Selective Deposition of Thin Copper Films onto Silicon with Improved Adhesion

Metallization plays a key role in the production process of integrated devices. Recently, copper (Cu) has been proposed as an alternative material to aluminum to address the need for metallic thin films with low resistivity and high electromigration resistance. 1 As a consequence, great attention has been paid to electrochemical processes, which enable the deposition of metallic copper at low temperatures with low costs. Copper electroless deposition has been shown to selectively plate silicon (Si) substrates and structures with high aspect ratios and large structural heights. 2,3 However, this method requires a pretreatment to activate the surface. Selectivity is achieved only to the extent that the activating treatment can be made selectively. In addition, many studies have also been carried out on the galvanic deposition of copper from fluoride containing solutions.4-11 In contrast to electroless deposition, galvanic displacement deposition requires no prior activation of the surface and is truly selective to silicon surfaces. Thus, it provides an attractive deposition method for copper interconnects or seed layers for subsequent metallization.4 Galvanic displacement is also a promising avenue for the integration of metals in micromechanical devices, due to its conformal nature and high substrate selectivity. Despite these attractive features, a crucial issue in the aforementioned processes is the lack of adhesion of copper to the Si substrate, which may severely constrain their application. 12 In particular, copper films deposited by galvanic displacement fail the qualitative Scotch tape test. In this paper, we report a process for the galvanic deposition of copper onto silicon from fluoride-containing solutions. Thin copper films with reflective and smooth surfaces and excellent adhesion to silicon are obtained. Results on plating of microelectromechanical systems (MEMS) after release are also presented. Polycrystalline silicon and single crystalline Si(100) and Si(111), p- or n-type, and analytical grade chemicals were used. Samples were ultrasonicated in acetone and, after drying with nitrogen flux, etched in concentrated hydrofluoric acid for 10 min. The hydrogenterminated surfaces thus obtained were rinsed, dried, and immersed in the plating solution. The following additives were used to prepare the aqueous solution: ammonium fluoride (NH 4 F 40%) 50 vol %, copper sulfate (CuSO 4 ·5H 2 O) 0.01 M, ascorbic acid (C 6 H 8 O 6 ) 0.01 M, sodium potassium tartrate (KNaC 4 H 4 O 6 ·4H 2 O) 0.005 M, and methanol 30 vol % (percentages are referred to the final solution volume).13 The pH of the solution was 7.5. Room temperature (25°C) and gentle agitation of samples were used. Samples were finally rinsed in deionized water and dried with nitrogen flux. The adhesion of the copper film to the substrate is strongly related to the presence of ascorbic acid in solution. The absence of the acid results in the formation of a copper film which fails the standard scotch tape test. Because we have observed a similar effect when ascorbic acid was substituted with fumaric acid, and both acids are