The growing packing density and power consumption of very large scale integration (VLSI) circuits have made thermal effects one of the most important concerns of VLSI designers The increasing variability of key process parameters in nanometer CMOS technologies has resulted in larger impact of the substrate and metal line temperatures on the reliability and performance of the devices and interconnections Recent data shows that more than 50% of all integrated circuit failures are related to thermal issues This paper presents a brief discussion of key sources of power dissipation and their temperature relation in CMOS VLSI circuits, and techniques for full-chip temperature calculation with special attention to its implications on the design of high-performance, low-power VLSI circuits The paper is concluded with an overview of techniques to improve the full-chip thermal integrity by means of off-chip versus on-chip and static versus adaptive methods

/pdf/thermal-modeling-analysis-and-management-in-vlsi-circuits-27gdag8m8n.pdf

Thermal Modeling, Analysis, and Management in VLSI Circuits: Principles and Methods

This paper deals with the identification of RC networks from their time- or frequency-domain responses. A new method is presented based on a recent approach of the network description where all response functions are calculated by convolution integrals. The identification is carried out by deconvolution (NID method). This paper discusses the practical details of the method. Special attention is paid to the identification and modeling of distributed RC networks, like the problems of conductive heat-flow. A number of examples make it easy to understand the operation and the capabilities of the NID method. Comparative considerations are given concerning the accuracy and expenses of the NID and the popular AWE (momentum-matching) methods.

Identification of RC networks by deconvolution: chances and limits

Miniaturization of electronic and optoelectronic devices and circuits and increased switching speeds have exasperated localized heating problems. Steady-state and transient characterization of temperature distribution in devices and interconnects is important for performance and reliability analysis. Novel devices based on nanowires, carbon nanotubes, and single molecules have feature sizes in 1–100 nm range, and precise temperature measurement and calibration are particularly challenging. In this paper we review various microscale and nanoscale thermal characterization techniques that could be applied to active and passive devices. Solid-state microrefrigerators on a chip can provide a uniform and localized temperature profile and they are used as a test vehicle in order to compare the resolution limits of various microscale techniques. After a brief introduction to conventional microthermocouples and thermistor sensors, various contact and contactless techniques will be reviewed. Infrared microscopy is based on thermal emission and it is a convenient technique that could be used with features tens of microns in size. Resolution limits due to low emissivity and transparency of various materials and issues related to background radiation will be discussed. Liquid crystals that change color due to phase transition have been widely used for hot spot identification in integrated circuit chips. The main problems are related to calibration and aging of the material. Micro-Raman is an optical method that can be used to measure absolute temperature. Micron spatial resolution with several degrees of temperature resolution has been achieved. Thermoreflectance technique is based on the change of the sample reflection coefficient as a function of temperature. This small change in 10−4–10−5 range per degree is typically detected using lock-in technique when the temperature of the device is cycled. Use of visible and near IR wavelength allows both top surface and through the substrate measurement. Both single point measurements using a scanning laser and imaging with charge coupled device or specialized lock-in cameras have been demonstrated. For ultrafast thermal decay measurement, pump-probe technique using nanosecond or femtosecond lasers has been demonstrated. This is typically used to measure thin film thermal diffusivity and thermal interface resistance. The spatial resolution of various optical techniques can be improved with the use of tapered fibers and near field scanning microscopy. While subdiffraction limit structures have been detected, strong attenuation of the signal reduces the temperature resolution significantly. Scanning thermal microscopy, which is based on nanoscale thermocouples at the tip of atomic force microscope, has had success in ultrahigh spatial resolution thermal mapping. Issues related to thermal resistance between the tip and the sample and parasitic heat transfer paths will be discussed.

Microscale and Nanoscale Thermal Characterization Techniques

The insulated gate bipolar transistor (IGBT) modules are getting more accepted and increasingly used in power electronic systems as high power and high voltage switching components. However, IGBT technology with high speed and greater packaging density leads to higher power densities on the chips and increases higher operating temperatures. These operating temperatures in turn lead to an increase of the failure rate and a reduction of the reliability. In this paper, the static and dynamic thermal behavior of IGBT module system mounted on a water-cooled heat sink is analyzed. Although three-dimensional finite element method (3-D FEM) delivers very accurate results, its usage is limited by an imposed computation time in arbitrary load cycles. Therefore, an RC component model (RCCM) is investigated to extract thermal resistances and time constants for a thermal network. The uniqueness of the RCCM is an introduction of the time constants based on the Elmore delay, which represents the propagation delay of the heat flux through the physical geometry of each layer. The dynamic behavior predicted by the thermal network is equivalent to numerical solutions of the 3-D FEM. The RCCM quickly offers insight into the physical layers of the components and provides useful information in a few minutes for the arbitrary or periodic power waveforms. This approach enables a system designer to couple the thermal prediction with a circuit simulator to analyze the electrothermal behavior of IGBT module system, simultaneously.

Thermal component model for electrothermal analysis of IGBT module systems

The paper presents a methodology for simulating the static and dynamic performance of integrated circuits in the presence of electro-thermal interactions on the integrated circuit die. The technique is based on the coupling of a finite element method (FEM) program with a circuit simulator. In difference to other known simulator couplings a time step algorithm is used. Its implementation into simulation tools is described. The thermal modeling of the die/package structure and the extended modeling of the electronic circuit is discussed. Simulation results which indicate the capabilities of the methodology for electro-thermal simulation are compared to experimental results.

Electro-thermal circuit simulation using simulator coupling

Due to severe thermal problems of today's VLSI integrated circuits the need for reliable and quick thermal, electro-thermal and logi-thermal simulation tools is increasing, In this paper, we discuss the latest advances in the SISSI package (simulator for integrated structures by simultaneous iteration) which is a tool developed originally for analog VLSI design. The improvements include electro-thermal ac and transient simulation and the consideration of the thermal voltage of Si-Al contacts. Furthermore, we introduce a new module of SISSI, LOGITHERM, which is aimed at the self-consistent logic and thermal simulation of large digital VLSI designs. The features of our simulator package are highlighted by simulation examples that are compared in most cases with measurement results.

Electro-thermal and logi-thermal simulation of VLSI designs

Thermal mapping with liquid crystal method

Electro-thermal simulation: a realization by simultaneous iteration

An efficient thermal simulation tool for ICs, microsystem elements and MCMs: the μS-THERMANAL

Integrated microsystems raise new problems in thermal simulation. The frequently used structures such as cantilevers, membranes have quite different heat transfer properties than the simple silicon cubes of conventional ICs. Furthermore numerous functions are realized on these structures based on thermal principle. Quick and correct thermal simulation of these structures is needed during the design process. The paper presents the (mu) S-THERMANAL thermal simulation tool which is capable to simulate cantilever, bridge etc. microsystem structures both in steady-state and in the frequency-domain case. The algorithms of the program, based on the Fourier method, are detailed in the paper and numerous examples illustrate the capabilities of the tool.

A. Csendes

Papers

Electro-thermal and logi-thermal simulation of VLSI designs

Thermal mapping with liquid crystal method

Electro-thermal simulation: a realization by simultaneous iteration

An efficient thermal simulation tool for ICs, microsystem elements and MCMs: the μS-THERMANAL

μS-THERMANAL: an efficient thermal simulation tool for microsystem elements and MCMs