T. Parent

Design, development, and testing of real-time feedback controllers for semiconductor etching processes using in situ spectroscopic ellipsometry sensing

Real-time feedback controllers for two semiconductor etching processes are developed. Both controllers rely upon in situ spectroscopic ellipsometry measurements of sample thickness for their feedback variables. Spectroscopic ellipsometry (SE) is a commonly used nondestructive, noninvasive in situ sensor for dry etching. The first etching process we consider is the thermal chlorine etching of gallium arsenide. An empirical/first principles physics-based model for the etching process is developed. A linear-quadratic controller based on the model is designed and tested. The second etching process is the electron cyclotron resonance freon-14/oxygen (CF/sub 4/O/sub 2/) plasma etching of silicon nitride thin films. An adaptive etch rate controller for the fluorocarbon plasma etching process is designed, implemented, and tested.

Modeling and simulation of thermal chlorine etching of gallium arsenide with application to real-time feedback control

A dynamic model for the simulation of thermal chlorine etching of gallium arsenide is developed. The primary motivation for the development of the simulation is the design and testing of real time adaptive feedback controllers which rely upon in-situ optical measurements of etch depth obtained via spectroscopic ellipsometry. The basis for the model is an empirically derived relationship between etch rate, chlorine pressure, and substrate temperature. The chlorine pressure in the chamber is regulated by a throttle valve which determines the effective pumping rate of a turbo-molecular pump which is used to evacuate chlorine pressure dynamics and a second-order damped harmonic oscillator with zero-order hold valve position command inputs is used to model the dynamics of the throttle valve. An output equation is used to model the fact that ellipsometry based etch depth and chamber pressure can be observed at discrete time intervals. Unmeasurable parameters which appear in the model are identified, and the model is validated using experimental data. An adaptive linear-quadratic Gaussian based controller based on our model which forces etching to precede at a desired rate while estimating the often difficult to measure substrate temperature is designed and then tested using our simulation.

Multi-variable adaptive control of CF/sub 4//O/sub 2/ plasma etching of silicon nitride thin films

A two input real time feedback adaptive controller for the electron cyclotron resonance (ECR) CF/sub 4//O/sub 2/ plasma etching of plasma enhanced chemical vapor deposited (PECVD) silicon nitride thin films is designed and simulation tested. Variations in etch rate resulting from factors such as etch chamber wall seasoning which are inherent to plasma etching necessitate the use of feedback and adaptive control to achieve precise and reliable etching of ultra-thin films. In this paper, an adaptive controller is designed which maintains a constant desired etch rate by adjusting the microwave power which drives the plasma and a throttle valve which determines the pressure in the etching chamber. The controller is tested using a simulation based upon laboratory empirical data.

A neural-network-based approach to determining a robust process recipe for the plasma-enhanced deposition of silicon nitride thin films

We consider the problem of locating a process recipe that produces outputs which are, in some sense, least sensitive to small fluctuations in the process condition. Specifically, we determine the most robust process recipe for the plasma-enhanced chemical vapor deposition of silicon nitride thin films having specified optical properties. An appropriate sensitivity functional describing the relationship between the process inputs and outputs and their localized variability is defined in terms of response surfaces and the response surface gradients. Determining the most robust process recipes which produce films with a given refractive index is then formulated as a constrained minimization problem. The silicon nitride films are characterized by spectroscopic ellipsometry and the requisite response surfaces are obtained by training feedforward artificial neural networks with available data. Numerical findings are presented, validated via simulation, and discussed.

Spectroscopic ellipsometry (SE) based real-time control of CF/sub 4//O/sub 2/ plasma etching of silicon nitride

Real time feedback controllers for the electron cyclotron resonance CF/sub 4//O/sub 2/ plasma etching of plasma enhanced chemical vapor deposited silicon nitride thin films are designed and tested. Variations in etch rate resulting from factors such as etch chamber wall seasoning which are inherent to plasma etching necessitate the use of feedback control to achieve precise and reliable etching of ultrathin films. In this paper, a non-adaptive and an adaptive controller are proposed for this task. The controllers are implemented and tested both in simulation studies and in the laboratory on the actual etching chamber. The results of these tests are reported and discussed.

Locating sensitivity minimizing process inputs in the plasma enhanced chemical vapor deposition (PECVD) of silicon nitride thin films: a neural network based approach

We consider the problem of locating a process recipe that produces outputs which are least sensitive to small fluctuations in the process condition. A sensitivity functional describing the relationship between the inputs and outputs and their localized variability is defined using feedforward artificial neural network response surfaces. The most robust process recipe is formulated as a minimization problem. Numerical findings are presented for the problem of finding the most robust process recipe for the plasma enhanced chemical vapor deposition of silicon nitride thin films having a specified refractive index.

In Situ Spectroscopic Ellipsometry for the Real Time Process Control of Plasma Etching of Silicon Nitride

Spectroscopic ellipsometry (SE) is a commonly used non-destructive, non-invasive in-situ sensor for dry etching. SE measures the change in the polarization state of light reflected from a surface. Sample thickness is obtained by fitting a model to the experimental ellipsometry data. In this paper we describe the design, testing and evaluation of an SE based adaptive real time feedback controller for etch rate regulation in CF 4 / O 2 plasma etching of silicon nitride films. The feedback variable is the current etch rate as determined from the in-situ SE measurements of the films thickness. The controller compensates for drifts in etch rate which occur during a given etch, and adaptively adjusts for the run-to-run variability inherent to plasma processing. Experimental results are presented and discussed.

Feedback control of thermal chlorine (Cl/sub 2/) etching of gallium arsenide (GaAs) using in-situ spectroscopic ellipsometry sensing

Real time feed-back control of etching is becoming necessary to meet the degree of process reproducibility demanded by the increasingly strict process requirements of advanced semiconductor manufacturing. The feasibility of using in-situ spectroscopic ellipsometry, being sensitive to film thickness, surface roughness and substrate temperature, to achieve real-time feedback control of etch rate in dry-etching is investigated for the case of thermal Cl/sub 2/ etching of GaAs(100). The etch rate is modeled as a function of Cl/sub 2/ pressure and substrate temperature with all sample preparation and etch rate measurements made in-situ, thus minimizing surface contamination effects. The dynamics of the chamber pressure as controlled by the position of a gate valve in front of a pump is modeled. The resulting nonlinear model with states etch rate, pressure, valve position and valve position velocity, and commanded valve position as control is linearized. A discrete-time linear quadratic Gaussian (LQG) compensator is designed and simulated on the nonlinear system.