F. Aniel

On the use of cryogenic measurements to investigate the potential of Si/SiGe:C HBTs for terahertz operation

The transit times analysis, at room and cryogenic temperatures, of SiGe HBTs featuring various ftau/fmax trade-offs is performed. It allows to identify the principal development axes to reach half-terahertz at 300 K, a frequency already obtained at 40 K.

Full-band matrix solution of the Boltzmann transport equation and electron impact ionization in GaAs

Abstract A direct matrix solution of the Boltzmann transport equation is developed to investigate impact ionization and other hot carrier phenomena in compound semiconductor devices. The model is applied to GaAs. A full-band calculation of the semiconductor structure is performed using k · p theory. An original determination of the distribution function in k-space as a function of electric field and temperature is achieved. The dependence of the ionization rate on energy, of the drift velocity and of the ionization coefficient on electric field are in good agreement with Monte Carlo data at 300 K and validate the modeling approach. The variation of the impact ionization coefficient on temperature and electric field in GaAs is discussed.

High Frequency Analysis of InP Transistors versus Temperature

We present a device HF analysis versus temperature and its application to High Electron Mobility Transistor (HEMT) on InP. We give examples of temperature dependence of HF small signal equivalent circuit parameters. We underline the strong influence of the parasitic access resistance determination on the extracted values of device intrinsic elements . Introduction: The microwave analysis of III-V transistors at variable temperature brings useful informations on the carrier dynamics related to trapping effects, the improvement of transport and of noise properties, self-heating among other points. There is also a continuing interest in cryogenic operation, since device performances improve at low temperatures. But there is few information available on FET small signal parameter dependences versus temperature. Often designers of MMICs at low temperature have still to relie on 300K HF device modeling. In this study on InP HEMTs, we present first the specific aspect of the low temperature characterization techniques. Then we describe the analytic extraction tool and the impact of extrinsic parasitic element determination on the device parameter extraction acuracy. Finally we discuss the main trends of parameter temperature dependences and we give some temperature coefficients of small signal parameters. a) Characterization : Measurements are performed in a He circulation cryostat with nitrogen cooled microwave accesses for long term thermal gradient stability. A full HF calibration is performed only at selected temperatures. In the intervals a simple C-open measurement corrects the phase of error vectors of the nearest full calibration. Cryogenic device deembeding is a critical aspect. We measure specific test structures to correct for device access losses: long lines to determine the losses, a short and an open circuit to obtain the four S parameters of the quadrupoles representing the two access lines (150μm on each side). The modeling accuracy allows to take into account the skin effect. The procedure keeps the reference planes under the HF probes. It allows to control the sensitivity of the deembeding accuracy on the device small signal parameters. We have realised a driving software to perform sequences of dc/HF measurements sequentially. The device is biased back in a well defined quiescent off-state during a sufficiently long time between any two successive investigated operating points. This is intended to screen its previous bias history relatively to carrier trapping effects. The Fig. 1 shows the HEMT small signal equivalent circuit . We use for both gate and drain accesses a coplanar waveguide model and residual inductances to represent the contact due to gate metallization and the drain metallization over cap layers. b) Extraction of HF parameters : We have developed an efficient and robust analytic extraction procedure of the parameters using least square algorithms with no limited series expansion. Device parameter frequency dependences for Gm, Gd and Cds are unnecessary in fitting measurements. Otherwise the analytic extraction proceeds as in [2].The electrostatic capacitances are taken into account. The measurements allow to extract the electrostatic capacitances (Cgse and Cgde) related to the device accesses or to the device itself (between the gate foot and the cap layers). This relies on previous comparisons between experimental and simulated results [4]. c-r b t t b L r L-r b r t c

Apparent base resistance decomposition by means of small-signal and high-frequency noise analyses of submicron InP/InGaAs HBTs

We present an original and reliable technique to elucidate the different contributions to the apparent base resistance (RB?= RBx?+?XRBi) of double heterojunction bipolar transistors (HBTs) designed by Alcatel-Thales III?V Lab. The extrinsic base resistance (RBx) is quantified using small-signal measurements. The base-collector junction distribution factor (X) and the intrinsic base resistance (RBi) are extracted from high-frequency noise (HFN) measurements. This method was applied to three InP/InGaAs HBTs having different emitter surfaces (SE). The correct determination of RBx, X and RBi?may be a useful tool for compact and/or linear electrical modelling and may give some guidelines to designers to improve operation frequencies. Moreover, this strategy can be applied to any layout and technological variation of HBT; it can be also applied to homojunction bipolar transistors. Our results show that HFN analysis should be included to fully characterize bipolar transistors.

Experimental and modelling investigation of the influence of noise transit time and self-heating on microwave noise of Si/SiGe:C and InP/InGaAs HBTs

The results of a comprehensive experimental and modelling investigation of high-frequency noise (HFN) of Si/SiGe:C and InP/InGaAs heterojunction bipolar transistors (HBTs) demonstrate that the key parameters to correctly model HFN for both HBTs are the self-heating (SH) effect at high collector injection levels, i.e. JCxa0> 2–3 mA µm−2, and the noise transport time (τ) which must be included at all bias levels. In contrast, thermal noise contribution of the base–emitter junction of the Si/SiGe:C HBT, associated with the diffusive nature of electron transport at this junction because of the lack of conduction band discontinuity, is only important at operation frequencies lying in the frequency range f > fT/2. The noise analysis indicates that in order to correctly model HFN performances at any injection level and operation frequency, τ must be constant with bias and equal to the addition of base and collector transit times (τBxa0+xa0τC). On the other hand, at high JCxa0levels if SH is neglected, then the amplitude of the equivalent noise resistance (Rn) will be the most impacted noise parameter for both HBTs. Concerning the InP/InGaAs HBT, if SH is off at JCxa0= 5.5xa0mAxa0µm−2, the minimum noise figure (NFmin) of the InP HBT will be underestimated by 10%.

Cryogenic behavior of ultrashort gate AlGaAs/GaAs and pseudomorphic AlGaAs/InGaAs/GaAs HEMT's

High performance δ-doped double recess 0.1-0.4μm AlGaAs-GaAs S-HEMTs[1] and δ-doped single-recess 0.1-0.7μm AlGaAs/GaInAs/GaAs PM-HEMTs [2-3] are studied from 300K down to 60K. DC and 0.1-40GHz HF measurements followed by electric parameter extractions give extensive evolutions of HEMT electric parameters versus temperature T, gatelength 1g. The results are interpreted in terms of the transition regime from diffusive to strong overshoot/near ballistic carrier transport under the gate. In the longer devices (1g=0.4-0.7μm), the latter effects become important at low temperatures and bring frequency intrinsic performance improvement mainly in terms of an increase of maximum intrinsic transconductance gmo. The shorter devices (0.1-0.2μm) appear to be already well in the overshoot regime at 300K due to the high quality epitaxial layers, and their frequency performance improvement upon cooling comes mainly from a reduction of intrinsic capacitance Cgsint. The relative increase of intrinsic unity current gain cut-off frequency ftint upon cooling rates at a similar value of 30-40% for all investigated gatelengths(0.1-0.7μm).

Monte Carlo-Based Analytical Models for Electron and Hole Electrical Parameters in Strained SiGeC Alloys

Bipolar transistors in advanced BiCMOS technology rely on highly-doped SiGeC bases strained on Si. Modeling the electrical properties of such devices by TCAD means requires an accurate description of SiGeC transport parameters, including low-field mobility, saturation velocity and energy relaxation time. Since bipolar transistor operation involves electron and hole transport, both types of carriers must be adressed. A distinction between majority and minority carriers must also be physically accounted for. In this paper, transport properties of doped and strained Si1−x−yGexCy alloys are investigated by Full-Band Monte Carlo (MC) method with a view of deriving analytical models suitable for implementation in hydrodynamic- based TCAD simulators.

Scaling behavior of delta‐doped AlGaAs/InGaAs high electron mobility transistors with gatelengths down to 60 nm and source‐drain gaps down to 230 nm

Pseudomorphic delta‐doped AlGaAs/InGaAs/GaAs high electron mobility transistors with gatelengths ranging from 60 to 250 nm have been fabricated in submicron source‐drain gaps using high‐resolution electron beam lithography. As a consequence of short channel effects, the maximum transconductance gm is not improved by decreasing the gatelength from 250 nm (950 mS/mm) to 60 nm (800 mS/mm). When comparing high‐frequency (HF) and direct‐current values of the peak transconductances for devices with gatelengths shorter than 100 nm, the HF performance is remarkably higher. The difference can be explained by the fact that short channel effects are less important at high frequencies. Due to the decrease of the gate capacitance, the transition frequency fT increases as the gatelength is reduced.

Germanium content and base doping level influence on extrinsic base resistance and dynamic performances of SiGe:C heterojunction bipolar transistors

We describe a reliable technique to separate the different contributions to the apparent base resistance (RB?=?RBx?+?X RBi) of silicon germanium carbon (SiGe:C) heterojunction bipolar transistors (HBTs). The extrinsic base resistance (RBx) is quantified using small-signal measurements. The base-collector junction distribution factor (X) and the intrinsic base resistance (RBi) are extracted from high frequency noise (MWN) measurements. This method is applied to five different SiGe:C HBTs varying in base doping level and germanium content. The results show that high doping levels improve high frequency noise performances while germanium gradient helps to maintain outstanding dynamic performances. This method could be used to elucidate the base technological configuration that ensures low noise together with remarkable dynamic performances in state-of-the-art SiGe:C HBTs.

Procedure to derive analytical models for microwave noise performances of Si/SiGe:C and InP/InGaAs heterojunction bipolar transistors

We present a useful procedure to derive simplified expressions to model the minimum noise factor and the equivalent noise resistance of Si/SiGe:C and InP/InGaAs heterojunction bipolar transistors (HBTs). An acceptable agreement between models and measurements at operation frequencies up to 18 GHz and at several bias points is demonstrated. The development procedure includes all the significant microwave noise sources of the HBTs. These relations should be useful to model Fmin and Rn for state-of-the-art IV-IV and III–V HBTs. The method is the first step to derive noise analyses formulas valid for operation frequencies near the unitary current gain frequency (fT); however, to achieve this goal a necessary condition is to have access to HFN measurements up to this frequency regime.