Selection of measurement locations for the control of rapid thermal processor

doi:10.1016/S0005-1098(99)00197-1

Automatica

Volume 36, Issue 5, May 2000, Pages 705-715

https://doi.org/10.1016/S0005-1098(99)00197-1 Get rights and content

Abstract

Rapid thermal processing (RTP) offers a growing potential in integrated circuit manufacturing as the feature sizes in ULSI move further down into the subhalf micrometer technology. The main problem of RTP control lies in temperature uniformity. In this work, process characteristics of RTP systems are explored and difficulties in control are explained. Two important factors have to be taken into account in the RTP temperature control. They are: temperature uniformity across the wafer and the controllability of the resultant control structure. Measurement selection criteria are devised to achieve both these objectives. The conventional equal-spaced measurement selection criteria in general cannot guarantee temperature uniformity. A nonlinear measurement selection criterion is proposed to overcome this problem. An alternative procedure is also explored to achieve both objectives by varying the temperature set points. Once the control structure is determined, internal model control (IMC) principle is employed for the design of multivariable controllers. As a result of ramp inputs, this leads to a PID type of controller with double integrators: a PI²D controller. Nonlinear compensation is also implemented to deal with a wide range of set point changes. Simulation results show that a 42% improvement in temperature uniformity can be achieved by using the proposed design procedure.

Introduction

The continuing downscaling of design features in ULSI circuits places a challenging demand on the semiconductor manufacturing technology. Rapid thermal processing (RTP) offers a growing potential as one moves further into the subhalf micron technology (Roozeboom & Parekh, 1990). RTP performs single wafer thermal process operations including: annealing, oxidation, nitridation, chemical vapor deposition and cleaning (Gyurcsik, Riley & Sorrell, 1991; Roozeboom, 1992). Furthermore, RTP possesses the feature to significantly reduce the thermal budget while affording single-wafer granularity and cluster compatibility. Despite all the promises, the major obstacle to wide-spread applications is: inadequate temperature measurement and control capabilities for applications more critical than current silicides (Roozeboom, 1992; Badgwell, Breedijk, Bushman, Butler & Chatterjee, 1995). That means: maintaining temperature uniformity over a range of processing conditions is critical for the acceptance of RTP. In a rapid thermal processor, tungsten-halogen lamps are arranged in linear (Gyurcsik et al., 1991), square or pseudo-ring formation (Sorrell, Fordham, Öztürk & Wortman, 1992; Apte & Saraswat, 1992) and, typically, multiple banks of lamps are also employed (Roozeboom, 1992; Badgwell et al., 1995). The powers of the lamps are manipulated to control the wafer temperature during the RTP cycle.

The last decade has seen advances in the modeling, design and control of RTP. The importance of design for better temperature control was recognized by several researchers (Norman, 1992; Cho, Paulraj, Kailath & Xu, 1994). Pseudo-ring lamps arrangement (Badgwell et al., 1995), placement of radiation shield (Lord, 1988) and design of reflectors (Sorrell et al., 1992) are proposed to overcome temperature non-uniformity. Design parameters such as: chamber geometry, the lamp number and location, the reflector characteristics and wafer rotation speeds are explored by Dilhac, Ganibal, Bordeneuve and Nolhier (1995). Based on linear programming, a systematic procedure to determine the optimal lamps arrangement is proposed by Cho et al. (1994). The design and optimization steps are often carried out on first principle models (Lord, 1988; Gyurcsik et al., 1991; Cho et al., 1994; Merchant, Cole, Knutson, Hebb & Jensen, 1996).

The second factor that affects the temperature uniformity is inadequate temperature control. The multivariable nature of temperature control is addressed properly by Gyurcsik et al. (1991) and by Apte and Saraswat (1992). Multiloop PID control (Dilhac, Ganibal, Bordeneuve & Nolhier, 1992) and Internal Model Control (Schaper, Moslehi, Sarawat & Kailath, 1994) are proposed for the multi-zone RTP systems. Interaction and robustness analyses of multivariable temperature control are also explored by Schaper et al. (1994) and Edgar and Breedijk (1994). Gain scheduling is also employed to compensate process nonlinearity as the temperature set point changes (Schaper et al., 1994; Edgar & Breedijk, 1994). Control techniques such as long-range predictive control (Bordeneuve, Najim & Ganibal, 1991) and adaptive control (Guibe, Dilhac & Dahhou, 1992; Djebara, Dahhou, Babary, Khellaf & Ganibal, 1993; Morales, Dahhou, Roux & Babary, 1995) are also proposed for the control of RTP systems. As mentioned earlier, two possible ways to overcome temperature non-uniformity are: (1) design for better uniformity and (2) improved control algorithms to maximize uniformity. However, little is said about the selection of temperature measurement locations to improve temperature uniformity. The temperature distribution in an RTP can be viewed as a distributed parameter systems (Waldraff, Dochain, Bourrel & Magnus, 1998). Another well-known example is the inferential control in distillation column (Luyben, 1992; Doyle III, 1998). Such an analogy can be extended to the RTP control. However, it should be emphasized that the control objective plays a vital role in the selection of measurement locations. In an RTP system, the temperature profile is not uniform across the wafer and it is critical to select temperature measurement locations such that the following two objectives can be achieved: (1) maintaining the desired temperature profile and (2) possessing good controllability.

The purpose of this work is to study multivariable temperature control of RTP systems and emphasis is placed on the selection of temperature measurement locations for better uniformity. The remainder of this paper is organized as follows. Section 2 addresses the process characteristics and design aspects of RTP system. Procedures for the selection of temperature measurements are discussed in Section 3. PID type of controllers is proposed and nonlinear compensation is addressed in Section 4. Simulation results and an alternative control strategy are presented in Section 5 followed by conclusion.

Section snippets

Process

In this work, an axisymmetric RTP chamber is studied. A schematic diagram of the RTP system is shown in Fig. 1. In this system, powers are supplied to three rings of tungsten-halogen lamps. Energy is transferred through a quartz window onto a thin semiconductor wafer via direct and reflective paths.

The RTP system is a typical heat transfer process. However, unlike conventional chemical engineering systems, this is a radiation dominant heat transfer process. Lord (1988) is among the first to

Selection of measurement locations

As mentioned earlier, the design of the heating source poses an inherent limitation on the performance of RTP systems. In terms of control system analysis, this only constitutes part of a control structure. The powers of the lamps ( $P_{j}, j=1,…,m$ ) act as the manipulated inputs to a control system. In other words, the “design” only completes one aspect of the RTP temperature control. The other aspect is to select temperature measurements as controlled variables (outputs). A control structure is

Control system design

In this work, internal model control (IMC) principle (Morari & Zafiriou, 1989) is employed to design the PID type of multivariable controller. Before getting into the details of controller design, it is important to recognize a typical temperature cycle for RTP (e.g., Fig. 5). This cycle shows a temperature ramp to 900°C in 10 s followed by a process hold for 30 s at 900°C and, then, process is cool down in another 20 s. In this cycle, ramp up and cool down portion takes up 50% of the batch time

Results

Consider a typical RTP temperature cycle (e.g., Fig. 5) with a standard 50°C/s ramp from 400° to 900°C and a 30 s hold at 900°C followed by a 40°C/s cooling. For the control structure selected by equal-spaced method, T₁−P₁,T₁₆−P₂ and T₃₀−P₃, with a filter time constant τ_f=1, Fig. 5 shows that good set point responses can be achieved. A more important measure of the temperature uniformity is the averaged deviation of all temperatures $(T_{1} ∼T_{30}), Δ T_{avg}$ . Fig. 6B shows that the standard deviation of all

Conclusion

In this work, the importance of selection of temperature measurement locations is analyzed and the results indicate that the resultant control structure (from temperature measurements) imposes an inherent limitation on the temperature uniformity. A measurement selection criterion is proposed to achieve both objectives. Moreover, an alternative procedure is also explored by setting the temperature according to the desired temperature profile. Once the control structure is determined, IMC

Acknowledgements

This work is supported by National Science Council of Taiwan.

Chi-Jay Huang received the B.S. and M.S. degrees from National Taiwan University of Sci. Technol. in 1994 and 1997, respectively, both in chemical engineering. Since 1999, he has been with process engineering department of CTCI Corporation in charge of process design. His research interests include: process design and control of microelectronic processes.

References (32)

T.A. Badgwell et al.
Modeling and control of microelectronics materials processing
Computers in Chemical Engineering
(1995)
J.W. Chang et al.
The relative gain for non-square multivariable system
Chemical Engineering Science
(1990)
J.B. Guibe et al.
Adaptive control of a rapid thermal processor using two long-range predictive methods
Journal of Process Control
(1992)
S.H. Shen et al.
Indirect feedforward control: Multivariable systems
Chemical Engineering Science
(1992)
W. Waldraff et al.
On the use of observability measures for sensor locations in tubular reactor
Journal of Process Control
(1998)
P.P. Apte et al.
Rapid thermal processing uniformity using multivariable control of a circularly symmetric 3 zone lamp
IEEE Transactions on Semiconductor Manufacturing
(1992)
J. Bordeneuve et al.
Long-range predictive control of a rapid thermal processor
Internatioanl Journal of System Science
(1991)
Y.M. Cho et al.
Model identification in rapid thermal processing system
IEEE Transactions on Semiconductor Manufacturing
(1993)
Y.M. Cho et al.
A contribution to optimal lamp design in rapid thermal processing
IEEE Transactions on Semiconductor Manufacturing
(1994)
Crowley, J.L., DeBolski, T.J., Kermani, A., & Lassing, S.E. (1988). US Patent,...

J.M. Dilhac et al.

Temperature control in a rapid thermal processor

IEEE Transactions on Electronic Devices

(1992)

J.M. Dilhac et al.

Thermal modeling of a wafer in a rapid thermal processor

IEEE Transactions on Semiconductor Manufacturing

(1995)

K. Djebara et al.

Partial state model reference adaptive control of a rapid thermal processor

International Journal of Adaptive Control Signal Process

(1993)

Downs, J.J., & Moore, C. (1981). Steady state gain analysis for azeotropic distillation. Proceeding of the JACC,...

F.J. Doyle III

Nonlinear inferential control for process applications

Journal of Process Control

(1998)

Edgar, T.F., & Breedijk, T. (1994). Overview of process control issues in rapid thermal processing. Proceedings of...

Cited by (29)

Combined iterative learning and delta-operator adaptive linear quadratic Gaussian control of a commercial rapid thermal processing system
2017, Chemical Engineering Science
Citation Excerpt :
There are two important requirements for an RTP system: (1) a high degree of temperature uniformity over the wafer surface for combined time-varying and constant target temperature trajectories and (2) a short processing time, leading to rapid heating and subsequent fast cooling. Many research groups have presented various types of control methods, including decoupling control (Balakrishnan and Edgar, 2000), proportional-double-integral-derivative control (Huang et al., 2000; Lee et al., 2016), iterative learning control (ILC) (Cho, 2005; Cho et al., 2005a; Cho et al., 2005b; Lee et al., 2001; Yang et al., 2003), adaptive control (Choi et al., 2003), internal model control (Ebert et al., 2010; Schaper et al., 1999), linear quadratic Gaussian (LQG) control (Cho and Gyugyi, 1997; Won et al., 2009), and model predictive control (Dassau et al., 2006; Gwak and Masada, 2008; Jeng and Li, 2012; Jeng and Chen, 2013; Takagi and Liu, 2011), over the past two decades. However, as the wafer size increases and the surface patterns become finer and more complicated, the need for a new control technique that can meet the more refractory objectives of modern RTP systems, i.e., more stringent temperature uniformity and rapider ramping, arises.
Rapid thermal processing (RTP) units are used for various semiconductor fabrication steps mainly due to the short processing time, which results in a low thermal budget and improved product quality. In this study, we propose a comprehensive control system for the RTP unit by combining a linear quadratic Gaussian (LQG) controller, a constrained iterative learning controller (ILC), and a model parameter estimator. The control system is developed to resolve the issues caused by a short sampling time, which is required to meet the refractory control objectives of the modern RTP system (e.g., rapid ramping and a high degree of temperature uniformity). LQG control is selected because most required computations are conducted offline prior to each run and only minor computations are conducted in real time. The LQG control is constructed in a Smith predictor configuration so that a delayed model can be easily incorporated. The LQG control algorithm is modified from the standard formulation to include a quadratic penalty term for input deviation from the average value, to restrain the inputs from violating their constraints in real-time operation. The control algorithm and process model are developed in delta form so that the effect of the truncation error on the model accuracy can be alleviated at a high sampling rate. The constrained ILC updates the feedforward signal of the LQG controller in a batch-wise manner to further increase the temperature uniformity and to ensure that the inputs satisfy their constraints. To overcome model discrepancy from an actual system (patterned wafer), a small number of tunable parameters that can modify the major characteristics of the process model are incorporated into the identified state-space model (bare wafer) and are updated by a model parameter estimator. The performance of the proposed control system is validated in a numerical process, which is precisely constructed through identification experiments in a commercial 12-inch RTP unit.
Design and experimental implementation of an effective control system for thin film Cu(InGa)Se<inf>2</inf> production via rapid thermal processing
2016, Journal of Process Control
Citation Excerpt :
While this may be appropriate for controller design, an effective observer demands a higher-fidelity distributed parameter model if it is to provide accurate estimates of surface temperature. It is easy to show that the presence of the double integral endows this controller with the ability to eliminate any steady state offset in tracking ramped set points (c.f. [26,39]). Two alternative approaches for eliminating offset during temperature ramps may have be considered.
Copper indium gallium diselenide, Cu(InGa)Se₂ (CIGS) solar cells have achieved efficiencies of 22.3% at the cell level and 17.5% at the module level. CIGS-based modules are also in the early stages of commercialization, with >1 GW annual production capacity. The most common method for producing CIGS in industry is via precursor reaction, which consists of depositing Cu-In-Ga precursor films and reacting them with gas-phase H₂Se at 450–550 °C for 60 min or longer, and is commonly called selenization. Recently, interest has been growing in selenization by Rapid Thermal Processing (RTP), which is characterized by rapid temperature ramping, approximately 550 °C or higher temperature reactions, and improved process throughput. However, it has been difficult to commercialize RTP for CIGS film production in part because implementing a rapid, linear temperature ramp in the reacting thin film is complicated by two intrinsic process characteristics: (i) the temperature of the CIGS film cannot usually be measured directly; and (ii), the process is significantly nonlinear due to the dominance of radiative heat transfer at high temperatures. In this paper, we present the design and modeling, construction, and successful operation of a pilot-scale RTP selenization reactor utilizing a novel temperature control system.
Our two-fold approach to the unique temperature control challenges involves the design and implementation of (i) a first-principles, model-based observer to estimate the desired surface temperature; and (ii) a specialized controller to enable effective tracking of the desired linear temperature ramp set point. Our experimental results demonstrate that the control system is effective in tracking rapid temperature ramps accurately, with performance limited only by the physical constraints of the experimental system.
Improvement of temperature homogeneity of a silicon wafer heated in a rapid thermal system (RTP: Rapid Thermal Process) by a filtering window
2015, Applied Thermal Engineering
Citation Excerpt :
Various solutions have been deeply studied to obtain uniform wafer temperature as listed in Table 1, but they don't enable to achieve perfect temperature homogeneity and they usually remain very stringent [20]. Heating by several groups of lamps controlled separately allows getting uniform temperature at the surface of the wafer [9,10,21,22]. However, the mastering of the piloting requires a very thorough preliminary optimization and aging of the lamps is accelerated for the ones supplied with higher power.
Rapid thermal processes are used in various key stages in the microelectronics industry. In this study, the heat transfer in a rapid thermal system is modelled with the finite volume method. The influence of the radiative properties of the quartz window on the thermal profile of the silicon wafer is first investigated. The obtained temperatures are interpreted by analyzing the radiative properties according to wavelength and temperature. The wafer temperature profile for a non-optimized heating in steady-state is explained by a four-phase scheme where the radiative heat fluxes are depicted. From this scheme, a filter on the underside of the quartz window is envisaged to achieve temperature uniformity for the wafer. Two configurations are tested, one where the filter covers the entire lower surface of the quartz window and another where it is placed in a ring close to the reactor wall to confine the infrared radiations with wavelengths beyond 2.6 μm in order to raise the temperature at the edge of the wafer. Simulations demonstrate that the latter modification enables a more significant improvement of the wafer temperature homogeneity with less than 1% dispersion. The implementation of the filtering window is also discussed.
Control strategies for thermal budget and temperature uniformity in spike rapid thermal processing systems
2013, Computers and Chemical Engineering
Citation Excerpt :
They include model-based control (Balakrishnan & Edgar, 2000; Cho & Edgar, 2008), iterative learning control (Cho, Lee, Joo, & Lee, 2005; Choi & Do, 2001), adaptive control (Choi, Do, & Choi, 2003), internal model control (Schaper, Kailath, & Lee, 1999), and nonlinear model predictive control (Dasssu, Grosman, & Lewin, 2006). Huang, Yu, and Shen (2000) used a proportional–double integral-derivative (PI2D) controller to eliminate offset error during the heating step. Most studies focused on the soak RTP system, whereas only few studies have discussed the control of the spike RTP system (Emami-Naeini et al., 2003; Su et al., 2007).
Single wafer rapid thermal processing (RTP) is widely used in semiconductor manufacturing. A precisely applied thermal budget during RTP is crucial and relies on the temperature control of the wafer. However, temperature control in the RTP system with a spike-shaped temperature profile is a challenging task, and achieving perfect servo control is almost impossible because of the high temperature ramp-up/down rate and substantial nonlinearity of the process. This paper presents a novel method of control system design to provide a precise thermal budget in the spike RTP system. By tuning controller parameters and designing the set-point profile, the method targets thermal budget indices instead of temperature servo control. A nonlinear control strategy is proposed based on modeling the RTP system as a nonlinear Wiener model. Furthermore, a multivariable control structure is considered to maintain the temperature uniformity within the wafer. The simulation results show the effectiveness of the proposed control strategy and provide helpful guidelines for the design of a multivariable control configuration to achieve superior wafer temperature uniformity.
Control Strategy for Thermal Budget and Temperature Uniformity in Spike Rapid Thermal Processing Systems
2012, Computer Aided Chemical Engineering
Temperature control in the rapid thermal processing (RTP) system with spike-shaped temperature profile is a challenging task, and targeting perfect servo control is almost unachievable because of the high temperature ramp-up/down rate and significant nonlinearity of the process. This study presents a novel method of control system design to provide a precise thermal budget in the spike RTP system. A nonlinear control strategy is proposed based on modeling the RTP system as a nonlinear Wiener model. Further, multivariable control is considered to maintain the temperature uniformity within wafer. Simulation results demonstrate the effectiveness of the proposed control strategy, and provide guidelines for the design of the multivariable control configuration to achieve better wafer temperature uniformity.
Control relevant issues in semiconductor manufacturing: Overview with some new results
2007, Control Engineering Practice
The quality control of integrated circuit (IC) processing is becoming more and more important as the wafer becomes larger and the feature size shrinks. However, an advanced IC fabrication process consists of 300+ steps with scarce and usually difficult quality measurements. Thus product yield may not be realized until months into production while in-line measurements are available on the order of a millisecond. The series production nature and measurement setup lead to a unique process control problem. In this work, typical disturbances are explained and the possibility for inferential control is explored. This leads to a control architecture with multiple layers in a cascade structure. Next, the rapid thermal processing (RTP) is used to illustrate recipe generation and control structure design at the tool level. The resultant multivariable controller gives satisfactory setpoint tracking for a triangular-like temperature program. Effective delay in a feedback loop at the process level is also clarified which can be used to design a run-to-run controller or to prioritize the measurement queue for the metrology tool. In order to prolong the time between maintenance and to reduce rework, process trend monitoring of a tool is essential. Instead of using entire batch data, a key process variable is identified and an index is computed to capture dynamic behavior of the tool. An IC processing example is used to illustrate this approach and results clearly indicate that process trend is well predicted using the index-based time-series model. Finally, future research directions for improved semiconductor manufacturing are also described.

View all citing articles on Scopus

Cheng-Ching Yu received the B.S. degree from Tunghai University, Taichung, Taiwan, in 1979 and the M.S. and Ph.D. degrees from Lehigh University in 1982 and 1986, respectively, all in chemical engineering. Since 1986, he has been with the National Taiwan University of Sci. and Technol., where he is currently a professor of chemical engineering. His current research interests include: plantwide control, control of microelectronic processes and reactive distillation. Dr. Yu received the Young investigator award from the Chinese Institute of Chemical Engineers, and the Outstanding Research Awards from the National Science Council of Taiwan. He is on the editorial board of the J. Chin. Inst. Chem. Eng. and the J. Chin. Inst. Eng. and he is the author of Autotuning of PID Controllers (Springer-Verlag, 1999).

Shih-Haur Shen received the B.S. degree from Feng-Cha University, Taichung, Taiwan, in 1988 and the M.S. and Ph.D. degrees from National Taiwan University of Sci. Technol. in 1990 and 1995 all in chemical engineering. Since 1997, he joined Applied Material Taiwan Ltd., where he is currently a senior process engineer in the CMP department. He holds one patent and published over 20 technical papers. His current research interests include: designs and operation of RTP and CMP systems.

^☆: This paper was not presented at any IFAC meeting. This paper was recommended for publication in revised form by Associate Editor E. Ydstie under the direction of Editor S. Skogestad.

View full text

Brief PaperSelection of measurement locations for the control of rapid thermal processor☆

Abstract

Introduction

Section snippets

Process

Selection of measurement locations

Control system design

Results

Conclusion

Acknowledgements

Computers in Chemical Engineering

Chemical Engineering Science

Journal of Process Control

Chemical Engineering Science

Journal of Process Control

Rapid thermal processing uniformity using multivariable control of a circularly symmetric 3 zone lamp

IEEE Transactions on Semiconductor Manufacturing

Long-range predictive control of a rapid thermal processor

Internatioanl Journal of System Science

Model identification in rapid thermal processing system

IEEE Transactions on Semiconductor Manufacturing

A contribution to optimal lamp design in rapid thermal processing

IEEE Transactions on Semiconductor Manufacturing

Temperature control in a rapid thermal processor

IEEE Transactions on Electronic Devices

Thermal modeling of a wafer in a rapid thermal processor

IEEE Transactions on Semiconductor Manufacturing

Partial state model reference adaptive control of a rapid thermal processor

International Journal of Adaptive Control Signal Process

Nonlinear inferential control for process applications

Journal of Process Control

Brief Paper
Selection of measurement locations for the control of rapid thermal processor☆