On the accurate determination of the thermomechanical properties of micro-scale material: Application to AlSi1% chip metallization of a power semiconductor device

https://doi.org/10.1016/j.microrel.2009.02.012Get rights and content

Abstract

This paper presents a method and an experimental technique supporting the accurate measurement of the thermomechanical parameters (coefficient of thermal expansion (CTE), Young’s modulus, yield stress and internal stress) of a thin metallic film used in microelectronic chip fabrication. First the theoretical background of the method is presented along with the process of fabrication of bilayer microcantilevers. Then the experimental setup of measurement and the measurement results obtained on a thin aluminum layer are given and discussed. The accuracy of the method is demonstrated and enables us to observe mechanical property variations over a temperature range of 80 °C as well as metal hardening phenomena.

Introduction

Power semiconductor devices are made up of a plurality of materials that exhibit different thermomechanical properties and especially large mismatch of coefficients of thermal expansion (CTE) higher than one order of magnitude. As a result, even under normal operating conditions, these power devices are subject to temperature changes that generate high level of stress repeatedly causes mechanical breakdown. In this context thermomechanical simulations become key to life time prediction of power device but reliability of such thermomechanical simulations is critically linked to the accuracy of the mechanical properties of the device’s materials. For bulk material (copper lead frame, mold compound, silicon) these properties are well known or can be easily determined by standard techniques such as tensile test. On the contrary thin materials, like source metallization, are not easy to determine because standard techniques like uniaxial tensile testing cannot be used due to the micrometer scale dimensions of material. Moreover material properties of such thin deposited film are closely linked to deposition techniques and thermal treatments. For example, Young’s modulus (E) values reported in literature for aluminum with 1% silicon (AlSi1%) deposited by sputtering range from 30 GPa to 70 GPa [1], [2], [3], [4]. In order to overcome this issue, this paper presents a method and an experimental technique for the accurate determination of thermomechanical properties of thin materials that make up the power semiconductor. Results obtained on a 4 μm AlSi1% film deposited by DC magnetron sputtering are presented and discussed. Fig. 1 is a cross-sectional view of the AlSi1% metallization. One can see in Fig. 1 that AlSi1% film is a polycristalline material with a grain size of approximately 4 μm and with silicon precipitates mainly located at grain boundaries. This AlSi1% thin film is the source metallization of a power vertical MOSFET (Metal Oxide Semiconductor Field Effect Transistor) manufactured by Freescale Semiconductor. Wire bond lift off is the main failure mechanism caused by fatigue of power semiconductor devices thus the accurate determination of the thermomechanical properties of the AlSi1% layer is a mandatory step to investigate the reliability behavior of the interface between bonding wires and source metallization [5], [6].

Section snippets

Measurements of thin film thermomechanical properties, methods and techniques

Generally thin film’s thermomechanical properties are characterized by either a mechanical or a physical method and for each method can be used several techniques [7]. The most commonly used mechanical techniques are wafer curvature [8], nanoindentation [9], uniform pressure which induces membrane deflection [10], resonant frequency of Micro-Electro Mechanical System (MEMS) under forced harmonic vibration [11], deflection of released MEMS induced by temperature [12] or the electrostatic pull in

Theoretical background to temperature induced bilayer cantilever curvature

Consider a bilayer structure composed of an aluminum layer deposited on a single crystal silicon structure as shown in Fig. 2. The temperature induced bilayer cantilever curvature technique relies on the thermal expansion coefficient mismatch between two components of a sandwiched layer to provide displacement with a change in temperature. This effect is well known and can be modeled by the following expression [12] where α stands for CTE, E stands for Young’s modulus, e being the layer

Cantilever process of fabrication

Cantilever process of fabrication starts on a Silicon Over Insulator (SOI) wafer because the oxide buried layer is a key layer during the process of fabrication of microstructures. SOI wafers were selected to comply with industrial clean room requirement in terms of size, thickness and contamination. SOI wafer properties are listed in Table 1. The first fabrication stage is silicon epi layer growth to obtain a suitable silicon thickness according to the study presented previously. This is

Experimental setup

In this section, the measurement equipments and methodology are detailed. The aim of experiments is to monitor as accurately as possible the evolution of bilayer cantilever curvature with temperature. As a result there are two parameters, temperature and curvature, that need to be measured accurately. The temperature adjustment system of cantilever microstructures is shown in Fig. 6. It is composed of thermoelectric module that generates heat or cold to adjust the temperature of a copper mass

Experimental results for ΔT  30 °C

First set of experiments are done for a maximum temperature excursion of 30 °C in order to stay in the elastic domain of deformation as demonstrated in the next section. Cantilever profiles are measured first at room temperature and then after three successive temperature variations of 10 °C. Curvature variations are determined for each measurement using the method presented previously. According to Eqs. (1), (2) curvature evolution varies linearly with temperature excursion and therefore a

Experimental results for ΔT  30 °C

In this section, experiments deals with the same structures subjected to greater temperature excursion. The aim is to highlight the limitation of the method due to the variation the thermomechanical parameters of aluminum. First, measurements are done on cantilever with the highest ΔρT, i.e. S4 structures, because these structures exhibit the highest level of stress. Cantilever profiles are measured at room temperature, 50 °C, 80 °C and 100 °C and a linear regression is performed for each

Conclusion

The experiments presented in this paper have demonstrated the accuracy of the method and the experimental setup to characterize thermomechanical properties of a thin layer of AlSi1% used as source metallization in a power device. Young’s modulus and coefficient of thermal expansion were determined with only few percents of errors. The method limitations have been investigated and it has been demonstrated that plastic deformation of AlSi1% for ΔT > 40 °C is the main limitation. Yield stress of AlSi

Acknowledgements

The authors would like to thank the Regional Council of Midi-Pyrénées for financial support.

References (22)

  • J.J. Vlassak et al.

    A new bulge test technique for the determination of Young’s modulus and Poisson’s ratio of thin films

    J Mater Res

    (1992)
  • Cited by (0)

    View full text