The specification on mass flow controller valve closure references the amount of gas that will leak past the control valve when a zero flow set point is given and a specified pressure differential applied across the device. In this case the gas does not leak out of the flow device but past the valve and into the downstream flow path. This is not usually a problem with properly functioning elastomer seal devices since the amount of leakage is essentially zero. With metal seal devices positive shutoff is not possible because the sealing surfaces are metal-to-metal with a limited amount of closure force available. The manufacturers specify an accepted leak through rate, usually in percent of full scale or in mv output (50 mv equals 1% full scale on a 5 volt full scale output device). Since this is actually the measurement of a gas leak through an orifice (the gap in the valve seat) it is also very dependent on the density and pressure of the gas used for the test.
While most manufacturers of metal seal controllers use 1% of full scale as their specification (largely because some of the larger gas panel suppliers have established this as their requirement) there are some models that are unable to meet this specification. In some cases the specification for leakage is not referenced directly at all in their performance data, possibly because it is much higher than their competitors. Most metal seal devices have a statement in their performance specification that the device should not be used as a positive shutoff. This advice would be wise to follow in cases of metering corrosive or reactive gases into downstream reaction chambers as severe system contamination can occur over the life of a relatively small leak in a flow controller valve.
It is usually possible to achieve much better results than the published specification with a new device (depending on the gas density) since the sealing surfaces are new and unblemished by use. As the valve is subjected to the forces and wear of many closures there is inevitable wear on the surfaces that can cause deterioration in the effective closure and a gradual increase in the leakage rate over time. When such a device is returned for service it is possible that, although the closure has degraded appreciably, it will still meet the specification but will not be as good as it was when new. Unless there is some reason to believe that the valve closure required is better than the specification, the device is not adjusted as long as it meets the published specification.
Since the valve leakage spec is referenced as a percent of full scale, the leak rate will be proportional to the full-scale flow of the device. This means that the valve leakage on a 100 sccm device could be 10 times that of a 10 sccm device and still meet factory specs. A test based solely on a specific rate of rise in downstream pressure over time will selectively fail the higher flow device since it is looking for a specific flow rate as opposed to a flow rate at a certain percent of full scale. If your process procedures are setup to pass or fail on valve closure test results that are much lower than the factory spec, or performed at higher differential pressures, expect higher failure rates (even on new devices) and repair costs to replace valve components on older devices with some run time on them.
If you have been using elastomer type devices and are switching to metal seal devices it would be wise to modify your testing procedures for valve closure and install positive shutoff valves up and downstream of your flow devices.
In summary metal seal flow controllers are not positive shutoff devices and are not claimed to be. There are factory specifications for the accepted leak rate that is considered acceptable, although not always experienced, especially on new devices. Lack of knowledge about valve leakage issues and what your process requires can cause undue incoming test failures and repair costs.