Simulating surface tension of Sn-based lead free solder using an artificial neural network

Abstract

Purpose Because of the complexity of relationship between surface tension and its decisive factors, such as temperature, concentration, electronic density, molar atomic volume and electro-negativity, a reasonable predicting model of surface tension of Sn-based solder alloys has not been developed yet. The paper aims to address the surface tension issue that has to be considered if the new lead free solder will be applied for electronics. Design/methodology/approach Using an artificial neural network (ANN) model with back-propagation (BP) algorithm, the surface tension for Sn-based binary solder alloys was simulated, and the comparison between the simulating results and data from experiments and literatures was analyzed as well. In addition, the relationship between surface tension and its decisive factors would be discussed based on the ANN and orthogonal design methods. Findings It is shown that the predicting model of surface tension of Sn-based solder alloys is constructed according to the BP–ANN theory, and the predicted value from the BP–ANN is in excellent agreement with the experimental results. The surface tension of Sn-based solders is determined by five factors, i.e. temperature, concentration, electronic density, molar atomic volume and electro-negativity. Among of the factors, molar atomic volume is major factor, and the order of degree of influence on surface tension is molar atomic volume > electro-negativity > electronic > density > concentration > temperature. Moreover, a simply reasonable equation is proposed to estimate the surface tension for Sn-based solders. Originality/value The five decisive factors of surface tension for Sn-based binary solder alloys have been analyzed theoretically, and a reasonable model of surface tension for Sn-based binary solder alloys is proposed as well. It is shown that ANN theory will be applied well to simulate the surface tension of Sn-based lead free solder.

Prediction of viscosity of ternary tin-based lead-free solder melt using BP neural network

“…Péter and Oliver (2019) used an artificial neural network (ANN), an adaptive neuro-fuzzy inference system, to evaluate the hole filling aspects of solder paste technology, and the predictions obtained help to improve the reliability of "pluggable production." Based on the relationship between the surface tension of lead-free solder and its influencing factors, ANN techniques were used to determine the relationship between surface tension and temperature, elemental composition and physical properties (Wu and Su, 2016). In summary, it can be seen that, although researchers have used the ANN principle to study some properties and soldering processes of lead-free solder alloys, the melt viscosity of ternary tin-based leadfree solders has not been systematically studied.…”