Erwin Kessels

ALD Options for Si-integrated Ultrahigh-density Decoupling Capacitors in Pore and Trench Designs

This paper reviews the options of using Atomic Layer Deposition (ALD) in passive and heterogeneous integration. The miniaturization intended by both integration schemes aim at Si-based integration for the former and at die stacking in a compact System-in-Package for the latter. In future Si-based integrated passives a next miniaturization step in trench capacitors requires the use of multiple ‘classical’ MOS layer stacks and the use of so-called high-k dielectrics (based on HfO2, etc.) and novel conductive layers like TiN, etc. to compose MIS and MIM stacks in ‘trench’ and ‘pore’ capacitors with capacitance densities exceeding 200 nF/mm 2 . One of the major challenges in realizing ultrahigh-density trench capacitors is to find an attractive pore lining and filling fabrication technology at reasonable cost and reaction rate as well as low temperature (for back-end processing freedom). As the deposition for the dielectric and conductive layers should be highly uniform, step-conformal and lowtemperature (≤ 400 °C), ALD is an enabling technology here, by virtue of the self-limiting mechanism of this layer-by-layer deposition technique. This article discusses first a few examples of LPCVD deposition of conventional MOS layers with ONO-dielectrics and in situ doped polycrystalline silicon, both as single layers and multilayer stacks. In addition, a few options for ALD deposition of thin dielectric and conductive layers (e.g. HfO2- and TiN-based) will be discussed. The silicon substrates that were used contained high aspect ratio (≥ 20) features with cross-section and spacing of the order of 1 µm.

The ALU + concept: N-type silicon solar cells with surface-passivated screen-printed aluminum-alloyed rear emitter

Aluminum-doped p-type (Al-p⁺) silicon emitters fabricated by means of screen-printing and firing are effectively passivated by plasma-enhanced chemical-vapor deposited (PECVD) amorphous silicon (a-Si) and atomic-layer-deposited (ALD) aluminum oxide (Al<inf>2</inf>O<inf>3</inf>) as well as Al<inf>2</inf>O<inf>3</inf>/SiN<inf>x</inf> stacks, where the silicon nitride (SiN<inf>x</inf>) layer is deposited by PECVD. While the a-Si passivation of the Al-p⁺ emitter results in an emitter saturation current density J<inf>0e</inf> of 246 fA/cm², the Al<inf>2</inf>O<inf>3</inf>/SiNx double layers result in emitter saturation current densities as low as 160 fA/cm², which is the lowest J<inf>0e</inf> reported so far for screen-printed Al-doped p<inf>+</inf> emitters. Moreover, the Al<inf>2</inf>O<inf>3</inf> as well as the Al<inf>2</inf>O<inf>3</inf>/SiNx stacks show an excellent stability during firing in a conveyor belt furnace at 900°C. We implement our newly developed passivated Al-p⁺ emitter into an n⁺np⁺ solar cell structure, the so-called ALU⁺ cell. An independently confirmed conversion efficiency of 20% is achieved on an aperture cell area of 4 cm², clearly demonstrating the high-efficiency potential of our ALU⁺ cell concept.

Optimization of Plasma Enhanced Atomic Layer Deposition Processes for Oxides, Nitrides and Metals in the Oxford Instruments FlexAL Reactor

Hafnium oxide films deposited on silicon wafers from TEMAH and O2 plasma showed saturation at growth rate per cycle of 1.1Aa, which was independent of the plasma conditions. The same film deposited thermally using H2O as the oxidant saturated at 0.8Aa/cycle. By varying the plasma exposure time the compositional ratio of [O]/[Hf], as calculated from RBS measurements, changed from 2.0 to 2.13. The carbon content in plasma HfO2 films was < 2% compared to 8% in thermal HfO2 films. Titanium nitride films deposited on silicon wafers from TiCl4 and N2 / H2 plasma showed saturation at 0.33Aa/cycle, which was independent of plasma conditions and a resistivity of <170µΩ cm at 350{degree sign}C deposition temperature. The stoichiometry of the films can be changed from being slightly nitrogen rich to titanium rich by varying the N:H ratios in the plasma and limiting the amount of nitrogen available for the reaction. The chlorine impurity in TiN varied according to plasma exposure time (2.6% to 1.2%) and N2:H2 gas ratio in the plasma, with a corresponding change in resistivity (200µΩ - 150µΩ).