Laser Diode Cooling


Introduction

Cooling systems are necessary components in a wide variety of advanced devices.  These devices regulate the heat dissipation which is a common and detrimental produce of power input.  Today's cooling systems  take advantage of convection, conduction and/or radiation to move heat efficiently away from the heat generator.  Commercial systems available today fall into the following categories: 

Laser diodes are made of semiconducting materials which produce very high heat loads.  Optical efficiencies of these diodes are very highly temperature dependent, so precise temperature control is necessary.  Conventional cooling systems for high-power diode arrays typically use liquid cooled approaches to maintain the diode temperature near room temperature.


This project explores the thermal properties of a cooling package designed specifically for laser diode cooling.  If successful, my research will deliver a package having the following advantages over today's commercially available systems:

  1. Space efficiency - The system requires far less room.

  2. Very few moving parts - With the exception of a fan, this system has no moving parts.  Existing systems often require pumps, fans, and other components which greatly reduce the working lifetime of the system.

  3. Control - The system can be easily controlled and adjusted by a variable DC source.

  4. Maintenance - Less moving parts means maintaining this system is quite easy.


Difficulties

Laser diodes are found in high-output pulsed and continuous mode lasers.  Heat is generated as side-effect of the optical production.  With arrays of up to 25 individual diodes mounted on a single bar, bars can currently output up to 100W thermal power.  With the size of diode bars on the order of millimeters, this thermal generation results in heat fluxes up to 107 W/m2. 


Additionally, the optical efficiency of the diode is extremely temperature dependent, with optimum performances near room temperatures. Considering the high flux and relatively low temperature requirements, conventional heat sink cooling systems have proved inadequate.  Advanced cooling systems incorporate liquid cooling and cryogenic systems.  These systems are often large and bulky, and can easily break down due to a large number of moving parts.  Additionally, low working temperatures and liquid environments can easily introduce moisture into the system or neighboring components which lead to detrimental thermal or electrical conditions.

To put the scope of the problem in perspective, consider the current cooling solution for computer chips.  A 3 GHz Pentium IV computer chip generates 70-100 Watts of heat over a 35mm X 35mm footprint, requiring dissipation of about 8X104 W/m2.  The specifications for this cooling package require dissipation of 6X106 W/m2, or nearly 100 times the flux found in today's computer heat sinks.


In this project, a steady-steady state laser diode cooling package is characterized numerically and validated experimentally.  The package is designed to dissipate the heat generated by a 100 Watt diode bar which produces 60 Watts of heat.  The cooler is a multistage device which makes use of advanced low resistance materials, a thermoelectric cooler and a heat sink.


Numerical Analysis

Extensive 2D simulations have been performed to fully characterize the working conditions of the cooler.  Numerical simulations were done using the finite element package contained in the software package FIDAP by FLUENT Inc.  The numerical study can be broken into two parts: (1), the heat transfer and (2) the stresses induced due to mismatches between the coefficient of thermal expansion.

Heat Transfer Analysis

The 2D mesh was generated using GAMBIT by FLUENT.  The mesh was optimized to least number of nodes while preserving the accuracy of the result, at a specified convergence.  The optimized mesh was fed into FIDAP, along with the appropriate boundary conditions and material properties.



An adiabatic boundary condition was applied to all surfaces in contact with the air.  Although natural convection occurs, the heat loss to the surrounding air is negligible.  Along the bottom surface of the copper, two different boundary conditions were applied, depending on the analysis being done for the particular simulation run.  One boundary condition applied was a  large convective condition.  The other applied on a different run was a constant temperature, the value obtained from an experimental run.

Design modifications have been done using the numerical model such than the temperature along the bottom surface of the copper are uniform to within 1.5 Celsius.  Inclusion of the diamond in the design has shown to increase efficiency by about 35% over a conventional, copper only heat sink.

Thermal Stress Analysis

Due to mismatches within the Coefficient of Thermal Expansion, stresses are produced within the cooling package and in the diode during operation.  FIDAP uses the results from the temperature calculations it makes, and uses them to calculate the amount of expansion or contraction which will occur in each material.  Because the materials are bonded together, and thus cannot move freely, stresses result at the interfaces.  Thus, the thermal stress is a function of the CTE mismatch at the interface and deviation of temperature from the state of zero stress.



The highest predicted stresses will occur in areas of red.  The data indicates failure within the silicon or diamond, more likely the silicon due to the lower strength of this material.  Von Mises stress criterion are calculated within FIDAP.



Experimental Work

Fabrication of the cooling system was done by a commercial manufacturer to the specifications based on optimized numerical results.  An experimental testing platform was manufactured in the POF Lab.  This platform gives great control over placement of was completed and the results indicated behavior which correlates well with the numerical predictions.