562 IEEE ELECTRON DEVICE LETTERS, VOL. 30, NO. 5, MAY 2009 On the Temperature and Field Dependence of Trap-Assisted Tunneling Current in Ge p + n Junctions Eddy Simoen, Francesca De Stefano, Geert Eneman, Brice De Jaeger, Cor Claeys, and Felice Crupi Abstract—The temperature dependence of the trap-assisted tunneling (TAT) current component in Ge p + n junctions has been studied between 25 ◦ C and 140 ◦ C. It is shown that the impact of TAT reduces significantly due to the combination of the negative thermal activation of the TAT-enhancement factor and the exponential increase of the diffusion current with temperature. It is shown that the experimental data can be well described in the frame of the Hurkx analytical model, which allows a fairly easy assessment of the TAT current contribution to the junction leakage current at realistic operation temperatures of Ge CMOS circuits. Index Terms—Germanium, leakage current mechanisms, p + n junctions, trap-assisted tunneling (TAT). I. I NTRODUCTION W HILE the high low-field mobility of germanium is surely an asset that can be exploited in future sub- 22-nm CMOS, its small band gap has raised serious concerns about limitations imposed by band-to-band tunneling at high electric fields [1]. At the same time, this may open up the door for new opportunities, like impact ionization [2], [3] or tunneling transistors [4], so that it is worthwhile to investigate in more detail the leakage current and breakdown mechanisms in germanium-based devices. It has recently been shown that the reverse current of Ge p + n highly doped drain (HDD) junctions at room tempera- ture is dominated by trap-assisted tunneling (TAT) for typical halo doping conditions, which are suitable for submicrometer pMOSFETs and for drain voltages up to 1 V [5], [6]. In good approximation, the area leakage current density J A can be de- scribed by the Hurkx model, which is originally developed for silicon devices [7], [8]. In this case, the Shockley–Read–Hall (SRH) generation rate by traps in the depletion region extending Manuscript received November 26, 2008; revised February 25, 2009. First published April 7, 2009; current version published April 28, 2009. The review of this letter was arranged by Editor C. Bulucea. E. Simoen, F. De Stefano, and B. De Jaeger are with IMEC, 3001 Leuven, Belgium. G. Eneman is with IMEC, 3001 Leuven, Belgium. He is also with the Depart- ment of Electrical Engineering, Katholieke Universiteit Leuven, 3001 Leuven, Belgium, and with Fonds Wetenschappelijk Onderzoek (FWO) Vlaanderen, 1000 Brussels, Belgium. C. Claeys is with IMEC, 3001 Leuven, Belgium, and also with the Depart- ment of Electrical Engineering, Katholieke Universiteit Leuven, 3001 Leuven, Belgium. F. Crupi is with the Università degli Studi della Calabria, 87100 Arcavacata di Rende, Italy and also with IMEC, 3001 Leuven, Belgium. Color versions of one or more of the figures in this letter are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/LED.2009.2017040 from the junction (x = 0) to x = W is modified by the electric- field-enhancement factor Γ(x) according to R trap (x)=[1 + Γ(x)] R SRH (x) (1) with Γ= ΔE T kT 1  0 exp  ΔE T kT u - 4 3 √ 2m ∗ (ΔE T ) 3/2 q|E| u 3/2  du (2) where kT is the thermal energy, q the absolute value of the electron charge,  is the reduced Planck constant, and m ∗ is the tunneling effective mass of the carriers. ΔE T is related to the trap level and equals E g /2 for midgap states. E is the local electric field, and E g is the band-gap energy. However, when investigating the current–voltage (I –V ) characteristics of a Ge p + n junction at higher temperatures, it appears that the reverse current changes from an approximately exponential increase with reverse bias at 25 ◦ C to a nearly bias-independent behavior at 140 ◦ C (Fig. 1). The latter is typical for a diffusion-current-dominated leakage current [9]. In other words, for realistic operation conditions, field-dependent junction leakage mechanisms in sub-22-nm Ge pMOSFETs appear to be less cumbersome than expected. In order to support this idea, the TAT leakage current is studied here in Ge p + n junctions as a function of temperature. It is demonstrated that the experimental Γ factor reduces exponentially with temperature, with thermal activation energy E A being in the range of 0.15–0.2 eV for the maximum electric fields studied (∼2 · 10 5 V/cm). This corresponds with a reduction by a decade of the field-enhancement factor between 25 ◦ C and 140 ◦ C, explaining the weaker bias dependence at high temperatures in Fig. 1. Fitting (1) and (2) to the experimental Γ factor yields an acceptable tunneling effective mass of 0.15 m 0 (m 0 is the rest mass of the electron). II. EXPERIMENTAL DETAILS Junctions have been processed on 200-mm epitaxial Ge-on- silicon substrates, using the CMOS processing described in de- tail in [10]. Active areas have been defined by windows etched in a deposited SiO 2 isolation layer. A 7.5-keV 4 · 10 15 cm -2 B ion implantation in Ge preamorphized substrates has been employed to fabricate the HDD junctions in the n-well. Stan- dard halo implantation has been performed [10]. Activation annealing was performed for 5 min at 500 ◦ C in a nitrogen 0741-3106/$25.00 © 2009 IEEE Authorized licensed use limited to: IMEC. Downloaded on April 27, 2009 at 05:22 from IEEE Xplore. Restrictions apply.