Niklas Rorsman

Accurate small-signal modeling of HFET's for millimeter-wave applications

In this paper we discuss the small-signal modeling of HFETs at millimeter-wave frequencies. A new and iterative method is used to extract the parasitic components. This method allows calculation of a /spl pi/-network to model the heterojunction field-effect transistor (HFET) pads, thus extending the validity of the model to higher frequencies. Formulas are derived to translate this /spl pi/-network into a transmission line. A new and general cold field-effect transistor (FET) equivalent circuit, including a Schottky series resistance, is used to extract the parasitic resistances and inductances. Finally, a new and compact set of analytical equations for calculation of the intrinsic parameters is presented. The real part of Y/sub 12/ is accounted for in these equations and its modeling is discussed. The accounting of Re(Y/sub 12/) improves the S-parameter modeling. Model parameters are extracted for an InAlAs/InGaAs/InP HFET from measured S-parameters up to 50 GHz, and the validity of the model is evaluated by comparison with measured data at 75-110 GHz.

Cryogenic wide-band ultra-low-noise IF amplifiers operating at ultra-low DC power

This paper describes cryogenic broad-band amplifiers with very low power consumption and very low noise for the 4-8-GHz frequency range. At room temperature, the two-stage InP-based amplifier has a gain of 27 dB and a noise temperature of 31 K with a power consumption of 14.4 mW per stage, including bias circuitry. When cooled to 15 K, an input noise temperature of 1.4 K is obtained at 5.7 mW per stage. At 0.51 mW per stage, the input noise increases to 2.4 K. The noise measurements have been repeated at different laboratories using different methods and are found consistent.

An empirical-table based FET model

A new large signal field effect transistor (FET) model combining empirical and table based models was developed and investigated experimentally. The Chalmers empirical model was used as a spline function for the table based model. Combining models improves accuracy and simplifies and speeds extraction. The procedure can be applied for other types of devices.

Investigation of the interface between silicon nitride passivations and AlGaN/AlN/GaN heterostructures by C"V… characterization of metal-insulator-semiconductor-heterostructure capacitors

Capacitance-voltage [C(V)] measurements of metal-insulator-semiconductor-heterostructure capacitors are used to investigate the interface between silicon nitride passivation and AlGaN/AlN/GaN heterostructure material. AlGaN/AlN/GaN samples having different silicon nitride passivating layers, deposited using three different deposition techniques, are evaluated. Different interface state distributions result in large differences in the C(V) characteristics. A method to extract fixed charge as well as traps from the C(V) characteristics is presented. Rough estimates of the emission time constants of the traps can be extracted by careful analysis of the C(V) characteristics. The fixed charge is positive for all samples, with a density varying between 1.3 x 10(12) and 7.1 x 10(12) cm(-2). For the traps, the peak density of interface states is varying between 16 x 10(12) and 31 x 10(12) cm(-2) eV(-1) for the three samples. It is concluded that, of the deposition methods investigated in this report, the low pressure chemical vapor deposited silicon nitride passivation shows the most promising results with regards to low densities of interface states

Large-signal modelling and comparison of AlGaN/GaN HEMTs and SiC MESFETs

The Large Signal (LS) model for GaN and SiC FET devices was developed and evaluated with DC, S, and LS measurements. Special attention was paid to improve the management of harmonics and to provide a more physical treatment of the dispersion. The model was implemented in a commercial CAD tool and exhibits good overall accuracy.

An AlGaN/GaN HEMT-Based Microstrip MMIC Process for Advanced Transceiver Design

A MMIC process in AlGaN/GaN technology for advanced transceiver design has been developed. The process is based on microstrip technology with a complete model library of passive elements and AlGaN/GaN HEMTs. The transistor technology in this process is suitable for both power and low noise design, demonstrated with a power density of 5 W/mm, and an NFmin of 1.4 dB at X-band. Process stability of subcircuits, complementary to power amplifiers and LNAs, in a transceiver system have been investigated. The results indicate that an all AlGaN/GaN MMIC transceiver is realizable using this technology.

A Wideband and Compact GaN MMIC Doherty Amplifier for Microwave Link Applications

This paper addresses the limitations and difficulties, in terms of DC-current density restrictions, and process limitations, associated with implementing impedance inverters with high characteristic impedance for monolithic microwave integrated circuit (MMIC) Doherty power amplifiers (DPAs). It is theoretically shown that impedance inverters with high characteristic impedance can be realized by utilizing the output capacitance of the active devices, together with a compact Tee-network of transmission lines with feasible linewidths. The utility of the impedance inverter is proven by design and fabrication of a GaN MMIC-DPA for microwave link applications. Continuous wave (CW) measurements demonstrate a maximum output power of 35±0.5dBm over a frequency range of 6.6-8.5 GHz. The power added efficiency (PAE) in 9 dB output power back-off (OPBO) is better than 30% in a frequency range of 6.7-7.8 GHz. Moreover, linearized modulated measurements, employing a 10 MHz 256-QAM signal with 7.8 dB peak to average power ratio (PAPR), demonstrate higher than 35% average PAE, with 27.5 ± 0.2 dBm average output power, and an adjacent channel power ratio (ACPR) less than -45 dBc, across a 6.8-8.5 GHz frequency range. The fabricated chip-size measures 2.1 mm × 1.5mm.

Electrical properties, microstructure, and thermal stability of Ta-based ohmic contacts annealed at low temperature for GaN HEMTs

Ta-based ohmic contacts to gallium nitride high electron mobility transistor (GaN HEMT) epitaxial structures were investigated. Two metallization schemes were considered: Ta/Al/Ni(Ta)/Au and Ta/Al/Ta. The latter was superior in terms of lower contact resistance (R-c) and wider process window. The metallizations were applied to two different heterostructures (GaN/Al0.14Ga0.86N/GaN and Al0.25Ga0.75N/GaN). The lowest measured R-c was 0.06 and 0.28 Omega mm, respectively. The main advantage of the Ta-based ohmic contacts over conventional Ti-based contacts was the low anneal temperature. The optimum temperature of annealing was found to be 550-575 degrees C. From optical and scanning electron microscopy, it was clear that excellent surface morphology and edge acuity were obtained at these low temperatures. This facilitates lateral scaling of the GaN HEMT. TEM images were taken of the contact cross sections onto which EDX measurements were performed. The aim was to investigate the microstructure and the contact mechanism. Storage tests at 300 degrees C for more than 400 h in air ambient showed no deterioration of R-c.

Comparison of the DC and Microwave Performance of AlGaN/GaN HEMTs Grown on SiC by MOCVD With Fe-Doped or Unintentionally Doped GaN Buffer Layers

In this brief, the authors present a comparative and comprehensive investigation of the effect of the type of resistive GaN buffers on the dc, dynamic, microwave, and power performance of AlGaN/GaN high electron mobility transistors (HEMTs). Two types of buffer layers were investigated: 1) a nonintentionally doped resistive GaN buffer and 2) an Fe-compensated buffer. The Fe modulation-doped buffer is shown to be favorable for better dc isolation. The RF small-signal performance of the HEMTs does not exhibit any significant dependence on the type of resistive GaN buffer. However, the type of GaN buffer influences considerably the dynamic large-signal characteristics of the processed AlGaN/GaN HEMTs. The continuous-wave output power density of the AlGaN/GaN HEMTs at 3 GHz was increased from 3.4 to 9.7 W/mm by using a nonintentionally doped buffer instead of an Fe-doped one. Based on this observation combined with pulsed current-voltage measurements, we ascribe this difference to the deep trapping of electrons by defects in the GaN buffer introduced by the incorporation of Fe

1 W/mm RF power density at 3.2 GHz for a dual-layer RESURF LDMOS transistor

In this letter, we present state-of-the-art performance, in terms of output power density, for an RF-power LDMOS transistor. The novel device structure has a dual-layer RESURF of the drift region, which allows for a sub-/spl mu/m channel length and a high breakdown voltage of 110 V. The output power density is more than 2 W/mm at 1 GHz and a V/sub DS/=70 V, with a stable gain of 23 dB at V/sub DS/=50 V. At 3.2 GHz the power density is over 1 W/mm at V/sub DS/=50 V and 0.6 W/mm at V/sub DS/=28 V. These results are to our knowledge the best ever for silicon power MOSFETs.