Search

With increasing thermal loads in electronic packages, liquid cooling is a preferred option for superior cooling capability. Closed loop liquid cooling systems with microchannel cold plates have demonstrated superior thermal performance. The widespread adaptation of this technology, however, is subject to long term reliability concerns, especially caused by corrosion with a water-based coolant. Galvanic corrosion, which is caused by the difference in electrode potentials is observed to be the dominant failure mechanism.In this paper, the design and development of flow loops suitable for capturing the corrosion behavior within the loop is discussed. Careful consideration of all components within the loop and the rationales behind the choices are highlighted. The experimental methodology details various sensors and measurements required to assess corrosion development in the loop. Experimental results from the flow loop tests are presented and engineering recommendations are made for flow loop design, choice of working fluid and measurements to be made in-situ. Although there is a clear indication of corrosion progression based on the experimental data of fluid chemistry degradation, no significant impact is observed on thermal performance of the current microchannel cold plate design for the duration of the current test conditions.

This paper reports the corrosion mechanism active in microchannel cold plates used in a liquid cooling system and proposes a kinetic model describing the rate of corrosion-induced failure as a function of testing conditions. The corrosion failure mechanism investigated in this paper is galvanic corrosion because the cold plate is typically made of Cu and is assembled using brazing alloys and there exists a galvanic potential between the Cu and the brazed area. A series of experimental characterizations indicates that the brazed joint is subjected to galvanic attack when exposed to a coolant, a mixture of water and propylene glycol (PG), with a galvanic potential sufficient to dissolve the braze component with an accelerated rate. Various testing on the galvanic corrosion finds that the braze (a ternary alloy of Cu, Ag, and P) becomes an anode in the galvanic pair and loses the component element by the process of dissolution. This type of galvanic corrosion is found to exist even with corrosion inhibitors present in the coolant, necessitating the corrosion assessment methodology that can predict the rate of cold plate failure with the use of the "accelerated testing" and the prediction model. Our research leads to the development of the micro-galvanic cell testing as well as the zero resistance ammeter (ZRA) methodologies. Our investigation with these testing methodologies presents clear evidence showing that the galvanic corrosion is the most active and serious form of corrosion in the cold-plate with the galvanic pair exerting as high as ~0.3V anodic potential on the brazed joint. It is also found that the rate of corrosion can be further accelerated with temperature and the external potential purposely applied across the Cu and the braze. The resulting galvanic corrosion kinetics collected in the form of current may be used to predict the corrosion rate at use conditions as they are found to follow the form of an Arrhenius-type kinetics model with a consideration of the corrosion acceleration factor.