Probabilistic Approach to Assessing Photolithography Quality in the Production of Printed Circuit Boards

Introduction. The current trend in the production of miniaturized electronic devices with improved computing power and performance leads to an increase in the density of interconnections on printed circuit boards (PCBs) and a reduction in the dimensions of such conductive pattern elements, as tracks and gaps, contact pads of components and vias. At the same time, the growing interconnection density decreases the reliability of devices and increases the number of defects in production. In this connection, the development of approaches to quantitative evaluation of the manufacturability of PCB blanks that meet the acceptance criteria represents a relevant research task. A significant share of defects is introduced at the photolithography stage; therefore, an a priori estimation of the number of defects before fabrication and determination of approaches to their reduction are of particular significance. Aim. Development and experimental verification of an analytical model for determining the yield probability of PCB blanks of acceptable quality for the photolithography stage. Materials and methods. An analysis of reasons for emergence of defects in the process of photolithography was conducted. On this basis, the manufacturing parameters that describe the defect value, i.e., conductive pattern distortion and conductor edge roughness, were established. A mathematical model describing the probability of defect-free manufacturing of PCB blanks was proposed. Conductor width and conductor gap size were used as estimated design parameters of PCBs. The quality criteria for the design and acceptance of PCBs were determined based on international standards. Results. A methodology for experimental verification of the proposed probabilistic model by means of processing and statistical analysis of photomask and blank images was developed. Difficulties associated with the creation of datasets and their processing were considered. The adequacy of the model for a laboratory production line was confirmed. For the investigated process, the dependencies of manufacturing parameters on the designed conductor width were determined and the corresponding adjustments of the process were introduced. This allowed the probability of obtaining PCB blanks of acceptable quality to be increased. Conclusion. The results of probability calculations obtained using the proposed model can be used as an indicator of required changes in the design of a printed assembly or for assessing the risks and reserves required by the manufacturer for the production of high-complexity specimens.