In-Situ Oxygen Monitoring of Rapid Thermal Process Chamber: Diagnosis of Gas Flow Dynamics and Wafer Processing
- PDF / 391,074 Bytes
- 7 Pages / 414.72 x 648 pts Page_size
- 68 Downloads / 137 Views
widely used. Illinois instruments Inc. model 3000 oxygen analyzer is employed as a monitoring apparatus. This model has an all-in-one construction, with the sensor itself build in a casing. The measureable oxygen range is 1 ppb to 100%. The apparatus was calibrated with 10 ppm standard 02 before installation and is connected to the exhaust gas line of a production RTP system. To minimize the effect of a dead volume between the sampling point and the inlet, the apparatus is placed as close to the exhaust line as possible. The sampling gas pipe is arranged to counter the incoming exhaust gas so as to sample unstagnant and fresh gas. The sampling gas pipe is purged by a small amount of pure N2 , during the RTP idle time, in order to suppress the effect of backdiffusion from the vent line. The RTP equipment used is an AST model SHS2000. This system has a rectangular and thin box-type quartz chamber(4). The gas flow is arranged to run unidirectionally from the inner side to the door opposed to that. The outlet is underneath the wafer. House nitrogen is used as a process gas throughout this study. DIAGNOSIS OF THE GAS FLOW IN THE CHAMBER The concept of the chamber diagnosis is to model the fluid system as an input-output system that makes a transient response. If the fluid is plug-flow, the tracer element added as input appears as it is at the outlet as output after a certain delay time (residence time). If the mixing in the chamber is perfect, the tracer element is dispersed at an infinite rate and the output shows an exponential decay. Real fluid systems are non-ideal and never follow these flow patterns. The deviations can be due to channeling of fluid, by recycling of fluid, or by creation of stagnant regions in the chamber. Elements of tracer taking different routes through the chamber require different lengths of time to pass though the chamber. These times for the tracer in the fluid stream leaving the chamber have the distribution, or the residence time distribution. When the tracer input is a pulse signal (5-function), the tracer output shows the residence time distribution itself. The response to an arbitrary input function is given by a convolution of the input and the residence time distribution. There are two important parameters of the residence time distribution. The most important measure is the mean residence time i, which is the time location of the distribution. Under steadystate and constant gas density conditions, which are assumed throughout this study, the mean residence time of the fluid with a flow rate F in a flow chamber with a volume V is simply given by
i =-V
(1)
F.
Obviously the mean residence time is independent of flow patterns or chamber geometry. The next important descriptive quantity is the spread of the distribution. This is commonly measured by effective diffusion coefficient or by Peclet number Pe (disucssed below). To characterize non-ideal flow within a chamber, we employ an one-dimensional dispersion model. This model is based on plug flow but mixing and stagnation of fluid
Data Loading...
