Brendan Dunne

Via first approach optimisation for Through Silicon Via applications

Through Silicon Via (TSV) is a very attractive solution for 3D stacking. Currently the main technique in industrial TSV processes is the via-last approach. But the via-first approach has also many advantages and in particular allows the use of high thermal budget materials for high voltage applications.

Mid-process through silicon vias technology using tungsten metallization: Process optimazation and electrical results

Through Silicon Via (TSV) is a one of the more important bricks for 3D stacking and offer different integration approaches. The via-last approach has been first introduced into production. Yet the via-first approach is also currently actively investigated since it has some advantages particularly the use of high conformal deposition materials for isolation and filling of the TSVs enabling higher density of connections or high voltage operations required for certain final product applications. We will show results on process development and integration of 70 ?m deep annular TSVs using tungsten as filling material on a dedicated test chip vehicle. First the complete process flow will be presented. Then, process development work and issues will be addressed. At first we will present developments on the annular trenches opening aiming at favorable slopes and minimum roughness. Deep RIE TSV etching process will be illustrated. For the isolation of the TSV a comparison between SACVD and DHDP deposition oxide will be then discussed. A special focus will be done on W filling sequence using multiple deposition and etch-back steps with different deposition process recipes and a final Chemical Mechanical Polishing (CMP) planarization of the TSVs. The backside process is also presented with the optimization of the back-lapping and CMP process to obtain a stress free silicon surface with no degradation of the TSVs as well as a minimum topology enabling a good back side contact. Backside interconnection is also presented featuring RDL (redistribution layer) and die-to-wafer attach with bumps technology. Then electrical characterizations will be presented. A specific test vehicle was designed to study the TSV density and proximity impact with different number of rings and ring width TSV designs. Daisy chains, specific structures to measure TSV resistance similar to Kelvin structures, interdigitated chains to measure via leakage, and special structures to stress at very high voltage (up to 1000 V), were designed. The electrical results from those specific structures will be discussed.

Finite element method as applied to the study of gratings embedded in complementary metal-oxide semiconductor image sensors

We present a new formulation of the finite element method (FEM) dedicated to the rigorous solution of Maxwells equations and adapted to the calculation of the scalar diffracted field in optoelectronic subwavelength periodic structures [for both transverse electric (TE) and transverse magnetic (TM) polarization cases]. The advantage of this method is that its implementation remains independent of the number of layers in the structure, the number of diffractive patterns, the geometry of the diffractive object, and the properties of materials. The spectral response of large test photodiodes that can legitimately be represented in 2-D has been measured on a dedicated optical bench and compared to the theory. The validity of the model as well as the possibility of conceiving in this way simple processible diffractive spectral filters are discussed.

The Finite Element Method as applied to the calculation of the quantum efficiency in optoelectronic imaging devices

We present a new formulation of the Finite Element Method (FEM) dedicated to the 2D rigorous solving of Maxwell equations adapted to the calculation of the diffracted field in optoelectronic subwavelength structures. The advantage of this method is that its implementation remains independent of the number of layers in the structure, of the number of diffractive patterns, of the geometry of the diffractive object and of the properties of the materials. The spectral response of large test photodiodes that can legitimately be represented in 2D has been measured on a dedicated optical bench and confronted to the theory. The representativeness of the model as well as the possibility of conceiving this way simply processable diffractive spectral filters are discussed.