Approximately 70% of failures are due to thermal loads on components.
The failure rates of most components can double with a 20C increase in the junction temperature operating at about 50% of their rated power.
Electrical performance of many components is temperature dependent.
Thermal Acceleration Factor for Typical IC Devices
Hotter, Smaller Packaging
Power densities continue to increase due to: • Larger chip sizes • Increased dissipated power • Smaller packaging technology
First SIA Roadmap predicts: • 40 W/die by 2001- already there • 200 W/die by 2007
Temperature limits have remained the same: • Maximum junction temperatures of 150175C. • Reliability requires lower junction temperatures.
Integrated Thermal Design
Successful design requires adequate thermal management from: • Junction to Case or case to substrate • Substrate (or heat sink) to External Cooling Source
Any one weak link in the thermal path can ruin the design.
Many times thermal management is an integral part of the structural design of the package.
Late changes to correct thermal problems can require costly redesign.
Methods of Heat Transfer
Conduction • Thermally conductive materials are used to transport heat from source to sink. • Heat is transferred atom to atom through a material.
Convection • Working fluid transports heat from source to sink. • Heat is transferred by both diffusion and advection.
Radiation • Heat is transferred by electromagnetic waves through liquids, gases, and vacuums.
Conduction Heat Transfer
Heat Input
Electrical/Thermal Conductivity
Heat Source T1
Area A
A) Thermal Flow
Area A
B) Electrical Flow
Thermal Conductivity of Various Materials
Effect of Temperature on Thermal Conductivity
Heat Transfer Methods
Five Material Construction with Electrical Analog
Parallel Heat Spreading
Effect of Heat Spreading
Spreading angle based on ratio of thermal conductivities of current layer to underlying layer. Often approximated as 45 o for rapid analysis
Equivalent Thermal Resistance Network
Thermal Impedance
RT ( also called Θ)= L/KA or 1/K ∫ dl/A where A is the surface of any crosssection of heat flow, L is the length and dl is the incremental length along the heat path ( conduction) RT = 1/hAs , where h is the heat transfer coefficient and A is the area (convection) Θjc = junction to case Θcs= case to sink Θja= junction to ambient
Convection Heat Transfer
The relationship between heat transfer and surface temperature is described by the Newtonian cooling law: Convective heat transfer can be determined by using numerical simulation of fluid flow and energy transfer Convective heat transfer is usually characterized by the Nusselt number, Nu=hL/k The heat transfer coefficients , h, can be determined for a number of conditions
Geometric and Flow Effects on Heat Transfer
Vertical plate height, H
Horizontal plate, heat transfer in upward direction
Horizontal plate, heat transfer in downward direction
Convective Heat Transfer Equations
Heat Sinks
Typical heat sink (Aavid 583068) for component mounting. Used in natural convection and forced conconvection, either in laminate flow (less that 180 feet/min flow) or turbulent flow. Most heat sinks commercially available have been characterized for natural and forced convection parameters
Heat Sink Performance
Heat Sink with natural convection
Heat Sink with Forced Convection (turbulent flow)
Fin Efficiency and Design
Efficiency = Actual heat transfer = ηf Heat which would be transferred if the entire fin were at the base temperature Important considerations in fin design • fin to fin spacing and number • height and surface area for convection • Thermal contact resistance • Heat transfer coefficients • Weight and other materials aspects
Heat Pipes
Heat Pipes allow large quantities of heat to be transported over long distances. Heat input vaporizes working fluid at evaporator, with condensation at other end. The working fluid returns by the action of a wick. By capillary action. Effective thermal conductivies approaching 10,000 W/moK
Radiation Heat Transfer
The transfer of radiative energy from a surface to ambient follows the Newtonian cooling relationship, with an added multiplier of a “gray body” shape factor. The radiation heat transfer coefficient follows the relationship: These two equations show that , to a first order, radiation follows a T4 relationship. This is predicted by black body radiation
Temperature Difference for Various Heat Transfer Methods
Thermal Analysis Techniques/Tools
Successful thermal analysis of electronic systems can be performed using one or a combination of techniques: • • • • •
Empirical data and correlation Closed form analytical solutions Finite difference analysis Finite element analysis Computational fluid dynamics
Thermal analysts will combine many of the techniques.
Analytical Solutions
Can provide accurate estimations for conduction and convection problems when the thermal paths are highly one dimensional.
Thermal Resistance Concept: • Relatively simple algebraic formulations can rapidly solve many problems and perform preliminary analysis.
Finite Difference Analysis can be developed for problems involving conduction and convection when: • Thermal paths are interdependent and multidimensional. • Correlations are available for convective transport. • Flow dynamics can be estimated reliably.
Analytical Solutions
Advantages: • “Quick and Dirty” estimates can be performed rapidly. • Low cost, requires basic PC’s and software.
Disadvantages: • Can become highly mathematical. • Have a practical limitation as to design complexity.
Empirical Correlations
Many correlations are available for very complex objects such as finned heat sinks, card cages, typical memory cards.
Correlations allow for convection effects to be incorporated relatively easily in analytical solutions.
Provide rapid solutions allowing for optimization and “what-if” scenarios.
Numerical Modeling
Finite Element Analysis is applicable for complex designs where: • Thermal paths are three dimensional. • Thermal interaction between components components.
Computational Fluid Dynamics is necessary when: • Convection is the primary method of heat transfer. • Flow dynamics are unknown and complex. • When fans and blowers are involved.
Numerical Modeling
Advantages: • Results can be made graphical for ease of interpretation. • Available codes can reliably solve very complex problems.
Disadvantages: • Software is generally expensive. • Engineers require training and experience. • Detailed modeling can be time consuming.
