Papers

AVS1997 Session EM+MS-FrM: Process Fundamentals for Microelectronics

Friday, October 24, 1997 8:20 AM in Room C3/4

Friday Morning

Start	Invited?	Item
8:20 AM		EM+MS-FrM-1 Process Modeling Fundamentals for Microelectronics Manufacturing L.J. Borucki (Motorola Advanced Customs Technologies) Process modeling needs in the microelectronics industry have evolved considerably since the era when 1D process modeling was considered adequate. It is quite common now for companies to establish dedicated organizations for the application of 2D vendor and university process modeling codes to the design and optimization of entire process flows. Yet this approach often fails to predict device performance well, even with extensive model calibration. Pressure to improve the situation is creating several trends. Emerging process modeling codes are all multidimensional and have relatively open model programming interfaces that allow one to quickly fix or extend a failing model or to build an entirely new model that fills an urgent and sometimes one-time need. Modeling analysis is going beyond bulk profile prediction and is encompassing the modeling of entire tools, such as RTP, CVD, and plasma reactors and plating tools in order to understand and control the factors that affect quality and cross-wafer uniformity. Multiscale approaches are being developed that connect such equipment models with feature scale models in order to predict patterning or loading effects. New numerical methods, such as level set algorithms, are making it possible to construct models to analyze 3D questions that were previously too hard. And finally, vast improvements in ab-initio methods and computing hardware are making calculation of formerly unknown and and experimentally inaccessible process or chemistry model parameters more feasible. Thus, the trend is toward a broader, deeper, and more flexible approach to process modeling.
9:00 AM		EM+MS-FrM-3 Modeling of Ultathin Oxynitride Gate Dielectrics Formed by Remote Plasma Nitridation of Silicon Dioxide D. Kapila, S. Hattangady, R. Kraft (Texas Instruments) A model was developed to predict the decrease in effective oxide thickness (delta(T_{ox,electrical}), due to increase in dielectric constant) during the process of remote plasma nitridation (RPN) of silicon dioxide. Three processing parameters (plasma power, plasma exposure time, nitrogen pressure) were varied. We used a three level CCD (Central Composite Design) design for modeling the nitridation process. With a CCD design we were able to get all the main and the quadratic interactions. The response surface model (RSM) was able to predict delta(T_{ox,electrical}) with an accuracy of R² (model fitting parameter) = 0.98 suggesting an excellent model for the predictability of the process. The within-wafer standard deviation of delta(T_{ox,electrical}) was small (sigma = 0.182 Å). The wafer-to-wafer standard deviation of the mean was also small (sigma = 0.25 Å). Both the values were comparable to that of the starting thermally grown oxide. The decrease in the effective oxide thickness is independent of the starting oxide thickness. From the models it can be seen that RPN is a highly predictable and uniform process.
9:20 AM		EM+MS-FrM-4 Combined Beam Profile Reflectometry, Beam Profile Ellipsometry, and Ultraviolet-Visible Spectrometry Characterization of Ultrathin Oxide-Nitride-Oxide Films on Silicon J. Leng, J. Opsal (ThermaWave, Inc.) We use beam profile reflectometry (BPR), beam profile ellipsometry (BPE), and deep UV spectrometry to characterize ultra thin oxide on nitride on oxide (ONO) films of total thickness less than 20 nm. The BPR and BPE capabilities are key elements to detect the multilayer nature of the ONO stack, while the deep UV capability together with the absorption edge of silicon nitride in this spectral region allows its thickness to be determined accurately. In modeling the three-layer stack we found that the insertion of a 1.2 nm interlayer between the Si substrate and the bottom oxide improves the fit to the data. Comparison of the optical measurements with cross- sectional transmission electron microscopy exhibited excellent agreement.
9:40 AM		EM+MS-FrM-5 Interface Studies of Titanium Nitride and Tungsten Nitride Composite Metal Gate Electrodes with Thin Dielectric Layers B. Claflin, G.L. Lucovsky (North Carolina State University) Future generations of deep sub-micron ULSI devices will require advanced gate electrode materials with demanding performance characteristics to maintain proper scaling behavior. In addition, because of the ultra-thin dielectric layers required in such devices, the interface between the gate and dielectric must be carefully controlled. Titanium nitride (TiN_x), tungsten nitride (WN_x), and their alloys are promising candidates as potential replacements for heavily doped polycrystalline silicon (poly-Si) for this application¹^,²; however, in order to control the threshold voltage in p- and n-channel devices, they will have to be used in stacked, or composite structures. In this work the initial formation and subsequent growth of reactive sputtered TiN_x/SiO₂ and WN_x/SiO₂ interfaces has been studied using high resolution Auger electron spectroscopy (AES). The integrity of these interfaces has also been investigated by rapid thermal annealing up to 850 C; TiN_x for example, does not chemically react with SiO₂ below 850 C. The electrical characteristics of TiN_x(WN_x)/SiO₂/Si MOS structures will be discussed as well as changes in the flatband threshold voltage for stacked metal electrode structures such as Al/TiN_x. Supported by ONR, NSF, and SRC. ¹D. H. Lee, S. H. Joo, G. H. Lee, J. Moon, T. E. Shim, J. G. Lee, Symp. VLSI Technology, pp 119-120, 1995. ²M. Wittmer, H. Melchior, Thin Solid Films 93, 397 (1982).
10:00 AM		EM+MS-FrM-6 High Resolution 2D Imaging of Dopant Profiles using Small Area Contact C-V Measurement Y. Li, D.J. Thomson (University of Manitoba, Canada) By combining the metal-semiconductor C-V profiling techniques with the two-dimensional scanning probe microscopy, a new technique for 2D delineation of semiconductor doping profiles based on the measurement of the local Schottky contact capacitance has been developed. When a metal probe is brought into contact with a semiconductor, a space-charged depletion region and therefore a capacitor is formed at the junction. By applying a small ac voltage, the voltage derivative of the contact capacitance can be measured with a lock-in amplifier. The amplitude of the derivative signal is a function of the ionized dopant concentration, and the sign gives the type of dopant. The local contact capacitance - voltage ( C-V) measurements on the standard doping samples show a monotonic behavior of capacitance derivative (dC/dV) versus doping concentration at the range of 10¹⁴ cm^-3 to 10¹⁸ cm^-3. Present work concentrates on the 2D doping profile imaging of MOSFET and BJT transistors. Lightly doped source/drain extensions and doping gradients in the channel region have been observed in PMOS transistor cross sections. Doping profiles of BJT active region have been imaged and doping gradients have been observed in the base region. A lateral spatial resolution of less than 50 nm is achieved by using a heavily boron doped diamond tip. The results demonstrate that this technique is capable of quantitative 2D characterization of semiconductor devices.
10:20 AM		EM+MS-FrM-7 Time and Length Scales in Plasma Etching S. Shankar, V. Singh, T. Phung (Intel Corporation) Low-Temperature plasmas or cold plasmas are used for microelectronic processing in deposition and in etching. The length scales range from centimeters (corresponding to reactor dimensions) to a few microns (corresponding to the feature that is being etched). In addition, there exist time scales ranging in magnitude from nanoseoconds (corresponding to plasma electron frequencies) to seconds (corresponding to etching times). It is the presence of these scales that make the plasma widely used in microelectronic processing. The interactions between the various time and length scales and their importance to understanding etching will be discussed in this paper. In addition, we evaluate the different models currently used to study these plasmas.
11:00 AM		EM+MS-FrM-9 Equipment Simulation of a Deep Silicon Trench Etch Reactor Chr. Werner (Siemens AG, Germany); B. Flietner (Siemens Components, Inc.) A deep trench etch process with high selectivity betweeen Si and SiO2 has been investigated by numerical simulation of gas flow in the reactor. Characteristic molecular demixing effects between O2 and the etch gas HBr across the very small holes in the gas distribution plate of the inlet were studied for different inlet designs. Moreover a simple chemistry model for the production of silicon bromide at the wafer and the deposition of a SiOBr passivating layer was implemented to allow a qualitative description of etch uniformity and selectivity. The model results were compared with detailed experimental data on specially prepared wafers with nonuniform trench distributions. It was found that the etch rate distribution is directly determined by the uniformity of the HBr species in the reactor while the concentration of the secondary product SiOBr as calculated in our simulations seems to be closely correlated with data on the selectivity in different regions of the wafer.
11:20 AM		EM+MS-FrM-10 Analytic Simulation of Silicon Etch Profiles from a HDP Plasma S. Abdollahi Alibeik, J.P. McVittie (Stanford University); V. Sukharev, P. Schoenborn (LSI Logic) Etch profiles in silicon for Cl2 etching in an inductively coupled high density plasma are simulated and compared to experiments using a simulator where there all 3D fluxes and the surface movement are calculated using fast analytic algorithms. The flux calculations include: direct from the gas phase, re-emitted based on a surface coverage/recombination model, angle dependent reflection, and redeposition of etch products. The etch model is based on reported beam measurements. The input fluxes for the profile simulator are obtained from hybrid plasma simulation for the specific chamber. It is found that redeposition of the the silicon etch product has a strong influence of the wall slope and the observed micro-trenching is not fully explained by ion reflection.
11:40 AM		EM+MS-FrM-11 Si Self-Interstitial Effective Diffusivity in the Presence of Carbon M.D. Johnson, T. Diaz de la Rubia (Lawrence Livermore National Laboratory) The goal of predictive process modeling requires that the fundamental processes be modeled on an atomic scale. These studies require the detailed understanding of the interactions of point defects and impurities. Furthermore, as the scale of micro-electronics shrinks, the need for more accurate models and hence increased complexity require the capability to simulate macroscopic features on an atomic scale. We have developed a flexible, high speed Monte-Carlo algorithm based on work by H. L. Heinisch¹, as the last step in a predictive process modeling effort which starts with ab-initio calculations, and includes molecular dynamics input as well. The first test of the algorithm is to examine a large system (up to 1Μm²) and observe the long time behavior. We have studied the effect of varying the carbon concentration on the effective diffusivity of Si self-interstitials produced by oxidation. We find that the profiles are sensitive to the specific atomic interactions included and that the effective diffusivity decays over very long time scales (0-10sec). We compare the results to recent experiments and also discuss the effect of varying the C concentration on the transient enhanced diffusion of boron implanted in Si. ¹H. L. Heinisch, J. Nucl. Mater., vol. 117, 46 (1983).