Impact of surface preparation for n-type Si:P and p-type SiGe:B semiconductors on low resistance silicide contacts

Highlights

• We present detailed XPS & TEM characterization of contact interfaces directly applicable to sub 10nm logic technology nodes.

• Establish impact of processing and interfacial treatments prior to contact metallization to improve contact resistivity.

• We correlate XPS characterizations with contact resistivity measurements using technological-relevant contact dimensions.

• Up to 20% contact resistance improvement seen using ammonia-based and HF-based surface treatments on n/p-type surfaces.

Abstract

In our study, we evaluate effective silicon and germanium oxide reduction by two surface treatments to achieve low contact resistivity at the semiconductor/metal interface. These chemistries, one alkaline and the other an acidic fluorine-based treatment, were utilized on epitaxial n-type Si:P and p-type Si_1 − xGe_x:B (x = 0.47) substrates to isolate any unique effects that may be present on doped, n-type and p-type semiconductor surfaces. To mimic plasma damage and surface conditions in an integrated process flow for actual logic devices, X-ray photoemission spectroscopy (XPS) characterization was performed on simplified blanket films after NF₃-based gas cluster ion beam (GCIB) exposure and subsequent aqueous treatments. Si:P and SiGe:B surfaces both demonstrated an increase in SiO₂ concentration after GCIB exposure, with SiGe:B surfaces showing a preferential SiO₂ surface oxidation. Subsequent acidic treatment showed reduction in SiO₂ concentration on both epitaxial surfaces, with the alkaline (basic) treatment showing little change in surface composition. Electrical characterization on simplified contact structures showed a benefit in contact resistivity of 15–23% in Si:P and 10–13% in SiGe:B for the chemistries evaluated.

Introduction

With the continued scaling of logic technologies, has come the ever-more demanding need for low resistivity contacts in metal-oxide-semiconductor field effect transistors (MOSFETs). Serving as the coupling of the metal interconnect to the Source/Drain (S/D) regions of the semiconductor device, metal silicides have long been established as the dominant material to decrease parasitic resistance [1]. Previous work has focused on optimizing silicide formation conditions and so-called “interfacial engineering”, where local dopant concentrations and activation conditions can lead to a lower Schottky barrier and a more favorable contact [2], [3], [4], [5]. Due to the sensitivity of the contact to this interfacial composition, it follows that the preparation (removal of resistive oxides) and passivation (minimizing unbonded Si/Ge atoms) of the S/D semiconductor surface is crucial for achieving low contact resistivity.

Removal of post etch residue (PER) is necessary in order to achieve clean semiconductor surfaces, improving yield and proper function of devices. It is estimated that over 50% of the yield loss in semiconductor manufacturing is due to micro-contamination [6]. Hence, subsequent wet cleaning steps are necessary to remove residues from the bottom and sidewalls in high aspect ratio (HAR) contact features while maintaining targeted trench critical dimensions (CD). This requires minimizing lateral etch without damaging other exposed materials, such as etch-stop layers (ESL), with aqueous cleaning solutions. Typically, dilute hydrofluoric acid (dHF) solutions have been used to remove post etch residues [7], [8]. However, controlling the CD of a contact via for sub-10 nm nodes is stringent and meeting these targets is extremely challenging [9]. Recent publications demonstrate that ultra-dilute HF solutions (< 0.05 wt%), in the presence of additives and at an elevated temperature, give excellent control of targeted CD for ultra-low k films (k = 2.7) [10]. This approach for PER removal provides a pathway to evaluate an ultra-dilute modified HF solution for S/D contact cleaning.

In common integration schemes, an etch-stop layer is used to protect semiconductor surfaces from oxygen and fluorine-based reactive ion etch (RIE) chemistries during the dielectric etch. However, a challenge remains for a highly selective and directional removal of ESL films without attacking the underlying S/D semiconductor surface. One method to achieve low-damage is the use of a gas-cluster ion beam (GCIB) [11], [12], [13]. A wide variety of GCIB applications in semiconductor processing (e.g. low energy implant, controlled etching, planarization, thin film deposition) can be found in literature [14], [15]. In a recent publication, controlled Cu etching at room temperature (RT) and high vacuum conditions was achieved by controlled oxidation of Cu surface using O₂-GCIB followed by the introduction of vapor acetic acid [12]. Acetic acid facilitated the CuO etching and the organic by-products formed during reaction were desorbed due to local and transient GCIB heating. These results indicate that the GCIB process can be highly effective in controlled etching of metal surfaces. However, O₂-containing GCIB processes are not suitable for exposed SiGe:B/Si:P surfaces prior to metallization in S/D contact applications. Therefore, determining the impact of non-oxidizing reactive gases on semiconductor surfaces is necessary in deciding its viability in logic technologies.

In order to achieve atomically clean semiconductor surfaces for low contact resistance, any damaged or amorphous surfaces formed as a result of GCIB processing must be removed. Previous efforts to remove these highly resistive layers, coincidently reducing Schottky barrier height for carrier conduction, have included traditional aqueous solutions [16], hydrogen plasmas [17], and gaseous chemical etching [18]. Few comprehensive studies exist that present extensive materials/chemical characterization of these initial damaging processes paired with possible surface treatments' and their impact on electrical performance. In this work, we evaluate the contact resistivity impact of GCIB and the effectiveness of two aqueous cleaning solutions, alkaline and hydrofluoric acid (HF)-based, to remove GCIB-derived residues. The removal of PER and SiGe:B/Si:P damage was characterized by using XPS to study the evolution of the elemental composition and bonding states of the surface. Transmission electron microscopy was used to directly compare structural impacts in high aspect ratio contact trenches. Finally, an optimized cleaning sequence was electrically tested via a transmission line model to determine changes in specific contact resistivity.