(Invited) Alternative High n-Type Doping Techniques in Germanium

Tuesday, 7 October 2014: 14:00
Expo Center, 1st Floor, Universal 17 (Moon Palace Resort)
G. Capellini (IHP-Innovations for High Performance Microelectronics, Dip. Scienze, UniversitÓ Roma Tre), W. M. Klesse, G. Mattoni, M. Y. Simmons, and G. Scappucci (School of Physics, University of New South Wales)
There is an international race both in industry and research labs to achieve ultra-high n-type doping of germanium driven by possible applications cutting across different fields, ranging from the development of nanoscale high-mobility transistors[1] to the fabrication of a Si-compatible laser[2].

Owing to the inherent difficulties to achieve heavy n-doping of Ge in a controllable way using traditional implantation techniques[3], alternative in-situdoping techniques have recently  attracted interest. The aim is to achieve high electrical activation of dopants  while limiting their diffusion and  avoiding crystal structure damage induced by the ion-implantation.

Embracing a radical bottom-up approach, where the chemistry of dopant precursors is studied at the atomic level, we have developed a novel technique for achieving ultra-high doping of Ge. Our approach consists in the in-situ adsorption of phosphorous-containing molecules onto a Ge(001) surface, following thermal incorporation of the phosphorus atoms into the Ge matrix, and a final encapsulation under an epitaxial Ge layer grown by molecular beam epitaxy[4].

This process results in the formation of a spatially-confined P d-layer. The high morphological quality of the crystal matrix allows to stack an arbitrary number of those P δ-layers separated by Ge spacer-layers of arbitrary thickness, resulting in a P:Ge layer featuring the desired doping density.

Using this technique we have demonstrated electrically active donor density up to ~2´1020 cm-3.

With the help of experimental results obtained by scanning tunneling microscopy (STM), secondary ion mass spectroscopy, atomic probe tomography (APT), and magneto-transport measurements, in this talk I shall discuss all the relevant aspects of this novel doping process.

I will first present the early stages of the doping resolved at the atomic scale (step (1) in Fig.1). In particular, I will focus on the different chemisorption pathways governing the interaction of PH3[5] and P2 molecules[6]with a clean Ge(001) surface and leading to different self-saturated donor density on its surface.

Subsequently, I will review the aspects concerning the incorporation of  P atoms within the Ge substrate (step (2)) and the subsequent Ge homoepitaxy leading to the formation of  a d-layer (3). I will specially focus on the impact of the process thermal budget on the donor spatial distribution, on their electrical activation, and on the surface morphology. In particular, we will demonstrate that in Ge is possible to find an optimal process temperature  window allowing, at the same time,  spatially-defined d-layers encapsulated in a matrix of high crystal quality (see Fig.2). This feature is a Ge key characteristic which enables the deposition of multiple stacked δ-layers and, thus, the realization of Ge:P doped layers of arbitrary donor concentration and thickness.

Moreover, I will discuss  the impact of the process parameters on the δ-layer electrical properties and the strategies used to maximize electron densities and minimize the  electrical resistivity  (see Fig. 3).

Bearing in mind the particular interest for industrial application, I will then discuss the challenges and opportunities in extending this doping process to other Ge based substrates, such as Ge/Si(001) and GeOI[7] , or to other deposition techniques which include reduced pressure chemical vapor deposition (CVD)[8] and ultra-high-vacuum CVD[9].

Figure 1: Repeated cycles of (1) PH3 or P2adsorption; (2) incorporation; and (3) encapsulation lead to the formation of a controlled stack of P-δ layers (schematics).

Figure 2: APT (top left) and relative donor density (bottom) as obtained in a sample containing 4 δ-layers. On the right side STM images of the top sample surface are displayed: notice the low surface roughness obtained along with very localized donor distributions. 

Figure 3: Resistivity of P δ-layers obtained using different deposition parameters, dopant gas, and post growth processes as a function of the electrically active carrier concentration. 

[1] R. Pillarisetty, Nature 479, 324  (2011)

[2] B. Dutt et al., J. IEEE Photonics 4, 2002 (2012)

[3] S. Brotzmann et al.,  J. Appl. Phys. 103, 033508 (2008)

[4] G. Scappucci et al., Nanoscale 5, 2600 (2013)

[5] G. Scappucci et al., Phys. Rev. Lett. 109, 076101 (2012)

[6] G. Mattoni et al., ACS Nano 7, 11310 (2013)

[7] W.M. Klesse at al., Appl. Phys. Lett. 102, 151103 (2013)

[8] Y. Yamamoto et al., Solid-State Electron. 83, 25 (2013)

[9] R. Camacho-Aguilera et al., Opt. Mat. Express 2, 1462 (2012)