For the past month, I've been struggling to reduce the slow leakage of outside air into the reactor. This was first seen in calibration and testing of MR2. After replacing most of the parts in the system, I've concluded that the leak must be from the KF flange where the capacitance pressure gauge is mounted. These fittings with a viton o-ring and are widely used in vacuum systems down to E-8 torr. The issue here is leakage rate rather than base vacuum, and the KF spec for leakage is typically 1.0E-9 mbar-l/sec. That is two orders of magnitude worse than the Conflat fittings used elsewhere in my setup. The attached chart shows the result of a series of long-term tests I did with a variety of different gauges attached directly to the vacuum system T and valve. The leakage measured of around 6 Pa/hr into 30 cc of enclosed volume seems to be of the right magnitude for the KF16 seal to be the source of leakage.
Because of the low hydrogen pressure (300 Pa) of the Mizuno protocol, and the small volume (300 cc) of my reactor, this amount of leakage is not acceptable for a test meant to last days or weeks. Therefore, I have to replace the capacitance gauge and its mount with a part using a VCR (metal face seal) fitting. This has been a painful lesson in vacuum engineering, and I hope my experience will help other replicators avoid the mistake I initially made. The chart also shows that my first Baratron capacitance gauge was damaged by over-pressure (3 bar absolute) during leak testing with hydrogen. Won't do that again!