Thermal analysis of GaAs power monolithic microwave IC's mounted with epoxy attachment

Abstract

The effect of chip-mounting attachment on the thermal resistance of GaAs power field effect transistor (FET) modules has been experimentally investigated. The thermal resistance was evaluated for different GaAs chip thickness of 150 and 250 m through an electrical method utilizing temperature dependence of Schottky-barrier in the GaAs metal semiconductor FET's (MESFET's). The thermal resistance of low-cost epoxy-mounted GaAs chips, suitable for uniplanar monolithic microwave IC's (MMIC's), was found not to increase even up to a chip thickness of 250 m, while that of AuSn-mounted GaAs chips increased as was conventionally expected. Numerical simulation has also been presented for the similar case of GaAs power MMIC's. The result of simulation suggests that lower thermal conductivity of attachment material, such as epoxy attachment, leads to larger optimum chip thickness that minimizes the total thermal resistance.

Precise and simple DC-electrothermal characterization of MMIC power amplifiers

“…Thermal coupling between stages 1 and 2 seems to be independent of temperature, because the measured relationship between ATI and P0~s2 looks rather linear even though measurements with different PoIsl are shown. This reinforces the fact that the influence of the variation of substrate thermal properties with temperature is almost negligible when poor thermal conductors like epoxy are employed as attachment [16]. Additionally different bias conditions lead approximately to the same thermal coupling response; suggesting that fiom the thermal point of view, in order to obtain a good approximation of the MMlC thermal behavior, the selection of the bias point is not critical when using the proposed method.…”

“…On the contrary, if solders with high thermal conductivity are employed, most of the temperature rise is due to the poor thermal characteristics of the GaAs substrate. In the latter case GaAs thermal conductivity variation with temperature has a definite impact on the amplifier peak temperature, [17], (thus, the thermal analysis has to include this fact), while in the former case the influence is insignificant, [16], (so, GaAs thermal conductivity dependence on temperature can he neglected). It is also shown the equivalent simplified thermal circuit.…”

“…Although its use is restricted to low-power applications, it can he considered for high-power applications [I51 if a precise thermal study is first successfully performed. Thermal simulations and experiments have indicated that in the case of using poor thermal conductors as attachment most of the temperature rise on the MMIC surface is due to the attachment layer [16]. On the contrary, if solders with high thermal conductivity are employed, most of the temperature rise is due to the poor thermal characteristics of the GaAs substrate.…”

Characterising thermal resistances and capacitances of GaN high‐electron‐mobility transistors through dynamic electrothermal measurements

“…f Gain(t 2 ) − f Gain(t 1 ) = T 01 e −t 1 /t th 1 − e −t 2 /t th 1 (4) f Gain(t 4 ) − f Gain(t 3 ) = T 01 e −t 3 /t th 1 − e −t 4 /t th 1 (5) where t 1 , t 2 , t 3 and t 4 are four points on the time axis in Fig. 4.…”

“…Using these techniques it is not possible to measure T c of a fully packaged device. A more convenient approach is to measure T c indirectly on the basis of relations between various electrical properties and T c [4,5]. However, these methods are not suitable for capturing dynamic T c responses and are therefore usually limited to the characterisation of thermal resistances.…”