2.1 History of HEMTs

As many other discoveries, the idea for a HEMT structure was a product of a research with different purposes and there were several factors superimposed. The late 70s saw the evolution of the molecular beam epitaxy growth technique and modulation doping together with a vivid interest in the behavior of quantum well structures [1] (the latter peaking in the work of Klitzing, Laughlin, Stomer, and Tsui).

At this time T. Mimura and his colleagues at Fujitsu were working on GaAs MESFETs. Facing problems with a high-density of the surface states near the interface, they decided to use a modulation-doped heterojunction superlattice and were able to produce depletion type MOSFETs [2]. While those structures were still plagued by several issues, the idea to control the electrons in the superlattice occurred to him. He achieved this by introducing a Schottky gate contact over a single heterojunction. Thus, the AlGaAs/GaAs HEMT was born [3]. Subsequently the first HEMT based integrated circuit was reported [4]. Alongside Fujitsu a number of other research facilities joined on the further development of the new structures: Bell Labs, Thomson CSF, Honeywell, IBM [5]. In order to counter different problems, several designs were proposed: AlGaAs/GaAs HEMTs, AlGaAs/InGaAs pseudomorphic HEMTs (pHEMTs), AlInAs/ InGaAs/InP HEMTs (ordered by increasing $ f_\ensuremath {\mathrm {t}}$) [6]. However, until the end of the decade HEMTs mainly found military and space applications [7]. Only in the 90s the technology entered the consumer market in satellite receivers and emerging mobile phone systems.

In the beginning of the last decade new methods for deposition of GaN on sapphire by MOCVD were developed. Thus, the production of AlGaN/GaN-based HEMTs was possible [8]. GaN has a wide band gap which brings the advantages of higher breakdown voltages and higher operational temperature. Due to the large lattice mismatch between AlN and GaN a strain in the AlGaN layer is induced, which generates a piezoelectric field. Together with the large conduction band offset and the spontaneous polarization this leads to very high values for the electron sheet charge density [9]. This large potential of AlGaN/GaN structures (and the indirect advantage of excellent thermal conductivity of the sapphire substrates) was realized very soon and the research focus partially shifted from AlGaAs/GaAs to AlGaN/GaN devices.

In the course of further development and optimization various techniques were adopted. An approach previously used in high-voltage p-n junctions [10], the field-plate electrode, significantly improved device performance by reducing the peak values of the electric field in the device. Thus, the breakdown voltage could be further increased. This technique was further refined to T-shaped [11] and subsequently Y-shaped gate electrodes [12]. Another step in optimization of the structure is the addition of a thin AlN barrier between the GaN channel and the AlGaN layer. It increases the conduction band offset and the two-dimensional electron gas (2DEG) density and decreases the alloy disorder scattering, thereby increasing the mobility [13]. An additional option to enhance the electron gas transport properties is the double-heterojunction structure [14]. The InGaN layer under the channel introduces a negative polarization charge at the interface, and thereby improves the carrier confinement in the channel.

While the depletion mode (D-mode) technology has been significantly improved, no comparable progress on the enhancement counterparts can be noted. However, such devices have advantages in certain applications and are therefore getting in the focus of research activities in the recent years. Several groups have proposed interesting approaches. Devices featuring very thin AlGaN layers [15] and Fluoride-based plasma treatment [16] have been proposed, however certain stability concerns remain. A very promising method is the recess gate structure reported by Kumar et al. [17]. Also recently, excellent results have been achieved with InGaN-cap devices [18].


S. Vitanov: Simulation of High Electron Mobility Transistors