The effect of furnace grown SiO2 layers on the optical properties of p- on p+ (100) Si substrates are investigated. The real part, n, of the complex refractive index n* = n + ik is calculated for radiation measured in the infra-red (IR) region between 3000 and 8000 cm−1 where the extinction coefficient, k, is negligible. The expression for n is obtained using the Fresnel coefficients for a three medium air-oxide-Si model. Strain in the silicon, which affects n, and caused by the stress in the SiO2 layer, increases with oxide thickness. X-ray diffraction (XRD) was used to measure the strain in Si for oxides layers ranging from native to 5124Å. The data showed a monotonically increasing normal compressive strain, εN (up to 0.47%) with oxide thickness, however, the corresponding change in n due to strain was not well defined. The effect of strain on the direct optical gap, Ed, at 3.46 eV when determined from results of other investigators by electroreflectance, suggests an average shift in Ed of about 25 meV.

Tunneling currents through thin silicon dioxide films on p-type silicon are measured at electric fields greater than 5 MV/cm. At the onset of the Fowler-Nordheim tunneling, oscillations in the current are observed. These oscillations are used for characterizing oxide films grown by three different processes: rapid thermal chemical vapor deposition, rapid thermal oxidation and thermal oxidation. We have explored the correlation between the oscillatory tunneling currents and the breakdown fields, and find a low field dc component to correlate with the breakdown fields and obscure the oscillations.

In this study, 6-150mm p-type <100> wafers were cleaned, laser-scribed, and pre-process measured for stress. The wafers were then processed in a tungsten-halogen lamp RTP system with a target Rapid Thermal Oxidation (RTO) thickness of 100Å. Three categories of whole-wafer measurement techniques were used to characterize these wafers: optical, electrical, and mechanical. Optical inspection techniques included spectroscopic reflectometry (reflectivity), and a combination of beam profile reflectometry and beam profile ellipsometry (thickness). Electrical techniques included C-V plotting with a mercury probe (oxide thickness from Cmax, breakdown voltage, and interface trap density) as well as laser-induced photo-current scanning (minority carrier lifetime, minority-carrier diffusion-length). Mechanical inspection included wafer warpage and stress measurements as well as optical imaging inspection using the magic mirror method (damage and defects). Wafer measurements from these instruments (i.e., contour and 3-d maps) are used to characterize integrity of the RTO process.

The electrical properties of shallow P+/N junctions formed by boron outdiffusion from polycrystalline and epitaxial CoSi2 contacts are discussed. The CoSi2 contacts are grown on (100) Si from a sputtered metal layer by rapid thermal annealing (RTA) at 900°C in forming gas. The epitaxial CoSi2 (epi-CoSi2) is made from layers of 15 nm Co / 2 nm Ti, and the polycrystalline material (poly-CoSi2) is made from a 15 nm Co layer with no Ti. Dopant is introduced by ion implantation into the silicide and the P+/N junctions are formed by a second RTA step. Junctions are found to have total leakage current densities as low as 4 nA/cm2 for poly-CoSi2 and 12 nA/cm2 for epi-CoSi2 at -5V and metallurgical junction depths of 60 nm beyond silicide/Si interface after 700-800°C annealing.

Here we investigate the possibility to use rapid thermal annealed spin-on doped glass (SOG) films as a doping source as well as a surface passivation layer. The dilution of the SOG liquid source with methanol allows to control the oxide film thickness and the dopant concentration. The results show that the combination of spin-on film (after dilution) and RTP can produce shallow lightly doped emitters. Junction depths less than 0.2 µm and surface concentrations in the range 1−20 × 1019 cm−3 are obtained. Sheet resistances lower than 150 Ω/□ are easily reached. Moreover, the minority-carrier diffusion length is improved due to the gettering effect induced by phosphorus diffusion. The use of the remaining SOG oxide film as a passivation layer makes this technique favourable for some application.

Cross section TEM and channeling analysis show that the heating rate (HR) of a rapid thermal annealing (RTA) cycle affects the residual defect distribution in Si implanted with As+ to a heavy dose (≈ 1016 cm−2). Two defect bands are observed after solid phase epitaxial growth (SPEG): the first one centered at a depth corresponding to the projected range of the As (band I), and the second one located at depth corresponding to the original amorphous crystalline (a-c) interface (band II). The density of defects in band I is found to increase with the As dose, and with the annealing temperature (550 - 650°C, furnace annealing). However, for RTA (800 - 1000°C) both the density and depth distribution of these defects are dependent on HR. We propose that Si self-interstitials (SiI) are created at the a-c interface when As becomes substitutional during SPEG. The SPEG velocity determines whether the SiI are accommodated in the amorphous Si layer (low velocities) or are captured by the regrowing c-Si (high velocities)

Silicon samples were single implanted with B+ (5 x 1014 cm−2, 50 keV) or co-implanted with C+ (5 x 1015 cm−2, 55 keV). Rapid thermal annealing (RTA) with heating rates (HR) of 1°C/s and 100°C/s with dwell time duration from 2 s to 15 min was used for B activation and the results compared with those provided by conventional furnace annealing (FA) for 30 min. In single implanted samples it was found that below 600°C or above 800°C the activation always increases with the annealing time. However, in the temperature range of 600-800°C and annealing times longer than 60 s a fraction of the initially activated B concentration deactivates. For temperatures in the range of 700-800°C the deactivation is followed again by another activation period. There is no noticeable difference between the activation yields after RTA with high HR for 15 min or FA for 30 min. In the C co-implanted samples the activation of boron saturates after few minutes. In addition the deactivation process is significantly reduced. A model assuming interaction of the B atoms with point defects and the C atoms with Si self- interstitial atoms is proposed to explain the results.

In this paper, low-energy (45 keV) implantations of phosphorous and boron into silicon were studied. A comparison of doping profiles, secondary defect formation, electrical activation and diode leakage was made between Rapid Thermal Annealing (RTA) and conventional furnace annealing. The samples were analysed by High-Resolution X-Ray Diffraction (HR-XRD), X-TEM, SIMS, spreading resistance (SRP) and sheet resistance measurements.

The non-destructive HR-XRD technique combined with the novel simulation software was a very useful tool for the defect characterisation and for the choice of the optimum annealing temperature. Furthermore estimations of electrically active dopant atoms were made with HR-XRD by measurement of the strain. With RTA a substitutional dopant concentration of a factor 2 to 4 higher than with furnace annealing can be obtained, for P and B respectively. Electrical measurements show that not all of the substitutional dopants are electrically active, however. Thus estimates of the electrically active dopant atoms with HR-XRD require further study. Furthermore it appeared that RTA was superior to furnace anneal for lowering sheet resistances, defect removal and dopant profile broadening. However, furnace anneal gave the best results for diode leakage currents. This indicates that RTA processing needs to be further refined or that combined RTA/furnace processes need to be developed.

Proximity rapid thermal diffusion (RTD) has been investigated as a doping technique for p-type boron doped junctions. The efficiency of RTD has been studied as a function of process variables (temperature, time, and ambient) and evaluated based on sheet resistance measurements, secondary ion mass spectroscopy (SIMS), spreading resistance (SR), and Fourier transmission infrared absorption (FTIR) in a spin-on-dopant source (SOD). The doping efficiency in source wafers is controlled by different mechanism than in processed wafers. Strong influence of dopant incorporation in the processed wafers on oxygen content in the diffusion ambient is observed especially at low diffusion temperatures.

The modelling of temperature distribution on semiconductor wafers in common RTP-equipment is described. The incident intensity distribution on the wafer is calculated using raytracing. Based on this distribution the temperature distribution on the wafer is determined solving the two-dimensional heat conduction equation. If the dependence of a considered material property on the process temperature is known, the calculated temperature distribution can be convened to a distribution of this parameter.

The distinctive feature of the described algorithms is the two-dimensional treatment of the distributions using a grid of ring segments, each with equal area. This grid is identical to the usual circular test patterns of multipoint measurement equipment. This is convenient since the evaluation of temperature uniformity in RTP equipment is done mostly by mapping an appropriate temperature dependent material property. All calculated distributions can be presented by contour plots as well as 3-D plots. This results in a very suitable method to compare simulated and experimental wafer maps.

The agreement between simulated and experimental temperature distributions is shown.

We have developed an analytical tool to predict temperature variations occuring both across the wafer and down the thermal cycles, based on very simple 3-dimensional physical models depicting the heat transfer processes taking place during a typical rapid thermal processing (RTP) step. This model takes into account the convective and the radiative heat losses along with the thermal conduction within the wafer. An hybrid control volume formulation has been used to solve the heat conduction equation to yield numerical solutions of temperature variation, both on and within the wafer. The effect of ambient gas flow direction and the related asymmetric convective heat loss on the temperature non-uniformity across the wafer at different stages of the thermal cycle is well analyzed. Finally, the temperature non-uniformity across a wafer patterned with oxide is examined.

In this work we examine the effect of various system and wafer parameters such as the flowing ambient, the spatial distribution of incident energy flux, the slip-free ring and doping density on temperature non-uniformity occurring during rapid thermal processing (RTP). The effect of using two inert ambients, i.e. argon and nitrogen, on temperature non-uniformity occurring during RTP cycle is studied in detail. The importance of dynamic lamp power control in providing time dependent spatial variation of incident energy flux as a measure to improve the temperature uniformity is also examined. Finally the simulation studies are also done in the presence of a slip-free ring. The nature of photon capture by the silicon wafer and the correlation between doping density and the steady state temperature for any given open loop condition is discussed in detail. This analysis has been done using an analytical tool developed by us based on simple 3-dimensional physical models depicting a typical rapid thermal cycle[1].

Microscale radiation effects are responsible for the dependence of absorption and temperature distributions on the geometry of the layering structures and the spectral characteristics of the heat source. The effect of patterned wafers, which may contain several different structures and materials, on the wafer absorption characteristics are investigated for rapid thermal processing. A numerical model to determine the thermal radiative absorptivity of the wafer for different structures and materials is presented for different heating conditions. The resulting transient effects are determined numerically for different rapid thermal processes. The changes in radiative properties for rapid thermal annealing and chemical vapor deposition are investigated for patterned wafers.

Gas flow engineering involves gas dynamics optimization for effective ambient change before heating and for homogeneous convective cooling of the wafers during the heating steps. Multiple gas buffle system, dynamical gas handling, low pressure operation, low temperature edge guard ring and independent top and bottom heater bank control are the proper tools for this optimization. Silicon surface or interface damage during inert gas anneal can be avoided by addition of a small amount of oxygen.

A methodology for predicting the spatial and temporal distribution of film thickness is given for Chemical Vapor Deposition (CVD), and for thermal oxidation in Rapid Thermal Processor (RTP) systems, e.g. RTPCVD and RTO. The methodology is based on a wafer thermal model for the heat transfer to, from and within the wafer, a geometric ray trace algorithm to predict the radiant heat transfer from the lamps and reflectors to the wafer, a process model for the deposition or oxidation, and a gas flow model to predict the flow field in the RTP chamber. The CVD process is based on the Arrhenius deposition model, and thermal oxidation is based on a parallel diffusion model. The methodology has been validated by comparison of measured and predicted final film thickness from a cylindrical RTP system. The methodology is based on physical principles, with a minimum reliance on empirical relations and experimental data. As such it can be used for optimization of existing RTP designs and for the evaluation of proposed RTP configurations, such as new or novel lamp, reflector or chamber geometry.

A fully three dimensional model has been developed for simulating the thermal behaviour of a RTP furnace. This model consists of two components to achieve the whole analysis. The first component models the radiation heat flux density at the wafer surface as a function of the system geometry, the lamp position and intensity, and the wall reflectivities. The second component solves the heat conduction equation at each point within the wafer using appropriate boundary conditions, including convective cooling which effect depends on the process conditions. A particular attempt is made upon the achievement of a flat temperature profile over the wafer by investigating the system parameters in order to improve the RTP equipement.

We present a general purpose Monte Carlo method for the simulation of radiation heat transfer in rapid thermal processing (RTP) chambers. Three-dimensional mesh generation software is used to discretize the surfaces within the system, allowing the simulation of realistic chamber and reflector designs. An adaptive subdivision of the chamber geometry reduces the number of raysurface intersections which must be computed. The method models internal reflection, absorption, and transmission within participating media, and includes wavelength, temperature, and material dependent optical properties. Radiation heat transfer simulations are used to examine a reflector assembly, and to test the assumptions of optical wafer temperature measurement techniques.

The results of emissivity of (a) Si and SiO2 /Si in the temperature range of 320 to 1240 K and (b). HgCdTe at temperatures of 303 and 440 K, using Fourier Transform Infrared(FTIR) spectroscopy, in the wavelength range of 1.5 to 20 μm, have been reported here. The measurements on SiO2 /Si have been performed for oxide thickness ranging from 60 to 500 nm. These results coufled with calculations of emissivity from first principles lead us to model the wavelength dependence of emissivity. Emissivity estimates have been obtained for HgCdTe using its fundamental optical constants.

There are fundamental differences between conventional furnace processing (CFP) and rapid isothermal processing (RIP). The radiation spectrum of a conventional furnace consists of photons in the infrared and longer wavelength regions, whereas the spectrum of the incoherent light sources used in RIP consist of some ultraviolet, visible, and infrared photons. As compared to CFP, the photophysical and photochemical effects associated with RIP provide the capability of lower temperature processing. Due to photoeffects in RIP, differences are observed in junction movement and defect evolution with different lamp configurations and different kinds of lamps having differing spectra. This implies that diffusion models, in addition to considering processing time and temperature, should also account for the lamp configurations and spectra of the heating sources. The fundamental understanding of photoeffects in RIP can be further exploited to reduce the overall thermal budget used in the manufacturing of semiconductor devices. In this paper we present our study of photoeffects in RIP, which can be of help in the design of the next generation of RIP equipment.