1 Introduction
 
Throughout the last years we can observe an increased demand of devices operating well above 1 GHz. Advances in material research now offer a wide variety of material systems which allow to realize devices operating in this frequency range. Up to now there is no consensus on what the best technology for which application will be. Circuits realized on silicon are now operating up to 60 GHz but the power capabilities and the efficiency is much lower than in other technologies [1]. Devices realized in other material systems like AlGaN/GaN [2, 3] show very good power capabilities and high frequency performance and thus are promising technologies for very high power applications in the future. But in order to satisfy not only the needs of some small niches the most important key for success is to realize a function as cheap as possible. Sales of discrete devices stagnate whereas the total volume of non silicon semiconductors increases dramatically [4]. This shows that the integration level is increasing and more functions previously realized by hybrid systems are integrated on chip.

The most commonly used semiconductor devices for applications in the GHz range now are GaAs based MESFETs, HEMTs and HBTs [4]. Although MESFETs are the cheapest devices because they can be realized with bulk material, i. e. without epitaxially grown layers, HEMTs and HBTs are promising devices for the near future. The advantage of HEMTs and HBTs is a factor of 2 to 3 higher power density compared to MESFETs which leads to significantly smaller chip size. As the die size of GaAs wafers is increased from 3'' to today's mostly used 4'' and further to 6'', which is expected in large volume in two or three years, HEMTs and HBTs are becoming more competitive even in the lower frequency range of 0.9 GHz and 1.9 GHz. These frequencies are used for mobile communication and thus represent one of the largest market for high frequency application.

In  Figure 1.1 the gain per amplifier stage and output power is shown versus the operation frequency for different FETs, HBTs, and HEMTs. It clearly demonstrates advantages of HEMTs for operation frequencies above 10 GHz and very good competitiveness for lower frequencies. Since it is quite expensive to keep different technologies established in a production line it is favorable to cover many applications with a single technology. This represents a major advantage of HEMTs if applications in the whole frequency range from 1 GHz to 100 GHz are requested.
 

In order to react fast on market needs and lower the cost there is a strong demand to minimize cycle times and the number of technology runs necessary for the development of a new product. Accurate device simulation can play an important role to meet these requirements.

In this work various aspects of GaAs based HEMTs are investigated. Not only trade-off between various device characteristics but also the optimization of device performance and cost of production are analyzed. The investigations are performed by means of simulations and measurements of manufactured devices. The simulations are performed using the generic device simulator MINIMOS­NT [5, 6].

Chapters 2 ­ 4 cover the principles of HEMTs, measurements and parameter extraction as well as a description of the models used in the simulator. In Chapter 5 the setup of the device geometry and the combination of the models used for the simulation is described. After the verification of the simulation results HEMTs for low noise, power and millimeter wave applications are investigated in Chapter 6. Based on very good agreement between simulated and measured data the simulation is used to predict performance of HEMTs which have not been fabricated yet.
 

   
Next: 2 The Principles of a HEMT Up: Dissertation Helmut Brech Previous: Acknowledgements