chemistry of additives in damascene copper plating, The

IBM Journal of Research and Development, Jan 2005 by Vereecken, P M, Binstead, R A, Deligianni, H, Andricacos, P C

Copper plating baths used for forming integrated circuit interconnects typically contain three or four component additive mixtures which facilitate the superfilling of via holes and trench lines during damascene plating. Extensive study over the last two decades has provided researchers with an understanding of the underlying mechanisms. The role of cuprous intermediates in the copper deposition reaction has long been acknowledged, but it is not yet fully understood. In this paper we describe the results of an electrochemical study of the interaction of the organic additives used with copper and copper ions in solution. It is shown that cuprous intermediates near the copper surface affect the overpotential and the kinetics of plating. The additives regulate the presence of cuprous species on the surface; levelers and suppressors inhibit Cu^sup +^ formation, whereas accelerating additives enhance Cu formation. Acceleration by the bis(sodiumsulfopropyl) disulfide (SPS) additive results from accumulation of cuprous complexes near the surface. Adsorbed cuprous thiolate [Cu(I)(S(CH^sub 2^)^sub 3^ SO^sub 3^H)^sub ad^] is formed through interaction of Cu^sup +^ ions and SPS rather than Cu^sup 2+^ and mercaptopropane sulfonic acid (MPS).

Introduction

The introduction of copper as a silicon chip interconnect material in 1997 by IBM resulted in a major change in silicon chip manufacturing [1]. Replacement of the traditional vacuum-deposited aluminum-based interconnects with plated copper-based interconnects was made possible by the integration of a thin liner material which acts as a diffusion barrier between copper and the underlying silicon. Key to the process was the invention and development of the damascene copper electroplating process for on-chip metallization, which was conceived at IBM in the early 1990s. The development of the copper damascene process [1,2] and the subsequent implementation of the copper-based interconnects had a considerable impact on the electrochemical community, with a renewed interest in the mechanism of copper electrodeposition and the role of the organic additives.

Copper-sulfate-based plating processes have been used for some time in the electronics industry for through-hole plating of printed circuit boards. It was found that similar processes could be used to fill micron and submicron inlaid features. The damascene copper plating process uses a thin copper seed which covers the complete wafer surface, including trench and via openings, and acts as the cathode for electroplating the copper. Small amounts of organic additives can increase the plating rate in features relative to the planar surface, when added in the proper concentrations. The observed differential plating kinetics are better known as superfilling, superconformai, or bottom-up plating. Typical copper-sulfate-based electroplating solutions contain small amounts of chloride ions, polyethers such as polyethylene glycol (PEG) and polypropylene glycol (PPG) as a suppressor additive, a sulfur-based organic molecule such as bis(sodiumsulfopropyl) disulfide (SPS) as an accelerator or brightening agent, and in most cases an aromatic nitrogen-based molecule or polymer [e.g., thiourea, benzotriazole (BTA), Janus Green B (JGB)] that acts as a leveling agent to produce mirrorlike plated surfaces. In combination, these additives can achieve accelerated, bottom-up electrodeposition of copper into submicron inlaid features, which permits void-free interconnect wiring in damascene structures.

Superfilling mechanisms

In recent years several models have been proposed in an attempt to describe the role of the organic additives in "superconformai" or "superfilling" plating. In general, differential plating rales inside and outside inlaid features can be obtained either by differential inhibition of the copper plating kinetics by diffusion-adsorption of strong adsorbing molecules such as levelers [2-5] or by differential acceleration of the plating kinetics by accumulation or generation of a catalyst inside the features [4, 6-9]. Schematics of both mechanisms are shown in Figure 1. Which of the two mechanisms prevails depends on the solution composition and the relative concentrations of the additives [4]. A system based on the action of a strong leveler additive such as JGB or BTA [10, 11] can be characterized approximately by the differential inhibition models, whereas systems with only SPS-type accelerators and PEG-type suppressors can be characterized more accurately by accelerator or catalyst models [12-15], For three-component systems with suppressor, accelerator, and leveler, the dominance of a system depends on the relative concentrations of accelerator and leveler.

The leveler-dominated models are based on diffusionadsorption leveling theories for single-component additives adapted to submicron damascene structures. Even though most copper plating baths contain multiple additives, experimental plated profiles from certain commercial baths can be well characterized by the differential inhibition model [4]. Differential inhibition is maintained through the diffusional flux of the leveler into the trench or via as a result of leveler consumption during plating. Characteristic of a leveler-dominated system are rounded profiles and relatively high impurity content in the plated copper because of leveler incorporation. Analysis by secondary ion mass spectroscopy (SIMS) has indeed shown higher C, O, S, and Cl impurities in copper films plated from leveler-dominated systems [4, 11]. Other systems investigated [4] showed poor correlation between plated profiles and modeling of differential inhibition. The authors explained this deviation by the possible accumulation of a catalyst inside the features [4].