· · · Repository hosted by TU Delft Library

ABSTRACT: The power of LEDs increases exponentially over the years,while the mean time to failure (MTTF) should remain >100000 hours. The reliability requirement limits the junction temperature and the thermo elastic stresses, which are roughly inversely proportional tothe thermal conductivity of the heat spreaders. The 3omega method has been set up to measure dynamically the thermal conductivity of anisotropic thin layers preferably in situ. CONCLUSIONS: The 3omega method measurement set up is operational from 10 Hz to 10 kHz The measured thermal conductivity of silicium deviates 50%, so the set upneeds to be improved. The length of the heater has already been checked. The goodness of fit of the sensor/heater calibration R² = 0.99998 The goodness of fit of the 3omega slope method R² = 0.99

LEDs are strong candidates for future illumination applications dueto their much lower consumption of energy compared to conventional lighting options. One of key problems in development of LEDs is successful thermal management during illumination. Therefore, current research ongoing related to high power LEDs is focusing on improvementof cooling performance of them to enhance light output efficiency, durability and reliability of these devices.The goal of current studyis a first characterization of the LEDs arrays cooling with microchanneled flows. The flow and heat transfer characterization of a 5 LED array on top of water driven microchannels are performed and the junction temperature change with flow rate and thermal resistance isstudied. Experimental results show that the microchannel cooler reduces the junction temperature of LED array and improves the heat dissipation capability of the LEDs themselves.X At very low Reynolds numbers, there are discrepancies found between literature data and ourexperimental results. These discrepancies are explained by uncertainty in general measurements at microscale, including geometrical dimensions and operating parameters. It is demonstrated that the uncertainty in f*Re is dominated by the microchannel width and height measurements. Even a very accurate pressure drop measurements are oftenovershadowed by the geometry measurement uncertainties. In addition,the channel geometry is often not exactly the desired geometry dueto manufacturing difficulties associated withsmall scales. Therefore, the flow behavior is different from the expectations based on theory. Nusselt number variation with Reynolds number is investigated. Nusselt number is almost linearly increasing with Reynolds number inthe low Reynolds number range as also observed in literature. In fact, Nusselt number is expected to be constant in fully developed laminarflow according to the classical laminar flow theory. This does not hold for several reasons such as entrance region effects, unclearboundary conditions in microchannel cross section, multi directional heat transfer, flow configuration as also explained in details previously.

Many photovoltaic cells and OLEDs are covered by cover glass or foil to protect the device against oxygen and moisture. The cover glassor foil is semitransparent for infrared radiation and conducts the heat and blurs lock-in thermograms. In lock-in thermograms of OLEDs with cover glass short circuits raise typically 10 mK in temperature,where as literature and observations suggest that polymers decompose into gas at way higher temperature behind the glass. A good interpretation of lock-in thermograms may be done by the understanding of the heat conduction and radiation through cover glass. Modelling of heat conduction and radiation around a short circuit in cover glass will be presented as well as first validation experiments where a short circuit has been imitated by a laser.

The dielectric constant of a metallization system can be reduced drastically by introduction of air gaps between metal lines. It was the objective of the activities reported here to demonstrate processes that allow to incorporate air gaps in single-metal layer structures. Two different methods were used for this purpose: Philips favoreda "sacrificial materials" approach where the dielectric material between the metal lines was removed underneath a porous hard mask layer by thermal decomposition; the air gaps of Infineon's "close-off" approach were etched between metal lines by RIE and subsequently sealed at the top by a novel selective oxide deposition. In both cases,air gaps could be created successfully and reproducibly. The Philips air gaps are less than 150 nm wide. Preliminary electrical measurements indicate that the capacitance between adjacent metal lines indeed can be reduced to a value that is compatible with the requirements of the ITRS roadmap for the 45 nm node. The reliability performancewill be evaluated in the remaining part of the project.

The market of LEDs increases exponentially over the years as well as their power. However, at the same time the mean time to failure should remain superior to 100,000 hours. The major difficulty comes from the mechanical stress in the LED structure, which is roughly proportional to the inverse of the thermal conductivity. Consequently, it is really important to determine thermal conductivity of the materials used in LEDs. In the same way, the thermal conductivity of films is lower than the thermal conductivity of the bulk material. Unfortunately, thermal conductivity measurement is difficult on two-dimensional structures. Then an appropriate method had to be developed. The 3 omega thermal conductivity measurement method has been used extensively to measure the thermal properties of bulk and thin film dielectric materials. Tests on different materials will allow to find materials with a higher thermal conductivity. The thermal conductivity measurement will be elaborated by two different methods defined the isotropic and anisotropic properties, the "slope method" and the 2D heat conduction.