There are numerous methods for measuring the temperature of an operating semiconductor device. The methods can be broadly placed into three generic categories: electrical, optical, and physically contacting. The fundamentals underlying each of the categories are discussed, and a review of the variety of techniques within each category is given. Some of the advantages and disadvantages as well as the spatial, time, and temperature resolution are also provided.

Temperature measurements of semiconductor devices - a review

The accurate determination of the channel temperature in field-effect transistors (FETs) and monolithic microwave integrated circuits is critical for reliability. An original accurate closed-form expression is presented for the thermal resistance of multifinger FET structures. The model is based on the solution of Laplace's equations in prolate spheroidal coordinates and elliptical cylinder coordinates. The model's validity is verified by comparing the results with finite-element simulations, and experimental observations from liquid-crystal measurements and spatially resolved photoluminescence measurements. Very close agreement is observed in all cases.

Accurate determination of thermal resistance of FETs

Photocurrent spectroscopy is applied for an analysis of both packaging-induced strain and strain-induced defect creation in InAlGaAs/GaAs high-power diode laser arrays. Strain profiles across 50 W diode lasers processed by different packaging procedures are measured and compared to model calculations. We demonstrate that packaging-induced strain giving rise to spectral shifts of the laser transition correlates with packaging-induced defects in the waveguide that are quantified via a sub-band gap absorption band. Packaging on a Cu–diamond multilayer heat spreader appears as optimized solution simultaneously minimizing strain and defect creation.

Simultaneous quantification of strain and defects in high-power diode laser devices

A new detailed approach has been developed, whereby the individual performances of the emitters comprising a laser bar are assessed separately as part of a larger coupled system. The results reveal the existence of a strain threshold, wherefore degradation of the emitter performances is only observed when the local packaging-induced strain is above a certain critical value. This degradation which occurs after aging, is manifested in an increase in the "apparent" threshold current and a decrease in the "apparent" slope efficiency of the individual emitters. This degradation is also revealed by the presence of facet defects as previously reported. These results suggest that measurement and control of packaging-induced strain will allow significant improvement in the degradation behavior of high-power AlGaAs laser bars.

Mounting-induced strain threshold for the degradation of high-power AlGaAs laser bars

Spatially resolved photoluminescence line scans were performed to determine the local stresses in AlGaAs laser diodes designed for high-power operation at 808 nm. In this approach, the sign and magnitude of the local stress are deduced from the spectral shift of the peak associated with band-to-band transitions in the n-type GaAs substrate. The sensitivity of the technique (minimal equivalent hydrostatic stress that can be detected) can reach 10 MPa or better. Correlations between solder-induced stress distribution in the devices and estimated lifetimes are demonstrated.

Microphotoluminescence mapping of packaging-induced stress distribution in high-power AlGaAs laser diodes

Spatially resolved photo-luminescence (PL) line-scans were performed in a specific optical micro-probe to determine the soldering-induced local stresses in GaAs/GaAlAs laser diode arrays designed for high-power operation at 808 nm. In this approach, the sign and magnitude of the local stress are deduced from the spectral shift associated with band-to-band transitions in the GaAs substrate. The sensitivity (minimal equivalent hydrostatic stress that can be detected) is better than 10 MPa. The spatial resolution of the micro-PL technique (of the order of 1 micrometer), together with the short acquisition times, allows for detailed investigations of the stress profiles along the whole laser bars with a large number of data points. Different aspects of the mechanical stress distribution at the various steps of the process could thus be revealed. Finally, correlations between solder-induced stress distribution and estimated lifetimes were established. In particular, > defects, which are known as a failure mechanism on this type of devices, were observed only on the laser bars for which the micro-PL indicates the strongest compressive stress. This leads to consider the micro-PL approach proposed here as a cost- effective screening technique for the high-power GaAs/GaAlAs laser diode arrays.

The microphotoluminescence (μ-PL) technique is proposed for mapping local stress distribution in GaAs/AlGaAs high power laser diode arrays (LDAs). This technique will be used to monitor the stresses that can be induced on the bars during the packaging process. We show herein that also a detailed study of the stress profiles that could exist in the bars before mounting and after aging can be achieved with this technique.

Packaging-induced stress distribution in high power AlGaAs laser diodes by photoluminescence mapping

In power FETs, channel temperature control is very important both in terms of output power and of operational lifetime, which is mainly determined by thermally activated degradation or failure mechanisms. This applies both for discrete devices and MMICs. The experimental approach currently relies on two methods, IR microscopy and liquid crystal thermography, but both are limited in terms of spatial resolution, causing them to fail when devices with sub-/spl mu/m dimensions are considered. Also, these methods require some type of coating to be investigated, either with a material of known emissivity or with a liquid crystal layer whose uniformity may be difficult to control. Thermal resistance calculations are performed to assess channel temperature and therefore the expected MMIC reliability. In this approach, analytical thermal resistance models and FEA descriptions have been proposed. Use of thermal resistance assumes uniform temperature throughout the circuit, which is in general untrue due to the detailed geometrical bounding conditions, and also because of process imperfections which may occur. FEA descriptions in general rely on an assumption about the heat dissipation zone dimensions, which is very difficult to check experimentally. The purpose of this paper is thus twofold: first, to demonstrate a new methodology for accurate measurement and mapping of local channel temperatures in GaAs-based FETs based on spatially resolved photoluminescence spectroscopy, and second, to discuss FET behaviour as a function of gate voltage and to show that using thermal resistance to characterize channel temperature during aging tests is insufficient.

E. Martin

Papers

Microphotoluminescence mapping of packaging-induced stress distribution in high-power AlGaAs laser diodes

Microphotoluminescence mapping of packaging-induced stress distribution in high-power AlGaAs laser diodes

Packaging-induced stress distribution in high power AlGaAs laser diodes by photoluminescence mapping

Temperature distribution in power GaAs field effect transistors using spatially resolved photoluminescence mapping

A new method for temperature mapping on GaAs field effect transistors