Transfer learning based 3D fuzzy multivariable control for an RTP system

A first-principles approach to the modeling of a rapid thermal processing (RTP) system to obtain temperature uniformity is described. RTP systems are single wafer and typically have a bank of heating lamps which can be individually controlled. Temperature uniformity across a wafer is difficult to obtain in RTP systems. A temperature gradient exists outward from the center of the wafer due to cooling for a uniform heat flux density on the surface of the wafer from the lamps. Experiments have shown that the nonuniform temperature of a wafer in an RTP system can be counteracted by adjusting the relative power of the individual lamps, which alters the heat flux density at the wafer. The model is composed of two components. The first predicts a wafer's temperature profile given the individual lamp powers. The second determines the relative lamp power necessary to achieve uniform temperature everywhere but at the outermost edge of the wafer (cooling at the edge is always present). The model has been verified experimentally by rapid thermal chemical vapor deposition of polycrystalline silicon with a prototype LEISK RTP system. The wafer temperature profile is inferred from the poly-Si thickness. Results showed a temperature uniformity of +or-1%, an average absolute temperature variation of 5.5 degrees C, and a worst-case absolute temperature variation of 6.5 degrees C for several wafers processed at different temperatures.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

The rapid thermal processing (RTP) system requires critical control of temperature uniformity. For most RTP system, the structure and heating modes of heating component determined the control strategy. Major types of the heating system of nowadays RTP systems were reviewed and compared. Then a heating system for RTP was proposed to improve the uniformity of heaving emission, which includes optimally designed light path and high efficient light reflecting system. The simulating results of that were also given.

Issues of rapid thermal processing (RTP) system design and process applications are reviewed. Temperature measurement is the most important and limiting factor in current RTP systems. Problems related to the temperature measurement and control and potential solutions are discussed. Process uniformity control is another important issue in RTP system design. Reactor chamber design, selection and arrangement of heat source, as well as issues related to dislocation generation, patterned and doped wafer, and ramp-up (down) thennal cycle are considered. Experimental results (RTO, RTCVD silicon nitride and polysilicon) based on an in-house RTP system developed in our laboratory are taken as examples to demonstrate the process applications and system requirements for single wafer processing.

A new proprietary rapid thermal processing (RTP) method has been developed and implemented for the fast switching of a gas ambient surrounding a wafer inside a RTP furnace operated in the low pressure or atmospheric pressure regime. By controlling the gas flow pattern around a semi-conductor wafer, the effective volume of process gas interacting with the wafer has been reduced by a factor of 33. Thus, real-time gas sequencing can be realized so that multiple rapid thermal processing steps, involving dry and wet rapid thermal oxidation (RTO), rapid thermal nitridation (RTN), rapid thermal annealing (RTA) and chemical vapor deposition (CVD), can be carried out sequentially in a single RTP cycle. Experimental data from ion implant anneal and thermal oxide thin film growth will be presented to illustrate the major technical challenges and solutions associated with the integration of the new method into the Summit/spl trade/ RTP systems. Using this approach, the benefits of rapid thermal processing with high productivity and quick ambient switching is realized.

This paper highlights the basic principles of rapid thermal processing (RTP) systems and the important areas of concern. The basic system characteristics, the fundamental physics involved, and the techniques for temperature measurement and control are extensively reviewed. We summarize the options currently available for 15 RTP equipment manufacturers and point out the latest developments in RTP system design and temperature measurement. Some novel options for temperature control (optical, fiber optical, and photoacoustic) are included.

A novel method for the estimation of the temperature distribution in a rapid thermal processing (RTP) system is developed in this paper. The proposed method uses a proper orthogonal decomposition algorithm to extract the dominant modes of the temperature distribution and a reduced-order model is obtained. Then, a reduced-order sliding-mode observer is developed to capture the dynamics of the dominant modes. The estimated dynamics of the dominant modes can be used to reconstruct the temperature distribution. It is proved that the estimation error would be drawn into a small boundary rapidly. Test results confirm the effectiveness of the proposed observer for the RTP system.

In this paper, a model-based control method for RTP (Rapid Thermal Processing) system is proposed. RTP system is a device used to heat semiconductor wafers uniformly. It can heat wafers rapidly and reduce the processing time. However, it is very difficult for a RTP system to achieve the extremely stringent requirement on temperature uniformity (±1[K] or less) without suitably controlling and coordinating the radiative heat flux from the halogen lamps. In this research, the location of lamps is optimized first. Then, a feedback control system is designed which provides optimal temperature control during the thermal processing. The design is carried out based on a linearized model and H2 optimal control. The effect of sensor location is also investigated. Numerous simulations show that satisfactory performances are obtained in both the tracking and the uniformity of wafer temperature.

Rapid thermal processing (RTP) is becoming increasingly more important in semiconductor wafer fabrication. A nonlinear physical model for a generic RTP system is described. This generic RTP system is representative of the state-of-the-art RTP systems. A low-order nonlinear model is developed. A model-based LQG controller is designed for the low-order model which successfully controls the full-order nonlinear simulation model.

ABSTRACTA 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.

The design of a decentralized controller has been studied as an integral part of the development of a commercial rapid thermal processing (RTP) system for manufacturing display glass. The glass RTP system uses bulb-type tungsten-halogen lamps for heating. A highly uniform temperature distribution with a maximum temperature difference of less than 6 °C between any two points is required during glass processing, regardless of the size of the glass. In designing a decentralized controller that satisfies this requirement, the problems of optimal sensor location, lamp grouping, optimization-based controller tuning, and time-varying bias input signals were identified and solved. The performance of the controller was investigated experimentally and numerically.

Multivariable fuzzy controller under Gödel's implication

The oxidation and nitridation processes of niobium films in a rapid thermal processing (RTP) – system were investigated. 200 and 500 nm niobium films were deposited via sputtering on sapphire-(1-102)-substrate. At first niobium films were oxidized in molecular oxygen at temperatures ranging from 350 to 500 °C and for times of 1, 2 and 5 min and then nitridated in ammonia at 1000 °C for 1 min using an RTP system. For characterisation of the niobium films complementary analytical methods were used: X-ray diffraction (XRD) for phase analysis, secondary ion mass spectrometry (SIMS) for determining the elemental depth profiles of the films, scanning electron microscopy (SEM) and atomic force microscopy (AFM) for characterisation of the surface morphology of the films. The influence of the substrate, single crystalline sapphire, on the reactivity of the niobium films was studied in dependence of temperature, time of reaction and film thickness. The possibility of existence of niobium oxynitride phase was investigated. According to XRD and SIMS data, there is evidence that an oxynitride phase is formed after oxidation and subsequent nitridation in the bulk of some Nb films. In some of the experiments crack formation in the films or even delamination of the Nb films from the substrates was observed.

We consider the recursive identification and feedback control of the rapid thermal processing (RTP) system where the asymmetry of the convective heat transfer coefficient, caused by gas flow in the chamber, is taken into consideration. The success of RTP depends on the precise control of water temperature by adjusting the strength of the heating lamps to minimize dopant redistribution as well as wafer warpage. An efficient method of recursive identification and optimal feedback control is developed by separating the radiation field from the temperature field of the wafer and converting the heat conduction equation for the wafer temperature to a reduced-order model by means of the Karhunen-Loève Galerkin procedure.

The purpose of transfer learning is to utilize the knowledge gained from the existing (source) domain to enhance the performance on a distinct but related (target) domain. Existing works on transfer learning are not capable of optimizing different quality measures (components) such as minimizing the marginal distribution, minimizing the conditional distribution, maximizing the target domain variance, modeling the manifold by utilizing the common geometric properties in the source as well as the target domain at the same time. Moreover, existing transfer learning methods use conventional approaches to determine the appropriate values of their parameters, which is very hectic and time-consuming. Therefore, in order to overcome the drawbacks of existing approaches, we propose a Particle Swarm Optimization based Parameter Selection Approach for Unsupervised Discriminant Analysis (UDATL-PSO) in transfer learning framework. In UDATL-PSO, all the quality measures are considered at the same time, as well as the PSO approach has been used to select the best values of their parameters. Extensive experiments on various transfer learning tasks show that the proposed method has a significant influence on state-of-the-art methods.

Rapid and potentially low-cost process techniques are analysed and successfully applied towards the fabrication of high-efficiency mono-and multicrystalline Si solar cells. First, a novel dielectric passivation scheme (formed by stacking a plasma silicon nitride film on top of a rapid thermal oxide layer) is developed that serves as antireflection coating and reduces the surface recombination velocity (S{sub eff}) of the 1.3 {omega}-cm p-Si surface to approximately 10 cm/s. The essential feature of the stack passivation scheme is its ability to withstand short 700 - 850degC anneal treatments used to fire screen printed (SP) contacts, without degradation in S{sub oeff}. The stack also lowers the emitter saturation current density (J{sub oe}) of 40 and 90 {omega}/{open_square} emitters by a factor of three and 10, respectively, compared to no passivation. Next, rapid emitter formation is accomplished by diffusion under tungsten halogen lamps in both belt line and rapid thermal processing (RTP) systems (instead of in a conventional infrared furnace). Third, a combination of SP aluminium and RTP is used to form an excellent back surface field (BSF) in 2 min to achieve an effective back surface recombination velocity (S{sub eff}) of 200 cm/s on 2.3 {omega}-cm Si. Finally, the above individual processes are integrated to achieve: (1) >19% efficient solar cells with emitter and Al-BSF formed by RTP and contacts formed by vacuum evaporation and photolithography, (2) 17% efficient manufacturable cells with emitter and Al-BSF formed in a belt line furnace and contacts formed by SP. (Author)