Badger Powder

Low-Cost Powder Dispensing System

Problem statement

Why is it still harder to dispense a few grams of powder than to pipette a microliter of liquid when both are essential steps in materials research? Automated solid-state materials synthesis demands precise, repeatable dispensing of multiple powders, yet powders behave far less predictably than liquids. Their flow depends on particle size, packing density, humidity, and electrostatic charge. Commercial powder dispensers can meet the accuracy requirements, but they typically cost hundreds of thousands of dollars and use permanent, specialized hardware that must be thoroughly cleaned between materials. Even promising research approaches, such as robot-arm spoon dispensing trained with domain randomization, have demonstrated milligram-level accuracy (0.1–0.2 mg error for 5–15 mg targets [1]), yet they still rely on complex learning pipelines and do not provide the low-cost, disposable powder-contact interfaces needed for practical, multi-powder laboratory use. This combination of high cost and limited accessibility leaves most research groups without a workable solution. Assistant Professor So Yeon Kim needs to have access to these dispensing abilities at about 1% accuracy for ideally down to a gram dosing, but does not have the budget for the current common practices. What is needed is a low-cost architecture that separates reusable actuation and measurement hardware from inexpensive disposable components, analogous to pipette tips, while incorporating robust gravimetric error-correction strategies such as fine dosing, overshoot recovery, and adaptive rescaling. We will change the way solid-state synthesis can be done by creating a modular dispensing, weighing, and transporting system within a controlled glovebox environment that delivers accuracy at the commercial level but at a fraction of the cost, enabling more laboratories to pursue autonomous materials discovery and accelerating the development of next generation solid-state compounds. [1] Y. Kadokawa, M. Hamaya, and K. Tanaka, Learning Robotic Powder Weighing from Simulation for Laboratory Automation. 2023. doi: 10.1109/IROS55552.2023.10342463. Links to an external site.

Team members

Dillon O’Brien – admin
Elijah Risberg – facilitator
Rachel Schulz – communicator
Tyler Stevenson – accountant

Client

So Yeon Kim
UW – Mechanical Engineering