulse and a small diameter ( 100 m) frequency-doubled Nd:YAG laser beam, working in the scanning regime. Processed films have been characterized in detail by different optical and microscopic techniques. Device performances indicate that the best results are achieved with the excimer laser leading to high mobility values (up to 140 cm²/Vs) which are much larger than in polysilicon TFTs fabricated onto the same quartz substrates by low-temperature thermal (630° C) crystallization of amorphous Si films (_fe55 cm²/Vs).     

Studies on the high-frequency properties of 〈111〉, 〈110〉 and 〈100〉 oriented GaAs IMPATT diodes

High-frequency analysis has been carried out to predict the rf performance of 111, 110 and 100 oriented p ⁺ nn ⁺, n ⁺ pp ⁺ (single drift region) and n ⁺ npp ⁺ (double drift region) GaAs IMPATT diodes for opertion at 35 and 60 GHz. The microwave performance is observed to be highly sensitive to crystal orientation in case of p ⁺ nn ⁺ and n ⁺ npp ⁺ diodes whereas orientation of the substrate has negligible effect on n ⁺ pp ⁺ avalanche diodes. The calculation shows that 111 oriented GaAs IMPATT diode would provide the largest magnitude of negative resistance and negative conductance for both SDR p ⁺ nn ⁺ and DDR n ⁺ npp ⁺ diodes which indicates that high microwave power with high conversion efficiency can be realised from these 111 oriented GaAs devices. This result can be explained from the experimental data of electron and hole ionization rates for different orientations in GaAs.     

Mechanism and kinetics of the diffusion of gold in silicon

The diffusion of Au in Si is known to take place via the interchange of Au atoms between substitutional (Au _s ) and interstitial (Au _i ) sites. So far it has generally been believed that this interchange involves lattice vacancies (V) and that it occurs via the Frank-Turnbull mechanism V+Au _i ⇆Au _s . It is stated in the literature that this model explains the observation that the Au _s concentrationC _s ^m in the centre of Au-diffused Si wafers increases with timet according to . We show that this statement is incorrect, i.e., the Frank-Turnbull model cannot account for the law. Such a dependence is expected in the case of Si wafers with a sufficiently low density of internal sinks for self-interstitials if the Au _i −Au _s interchange is controlled by the so-called kick-out mechanism Au _i ⇆Au _s +1. Since this mechanism involves self-interstitials (I) the present result is in accordance with the fact that under high-temperature equilibrium conditions the dominating intrinsic point defects in Si are self-interstitials and not vacancies as in Ge or metals.     

Comparison of profile tailing in SIMS analyses of various impurities in silicon using nitrogen,oxygen, and neon ion beams at near-normal incidence

We have performed a systematic SIMS study into the effect of (i) the chemical nature and (ii) the energy of the primary ions on the decay length which characterizes the exponential fall-off of impurity sputter profiles. The samples consisted of low resistivity, p-type Si covered with thin metallic overlayers. Bombardment was carried out at 2° off normal. Aspect (i) was investigated for tracers of Cu and Ga using N ₂ ⁺ , O ₂ ⁺ , and Ne⁺ primary ions at an energy of 5 keV/atom. The effect of the beam energy, aspect (ii), was studied for eight different tracer species and N ₂ ⁺ primary ions at energies between 2 and 5 keV/atom. In the case of Ga, was found to be shorter with N ₂ ⁺ or O ₂ ⁺ primary ions (=7.0 and 7.5 nm, respectively) than with Ne⁺ (=12 nm). This effect is attributed to beam induced formation of Si₃N₄ or SiO₂ layers, whereby the effective width of the internal distribution of intermixed Ga impurities in the Si subsystem is reduced significantly. In contrast to Ga, the decay length for Cu is smallest under bombardment with Ne⁺ (=16 nm), quite large with N ₂ ⁺ (26 nm) and extremely large with O ₂ ⁺ (2.2 m). Segregation of Cu atoms at the Si₃N₄/Si and the SiO₂/Si interface, respectively, is responsible for this depressed impurity removal rate. Within experimental accuracy the observed variation of the decay length with N ₂ ⁺ energy E [keV/atom] can be written in the form =kE ^p, where k and p are element specific parameters which range from k=1.2 nm for Pb to 10 nm for Cu and from p=0.6 for Cu and Ag to 1.0 for Pb. The results are discussed with reference to conceivable shapes of the distribution of intermixed impurity atoms.On leave from NTT Applied Electronics Laboratories, 3-9-11, Midori-cho, Musashino-shi, Tokyo 180, Japan     

Impurity distribution and electrical characteristics of boron-doped Si1−x Ge x /Si p +/N heterojunction diodes produced using pulsed UV-laser-induced epitaxy and Gas-immersion laser doping

A two-step pulsed UV-laser process which independently controls the metallurgical and electrical junction depth of a Si_1–x Ge _x /Si heterojunction diode has been implemented. Pulsed Laser-Induced Epitaxy (PLIE) combined with Gas-immersion Laser Doping (GILD) are used to fabricate boron-doped heteroepitaxial p ⁺/N Si_1–x Ge _x /Si layers and diodes. Borontrifluoride is used as the gaseous dopant source in the GILD process step. Boron incorporation and activation are investigated as a function of laser energy fluence and the number of laser pulses using SIMS and Halleffect measurements. The dose of incorporated dopant is on the order of 10¹³ cm^–2 per pulse. The B profiles obtained are flat except for a peak at the interface resulting from segregation effects. The B and Ge distributions are compared with shifts in the turn-on voltage of p ⁺/N Si_1–x /Si heterojunction diodes produced by the process. The GILD/PLIE process is spatially selective with the resulting diodes fabricated being quasiplanar. Hole mobilities in the heavily doped Si_1–x Ge _x films are found to be slightly lower than in comparable Si films.Presently at the Oregon Graduate Institute, Beaverton, OR 97006, USA     

High-rate laser-direct-write dry etching of titanium

Titanium surfaces can be etched spatially selective in a chlorine atmosphere under 488 nm cw Ar⁺-laser irradiation focused to 3 m with well-controlled etch depth and high etch rate. By scanning the substrate, patterns can be generated by laser direct writing with high scan speed. The dependence of the etch rate on various parameters, such as laser power, scan speed and chlorine pressure, is described, and the impact on three-dimensional structuring of titanium is discussed.     

Fast etching and metallization of SiC ceramics with copper-vapor-laser radiation

The etching of polycrystalline SiC is studied with the help of radiation of a copper-vapor laser either in air or under the layer of a liquid (H₂O, DMSO). The etching rate in air is as high as 0.24 m/pulse, in DMSO 0.07 gm/pulse at an energy density of 16 J/cm². The etched surface is characterized with Scanning Electron Microscopy (SEM) and X-ray diffractometry. Etching of SiC ceramics in air revealed the partial amorphization of SiC and the formation of microcrystals of elementary Si with an average size of 300 Å. The etched surface of SiC ceramics takes on the ability to reduce Cu from a corresponding electroless plating solution. The adherence of the deposit is as high as 30 N/mm² and is a function of the scanning velocity of the laser beam.     

Lattice disorder in silicon induced by 2.0 MeV Cu+ irradiation

The effect of 2.0 MeV Cu⁺ irradiation on Si(100) crystal has been studied by the Rutherford backscattering/channeling technique. Analysis of the lattice disorder distribution has been performed under 100 direction of tilting off from the target normal: 7°, 30°, and 45° as well as different doses. The lattice disorder distributions in Si(100) have been compared with TRIM'89 simulation. The results show that the lattice disorder distributions in Si(100) under different irradiation angles seem to be in good agreement with TRIM'89 simulation. When the dose increases up to 8.7×10¹⁴ ions/cm², the defect concentration increases leading to the formation of an amorphous layer.     

Laser projection-patterned etching of (100) GaAs by gaseous HCl and CH3Cl

Laser projection-patterned etching of GaAs in a HCl and CH₃Cl atmosphere performed using a pulsed KrF-excimer laser (=248 nm, =15 ns) and deep-UV projection optics (resolution 2 m) is reported. The etching process carried out in a vacuum system having a base pressure of 10^–6 mbar is shown to result from a purely thermochemical reaction. Etching takes place in two steps: (i) between the laser pulses, the etchant gas reacts with the GaAs surface-atomic layer to form chlorination products (mainly As and Ga monochlorides), (ii) local laser surface heating results in the desorption of these products allowing further reaction of the gas with the surface. The influence of the etching parameters (laser energy density, gas pressure and pulse repetition rate) on the etch rate and the morphology of the etched features was studied. Etch rates up to 0.15 nm per pulse, corresponding to the removal of 0.5 GaAs molecular layer, are achieved. The spatial resolution of the etching process is shown to be controlled by the heat spread in the semiconductor and by the nonlinear dependence of the etch rate on the surface temperature. As a result, etched features smaller or larger than the projected features of the photomask are achieved depending on the laser energy density. Etched lines having a width of 1.3 m were obtained at low fluences by the projection of 2 m wide lines onto the GaAs surface.     

Hot carrier effects in double injection phenomena

Carrier heating is shown to be responsible for unusualI(V) characteristics observed in small-sizep ⁺ nn ⁺ silicon on sapphire (SOS) devices. The classical quadratic law of the semiconductor regime becomes linear for high fields. The influence of dimensions and doping is experimentally checked and a model, based on the regional approximation method, is proposed. The key role is assumed by the hot-carrier region, growing from the cathode, where the electron and hole mobilities are field dependent: µ~E^–. A full agreement with the experiment in SOS is found for=0.5. The operating of hot carrierp ⁺ nn ⁺ devices can be described concretely with usual formalism by using the concept of effective carrier mobilities, which depend on the applied voltage.     

On the theory of the diffusion of gold into dislocated silicon wafers

The paper presents a theoretical study of the diffusion of gold into dislocated silicon wafers in terms of the kick-out mechanism Au _i Au _s +I, where Au _i , Au _s , andI mean Au interstitials, substitutional Au atoms, and Si self-interstitials, respectively. In agreement with experiments it is found that the Au _s concentration in the centre of a wafer,C _s ^m , increases with the durationt of the diffusion anneal according toC _s ^m =C _s ^eq (k ₀ t)^1/2 except forC _s ^m values in the vicinity of the solubility limitC _s ^eq of Au _s . Approximate analytical expressions fork ₀ as a function of the densityN _I of the dislocations acting asI sinks are given for the entire regime 0N_I<+t8.     

Decoupling crystalline volume fraction and VOC in microcrystalline silicon pin solar cells by using a µc‐Si:F:H intrinsic layer

Microcrystalline silicon thin film pin solar cells with a highly crystallized intrinsic μc‐Si:F:H absorber were prepared by RF‐plasma enhanced chemical vapour deposition using SiF₄ as the gas precursor. The cells were produced with a vacuum break between the doped layer and intrinsic layer depositions, and the effect of different subsequent interface treatment processes was studied. The use of an intrinsic μc‐Si:H p/i buffer layer before the first air break increased the short circuit current density from 22.3 mA/cm² to 24.7 mA/cm². However, the use of a hydrogen‐plasma treatment after both air breaks without an interface buffer layer improved both the open circuit voltage and the fill factor. Although the material used for the absorber layer showed a very high crystalline fraction and thus an increased spectral response at long wavelengths, an open‐circuit voltage (V_OC) of 0.523 V was nevertheless observed. Such a value of V_OC is higher than is typically obtained in devices that employ a highly crystallized absorber as reported in the literature (see abstract figure). Using a hydrogen‐plasma treatment, a single junction μc‐Si:F:H pin solar cell with an efficiency of 8.3% was achieved.

     

Diffusion of gold in dislocation-free or highly dislocated silicon measured by the spreading-resistance technique

The diffusion of Au in dislocation-free or plastically deformed Si (10¹¹ to 10¹³ dislocations/m²) was measured with the aid of the spreading-resistance technique. The Au profiles produced indislocation-free Si slices by in-diffusion from both surfaces possess nonerfc-type U shapes as predicted by the so-called kick-out diffusion model. This model is used to calculate the contribution of self-interstitials to the (uncorrelated) Si self-diffusion coefficient,D _I ^SD =0.064×exp(–4.80 eV/kT)m² s^–1, from the present and previous data on the diffusivity and solubility of Au in Si in the temperature range 1073–1473 K. Inhighly dislocated Si the diffusion of Au is considerably faster than in dislocation-free Si. From the erfc-type penetration profiles found in this case, effective Au diffusion coefficients were deduced and combined with data on the solubility of Au in Si. ThusC _i ^eq D _i=0.0064 ×exp(–3.93 eV/kT)m² s^–1 was obtained in the temperature range 1180–1427 K, whereC _i ^eq andD _i are the solubility and diffusivity of interstitial Au in Si.     

Rapid thermal annealing of electrically-active defects in virgin and implanted silicon

Chemical etching of single-crystalline (100)Si induced by pulsed laser irradiation at 308, 423, and 583 nm has been investigated as a function of the laser fluence and C1₂ pressure. Without laser-induced surface melting, etching requires Cl radicals which are produced only at laser wavelengths below 500 nm. With low laser fluences ((308 nm)<100 mJ/cm²) etching is non-thermal and based on direct interactions between photocarriers and Cl radicals. For fluences which induce surface melting ((308 nm)>440 mJ/cm²) etching is thermally activated. In the intermediate region both thermal and non-thermal mechanisms contribute to the etch rate.     

Excess capacitance of ZnO-Au varactors

Doped ZnO single crystals were deposited with gold and indium in 1×10^–8 Torr vacuum. The lithium-doped ZnO single crystals and the gold interface revealed not only a Schottky diode but also varactor characteristics. TheI-V andC-V characteristics of ZnO:Li-Au devices were determined in the 0–140 mV and 0–1.5 V ranges.The frequency dependence of ZnO:Li-Au varactors was investigated in the 6–550 kHz range and the value of the most efficient varactor frequency was found to be 50 kHz for the lithium-doped samples prepared.To bring further insight into the matter the concept of excess capacitance was introduced and 1/C ²=f(–V) curves were rearranged between 0–150 mV where Schottky characteristics are non-linear. The excess capacitance values of lithium-doped varactors were determined at four different frequencies and ranged from 26 pF at 50 kHz to 70 pF at 6kHz.Finally, the bulk donor concentrations of the single crystals were calculated from the modifiedC-V curves to beN _D= 3×10²⁰ m^–3. On the other hand, the bulk donor concentration determined from the non-modifiedC-V curves wasN _D=1.02×10²² m^–3.     

The influence of graphite boat material on the purity of LPE InGaAs

This paper describes the purity of LPE InGaAs layers grown in graphite boats, machined from various graphite materials. The influence of the material is clearly visible if the growth solution is sufficiently pure. Carrier concentrations n<2×l0¹⁵ cm^–3 and mobilities(77 K)> 38000 cm²/Vs are routinely achieved for suitable graphite materials already from the third run of a new large boat applying a prebake of only 15 h. Small boats yield even better results (n=0.5×10¹⁵ cm^–3 and(77 K)=49500 m²/Vs). The sticking of In-rich solutions to the graphite does not depend on the material but is solely dependent on the surface roughness. The problem of graphite particle abrasion is discussed.     

Current transport in Pd2Si/n-Si(100) Schottky barrier diodes at low temperatures

The forward current-voltage (I–V) characteristics of Pd₂Si/n-Si(100) Schottky barrier diodes are shown to follow the Thermionic Emission-Diffusion (TED) mechanism in the temperature range of 52-295 K. The evaluation of the experimentalI–V data reveals a decrease of the zero-bias barrier height ( _b0) and an increase of the ideality factor () with decreasing temperature. Further, the changes in _b0 and become quite significant below 148 K. It is demonstrated that the findings cannot be explained on the basis of tunneling, generation-recombination and/or image force lowering. Also, the concepts of flat band barrier height and T ₀-effect fail to account for the temperature dependence of the barrier parameters. The 1n(I _s /T ²) vs 1/T plot exhibits nonlinearity below 185 K with the linear portion corresponding to an activat ion energy of 0.64 eV, a value smaller than the zero-bias barrier height energy (0.735 eV) of Pd₂Si/n-Si Schottky diodes. Similarly, the value of the effective Richardson constant A** turns out to be 1.17 × 10⁴ A m^–2 K^–2 against the theoretical value of 1.12 × 10⁶ A m^–2 K^–2. Finally, it is demonstrated that the observed trends result due to barrier height inhomogeneities prevailing at the interface which, in turn, cause extra current such that theI–V characteristics continue to remain consistent with the TED process even at low temperatures. The inhomogeneities are believed to have a Gaussian distribution with a mean barrier height of 0.80 V and a standard deviation of 0.05 V at zero-bias. Also, the effect of bias is shown to homogenize barrier heights at a slightly higher mean value.     

UV-CVD deposited SiNH films on InP: Optimization of the physico-chemical and interface properties

High quality silicon nitride films are deposited at low temperature on InP substrates by direct photolysis at 185 nm of a NH₃-SiH₄ gas mixture. The composition of the films is measured by nuclear analysis. The thickness and refractive index are obtained by ellipsometry at 632.8 nm. As-deposited and post annealed samples are electrically characterized: quasi-static I(V) at 5×10^–4 Hz and C(V) characteristics at 1 MHz are performed on InP MIS diodes structures in order to optimize bulk and interface properties. At 250° C and 4 Torr, it is found that the highest critical field (measured for a leakage current density of 10^–9 A/cm²) is obtained for the injected ratio [SiH₄]/[NH₃]=2%. For these conditions, the film is stoichiometric, the critical field is 4 MV/cm and the resistivity is 6×10¹⁵ cm. The interface state density (N _ss) on InP is deduced from Terman analysis. The annealing conditions and the surface cleaning of InP have been optimized in order to reduce the N _ss which is, for our best conditions, as low as 2×10¹¹ eV^–1 cm^–2.     

Investigation of three-terminal voltage-controlled switching devices prepared by molecular beam epitaxy

Three-terminal GaAs switching devices prepared by molecular beam epitaxy using p ⁺-n ^–-(p ⁺)-n ^–-n ⁺ structures are fabricated. The effects of the third-electrode position and the possible voltage-controlled operation on the device performance are discussed. Concepts are proposed to obtain new and improved voltage-controlled properties. The internal barrier of one proposed structure can be modulated directly and is found to be effective for the studied structures. The position of the third-electrode is found to affect the electrical properties profoundly due to different dominant mechanisms. Comparisions are made by defining a control efficiency. Due to the idea of varying the gate position, a conceptual understanding of such a set of results would enhance our understanding of the physics of bulk barrier devices in general.     

Computer simulation of the small-signal admittance and negative resistance of double avalanche region IMPATT diodes

A generalized small-signal computer simulation of double avalanche region (DAR) n ⁺-p-v-n-p ⁺ Si and InP IMPATT diodes has been carried out for different frequencies and current densities taking both drift and diffusion of charge carriers into account. The simulation results show that both symmetrically and asymmetrically doped devices based on Si and InP exhibit discrete negative conductance frequency bands separated by positive conductance frequency bands. The magnitudes of both negative conductance and negative resistance of InP devices are larger than those of Si devices in case of symmetrical and asymmetrical diodes. Further, the negative resistance profiles in the depletion layer of these diodes exhibit a single peak in the middle of the drift layer in contrast to double peaks in double drift region diodes.