ALBERT

Notes: Analytical solutions of Fick's one-dimensional diffusion equation for a semi-infinite medium with an exponentially decaying initial impurity concentration profile and different boundary conditions are presented. The properties of the solution for a constant surface concentration are discussed more extensively. The theoretical results are applied to the diffusion of nitrogen incorporated in silicon by laser melting. The diffusion coefficient of nitrogen in silicon near its melt temperature is estimated to be of the order of 1×10−6 cm2/s. This confirms the recently reported high diffusion coefficient values for nitrogen in silicon.

Notes: Stress in local isolation structures is studied by micro-Raman spectroscopy. The results are correlated with predictions of an analytical model for the stress distribution and with cross-sectional transmission electron microscopy observations. The measurements are performed on structures on which the Si3N4 oxidation mask is still present. The influence of the pitch of the periodic local isolation pattern, consisting of parallel lines, the thickness of the mask, and the length of the bird's beak on the stress distribution are studied. It is found that compressive stress is present in the Si substrate under the center of the oxidation mask lines, with a magnitude dependent on the width of the lines. Large tensile stress is concentrated under the bird's beak and is found to increase with decreasing length of the bird's beak and with increasing thickness of the Si3N4 film.

Notes: The different steps that have to be taken in order to derive information about local mechanical stress in silicon using micro-Raman spectroscopy experiments, including theoretical and experimental aspects, are discussed. It is shown that the calculations are in general less complicated when they are done in the axes system of the sample. For that purpose, the secular equation is calculated in the axes system [110], [−110], [001], which is important for microelectronics structures. The theory relating Raman mode shift with stress tensor components is applied using two analytical stress models: uniaxial stress and planar stress. The results of these models are fitted to data from micro-Raman spectroscopy experiments on Si3N4/poly-Si lines on silicon substrate. In this fit procedure, the dimensions of the laser spot and its penetration depth in the substrate are also taken into account. © 1996 American Institute of Physics.

Notes: We discuss recent advances made in the theory and measurements of stresses and strains in Si-based heterostructures containing submicron- and micron-size features. Several reports on theoretical as well as experimental studies of stresses in the substrates with local oxidation of silicon structures on the surface have been published recently. With the advent of GeXSi1−X strained layers and stripes extensive studies of both the stripe and the substrate stresses have also been made. Unlike the previous calculations and analytical models, recent finite element (FE) calculations take into account the coupling between the film–substrate stresses without making the approximation that the interface is rigid or that there is no variation of stresses in the stripes in a direction perpendicular to the interface. The results of these calculations have been compared with the analytical models and limitations of the analytical models have been pointed out. Micro-Raman measurements of the stresses in the stripes, quantum wires, quantum dots, and substrates have been made. The measured values of stresses in GeSi stripes and quantum structures agree well with the calculated values by the FE method. The micro-Raman measurements showed that as the ratio R=2l/h (2l is the width and h is the thickness of the stripe) decreases, the shape of the measured normal stresses in the substrate under the stripe (plotted in a direction parallel to the interface) changes dramatically, from concave upward to convex upward. Generation of dislocations in laterally small layers is also discussed briefly. FE calculations of trench-induced stresses which include the effect of the anisotropy of Si have also been made recently. In these calculations realistic experimental conditions were simulated to determine the oxide shape, oxide–interface stresses, and intrinsic and thermal stresses of the polysilicon fill. These values were then used as inputs for the FE calculations. Calculations of stresses induced by oxide-filled trenches were also made assuming that Si is isotropic and that the oxide fill has the same elastic constants as Si. These calculations and results of an earlier analytical model implemented under the same assumptions gave identical results; however, the calculated stress values were in error of 20%–30%. The maximum resolved shear stress for the 60° dislocation induced by a trench is 30% more if it is aligned in 〈110〉 direction rather than in the 〈100〉 direction. This explains the common observation that the 〈100〉-oriented trenches cause fewer dislocations than the 〈110〉 trenches. The characteristics of trench isolated as well as junction isolated bipolar transistors have been studied. The trench isolated transistors had 20% higher gain; however, the collector–base capacitance was higher by up to 50% in the trenched transistors. The increase in capacitance was caused by the anomalous diffusion of the antimony dopant from the buried collector layer induced by the stress field of the trenches. The effect could be eliminated by increasing the depth of the trench. The trenched devices also had higher emitter–collector leakage current caused by the dislocations generated by the trench induced stress field. © 1996 American Institute of Physics.

Notes: Critical thickness hc has been calculated for capped and uncapped lattice mismatched II–VI semiconductor epilayers. Both the old equilibrium theory and the improved theory have been used. The calculated values are compared with the experimental data on epilayers of several II–VI semiconductors and alloys. The observed values of hc are larger than the calculated values, a result similar to that observed with GeSi and InGaAs strained layers. The discrepancy is attributed to the difficulty in nucleating the dislocations. Strain relaxation in layers with thickness h〉hc is also calculated. Observed strain relaxation in ZnSe layers grown on (100) GaAs shows good agreement with the equilibrium theory. In other cases, the observed relaxation is sluggish and the residual strain is larger than the calculated value. Many authors have observed that strain near the surface of the II–VI epilayers is small and increases as the depth increases. We describe an improved model to explain this observation. The agreement between the prediction of our model and the observed strain distribution is excellent. A new model based on continuum elasticity theory is described to explain strain oscillations during the initial stages of growth of highly mismatched layers. In highly mismatched layers, the dislocations are distributed uniformly. A model to interpret this observation is suggested. © 1998 American Institute of Physics.

Notes: The first results are presented of a comparative study of separation by implanted oxygen structures using spectroscopic ellipsometry (SE) and transmission electron microscopy. The strength of SE to measure the layer thicknesses of multilayer structures nondestructively is illustrated. Some limitations of the technique are also indicated.

Notes: In this paper a semiempirical model is presented which describes the amount of silicon that is molten in laser recrystallization of polycrystalline silicon layers. The model is based on the fitting of rather simple, physically meaningful expressions to experimentally obtained values for the width of the molten zone induced by a scanning laser beam. The influence of laser power, scan velocity, preheating temperature, and capping layer structure can be described.

Notes: Zone melting recrystallization (ZMR) of polycrystalline silicon on SiO2 can offer an interesting, i.e., cheaper, alternative to the dielectric isolation technology used for high-voltage integrated circuits or smart power devices. For that purpose crystalline Si layers of 10–15 μm thickness are needed. In this work a mercury-arc lamp stripheater was used to recrystallize 10 μm thick polycrystalline silicon films. In unseeded layers, grain boundaries and subgrain boundaries appeared. By applying seeding, single crystalline areas of at least 1 mm by 1 mm were obtained. In these layers stacking faults were revealed as being the major crystal defect. In ZMR oxygen, nitrogen, and carbon are the major impurities which are incorporated into the silicon either intentionally or unintentionally. Among these impurities, nitrogen and carbon are believed to play a crucial role in promoting the wetting of the insulator and cap layer by liquid silicon. In this paper the distribution and transport of oxygen and nitrogen during ZMR are studied. It is shown that nitrogen can cause dendritic growth and that oxygen is unlikely to do so.

Notes: Deep level transient spectroscopy of electron irradiated p-type silicon reveals a defect level at Ev+0.19 eV, which during anneal treatments at 200 °C gradually transforms into a band with Ev+0.24 eV. Both energy levels however, are reported in literature to be the donor level of the divacancy. In the present study it is proposed that during the low-temperature anneal the divacancy interacts with oxygen, forming a V2O complex. During heat treatments at temperatures in the range between 250 and 450 °C a further shift of the deep level to higher energy positions is observed which might be related with other vacancy-oxygen complexes. © 1995 American Institute of Physics.

Notes: In this letter, it is shown that micro-Raman spectroscopy allows easy, nondestructive determination of the C49 and C54 phase of titanium silicide with μm resolution within single structures with area dimensions down to 1×1 μm2 and along isolated line structures with widths down to 0.25 μm. The micro-Raman spectroscopy technique is used to study isolated 0.25–5-μm-wide TiSi2 lines with thicknesses as small as 16 nm that are formed in both crystalline Si and polycrystalline Si. The phase mapping ability of the technique is demonstrated on several 80-μm-long, 0.35-μm-wide TiSi2 lines that are part of four-terminal line resistance devices created using complementary metal–oxide–semiconductor processing. © 1997 American Institute of Physics.