M.H. Somerville

Direct correlation between impact ionization and the kink effect in InAlAs/InGaAs HEMTs

We present new, unmistakable experimental evidence directly linking the kink effect with impact ionization in the channel of InAlAs-InGaAs HEMTs on InP. Through the use of a sidegate structure, we confirm that the impact ionization coefficient obeys the classic exponential dependence on the inverse electric field at the drain end of the gate, and that the onset of the kink strongly coincides with the onset of impact ionization in the devices we consider. These measurements illuminate the functional relationship between the kink and impact ionization, and therefore will allow assessment of the numerous impact-ionization related kink mechanisms that have recently been suggested in the literature.

On-state breakdown in power HEMTs: measurements and modeling

A new definition of and measurement technique for on-state breakdown in high electron mobility transistors (HEMTs) is presented. The new gate current extraction technique is unambiguous, simple, and non-destructive. Using this technique in conjunction with sidegate and temperature-dependent measurements, we illuminate the different roles that thermionic field emission and impact ionization play in HEMT breakdown. This physical understanding allows the creation of a phenomenological model for breakdown, and demonstrates that depending on device design, either on-state or off-state breakdown can limit maximum power.

A new gate current extraction technique for measurement of on-state breakdown voltage in HEMTs

We present a new simple three-terminal technique for measuring the on-state breakdown voltage in HEMTs. The gate current extraction technique involves grounding the source, and extracting a constant current from the gate. The drain current is then ramped from the off-state to the on-state, and the locus of drain voltage is measured. This locus of drain current versus drain voltage provides a simple, unambiguous definition of the on-state breakdown voltage which is consistent with the accepted definition of off-state breakdown. The technique is relatively safe and repeatable so that temperature dependent measurements of on-state breakdown can be carried out. This helps illuminate the physics of both off-state and on-state breakdown.

Breakdown in millimeter-wave power InP HEMTs: a comparison with GaAs PHEMT's

In spite of their outstanding transport characteristics, InP high-electron mobility transistors (HEMTs) deliver lower output power than GaAs pseudomorphic HEMTs (PHEMTs) throughout most of the millimeter-wave regime. However, the superior power-added efficiency of InP HEMTs when compared with GaAs PHEMTs makes this technology attractive for many applications. The reason for the relatively inferior power output of InP HEMTs lies in their comparatively small off-state and on-state breakdown voltages. This paper reviews the state of knowledge regarding the physics of breakdown voltage in InP HEMTs, placing it in contrast with GaAs PHEMTs. It also presents current understanding regarding burnout, a closely related phenomenon. This paper concludes by discussing strategies for improving the breakdown voltage and the power output of InP HEMTs.

A model for tunneling-limited breakdown in high-power HEMTs

We present a new predictive model for off-state breakdown in InAlAs/InGaAs and AlGaAs/InGaAs power high electron mobility transistors (HEMTs). The proposed model suggests that electron tunneling from the gate edge, and not impact ionization, is responsible for off-state breakdown in these devices. The model indicates that the crucial variables in determining the off-state breakdown voltage of power HEMTs are the sheet carrier concentration in the extrinsic gate-drain region, and the gate Schottky barrier height. Other design parameters have only secondary impact on the breakdown voltage for realistic device designs.

A new physical model for the kink effect on InAlAs/InGaAs HEMTs

New measurements providing direct evidence linking the kink effect and impact ionization in InAlAs/InGaAs HEMTs are reported. Current kink models are not consistent with our findings. We propose a new mechanism, barrier-induced hole pile-up at the source, to explain the kink. The new model is shown to be consistent with both room temperature and low temperature measurements. These results allow formulation of a simple equivalent circuit model of the kink.

Off-state breakdown in power pHEMTs: the impact of the source

Summary form only given. Conventional wisdom suggests that in pseudomorphic high electron mobility transistors (pHEMTs), the field between the drain and the gate determines off-state breakdown, and that the drain to gate voltage therefore sets the breakdown voltage of the device. Thus, the two terminal breakdown voltage is a widely used figure of merit, and most models for breakdown focus on the depletion region in the gate-drain gap, while altogether ignoring the source. We present new measurements and simulations that demonstrate that for power pHEMTs, the electrostatic interaction of the source seriously degrades the devices gate-drain breakdown, and must be taken into consideration in device design. As a vehicle for this study we have used a state-of-the-art L/sub G/=0.25 /spl mu/m double heterostructure pHEMT with excellent power performance (P/sub 0/= 1W, Gain= 11 dB, and PAE=60% at 10 GHz for W/sub G/=1200 /spl mu/m) and high breakdown voltage (BV/sub DG/=21 V at I/sub D/=1 mA/mm).

Temperature dependence of breakdown voltage in InAlAs/InGaAs HEMTs: theory and experiments

We present results of an experimental and theoretical study of the temperature dependence of the off-state breakdown voltage of InAlAs/InGaAs high electron mobility transistors (HEMTs). We find that the breakdown voltage (BV) has a negative temperature coefficient that is more prominent for lower values of the extrinsic sheet carrier concentration (n/sub s/). Structural parameters such as the insulator thickness and top-to-bottom delta doping ratio have little effect on BV if n/sub s/ is held constant. These results are consistent with an extension of a new tunneling model for breakdown in HEMTs to include thermionic-field emission.

Degradation Uniformity of RF-Power GaAs PHEMTs Under Electrical Stress

We have studied the electrical degradation of RF-power PHEMTs by means of in situ 2-D light-emission measurements. Electroluminescence originates in the recombination of holes that have been generated by impact ionization. The local light intensity, thus, maps the electric-field distribution at the drain side of the device. This allows us to probe the uniformity of electrical degradation due to electric-field-driven mechanisms. We find that electrical degradation proceeds in a highly nonuniform manner across the width of the device. In an initial phase, degradation takes place preferentially toward the center of the gate finger. In advanced stages of degradation, the edges of the device degrade at a preferential rate. We identify the origin of this behavior as a small systematic nonuniformity in the recess geometry that impacts the magnitude of the electric field on the drain of the device. Our research suggests that a close examination of the width distribution of electric field in RF-power PHEMTs (and FETs in general) is essential to enhance their long-term reliability.

Temperature and carrier density dependence of mobility in a heavily doped quantum well

Interest in heterostructure field‐effect transistors (HFETs) utilizing narrow, heavily doped channels motivates a study of mobility in heavily doped quantum wells. We have measured electron mobility as a function of carrier concentration and temperature in an In0.15Ga0.85As quantum well with a doping of ND=6×1012 cm−2. Mobility is found to rise significantly as the ratio of electron to impurity concentration increases. Even at T=300 K, μ climbs by nearly a factor of 2 as carrier concentration in the well is increased from 1×1012 to 5×1012 cm−2. The results agree qualitatively with recently published theoretical predictions, and suggest that device models utilizing constant mobility are not appropriate for HFETs using doped two‐dimensional channels.