(Invited) Assessing Corrosion Risk of Cu Heat-Exchanger Assembled with Cu-P Brazing Alloys

Abstract

Cooling system based on recirculating water has a long history of application in various industrial sectors with its configurational/material simplicity, cost effectiveness and extreme cooling efficiency. While such systems are in use with many variations in system specifics such as the cooling media and component materials, the one with emerging interest is the configuration using microchannel Cu coldplate as a heat-exchanger because of its demand in high performance computing facilities. These facilities have to remove the intense amount of heat generated by a number of CPUs in parallel connection, forcing them to rely on the water based fluid and microchannel Cu coldplate attached to CPU as heat exchanger. One of the critical but less well address concern in such system is the danger of losing reliability by the process of corrosion. The concern exists especially because the Cu coldplate is likely to be subjected to a galvanic corrosion due to the brazing used during the assembly process. The brazing alloy is chemically different from pure Cu and may produce electrochemical potential sufficient to result in accelerated corrosion by the way of galvanic reaction. While such possibility has been noted at early stage of the system development, its risk to the system reliability and practical method of its suppression is not well known with lack of fundamental studies.Motivated by the growing need to better understand the corrosion mechanism active in such system, we have investigated the nature of corrosion in galvanic pair created by the braze. The braze is likely to suffer from galvanic corrosion due to multiple galvanic pairs. As shown in Fig.1-1, where microstructure of braze created by Cu-Ag-P alloy is shown, the braze consists of multiple phases. The inter-phase galvanic potential is expected. Also, the braze and surrounding Cu can form a galvanic pair. The existence of multiple galvanic pairs makes the study of corrosion to be extremely difficult and require new characterization strategies. For this, we have employed two new measurement techniques in order to isolate the galvanic corrosion mechanism from other background corrosion. Using these techniques we carried out extensive studies on braze/Cu corrosion with variation in coolant, filler materials and temperature. The first method was to measure the corrosion current under the configuration of ZRA (zero-resistance-analysis). The ZRA uses the Cu and the joint as the counter/working electrode as shown in Fig.1.2 and measures the galvanic potential and current across the joint. The second is to make a mini-galvanic cell consisting of a pure Cu plate and a Cu plate with the brazed part. Spaced by an O-ring, these two plates are forced to make a galvanic cell with a liquid filled the gap and an external short circuit. We were able to measure the resulting corrosion current, direction and the amount of which can be related to the galvanic corrosion created by the braze.This study presents key findings of ours evidencing that the galvanic corrosion can be better characterized by the use of techniques employed in our study. An example of such evidence can be found in Fig.1-3 where the Cu plate with the braze strip after microcell testing is shown. Note the corrosion happened along the interface between the strip and Cu plate. It is also found that the cell current corelates very well to the degree of corrosion as evidenced in Fig.1-4, enabling quantitative measurement of corrosion rate of the galvanic pair formed between Cu and the braze. A detailed description of such findings along with theoretical explanation will be provide in this paper. Figure 1