Optimizing Aerosol Jet Printing for SAW Sensor Fabrication for Nuclear Applications

If nuclear is to move forward and remain competitive in a diversifying energy economy, it must be able to shed its analogue systems in favor of digital control. Recent computational hardware developments have brought forth renewed interest in machine learning and digital twins as methods for safely bringing nuclear to digital operation. However, such technologies require large volumes of data which can be hard to retrieve due to the harsh conditions generated by nuclear reactors.

Surface acoustic wave (SAW) sensors have shown promise as environmentally robust wireless sensors capable of measuring a wide range of parameters including temperature, chemical, pressure, structural, radiation dose, and more. But like any advanced technology for nuclear, they must undergo rigorous testing and research before they can be widely used in reactor systems. Traditional manufacturing methods for micro-electronics like SAW devices include the use of lithography techniques, typically requiring clean rooms for operation and incurring high equipment costs which can disqualify most university research labs from even attempting prototyping of SAW sensors.

Additive manufacturing techniques such as aerosol jet printing (AJP) have been utilized in recent years to produce SAW sensors and similar devices, making it a strong candidate for university lab prototyping. While exciting, AJP is still an emerging field and presents its own set of challenges. Difficulties maintaining print consistency and an inability to utilize new inks without additional extensive testing hinder the technology.

In this thesis, large datasets were created to explore the capabilities of the commercial Nanojet AJP system and two separate silver inks, while making attempts to maximize quality for prints made at a 10-micron resolution. The data considered commonly tested parameters of flow rates, print speeds, bed temperature, and nozzle height. Additional considerations such as process drift variables were incorporated into later datasets as they were encountered. The extensive parameter testing lead to successful printing of at least one 10-micron high quality print. Future work would involve further parameter testing and incorporation of more complex methods of feature identification and quantification. Machine learning algorithms may be utilized for the creation of automated process monitoring as well. Additionally, design and testing of printed SAW devices is warranted.