I wanted to share my latest experience on hunting leaks on a GC-MS coupled to a thermal desorption unit with the community
Please comment if needed.
Let me first describe the system according to the carrier gas (He) flow:
main He outlet (@70psi) - gas clean filter - big universal trap - splitter - pressure reducing valve (@10psi) - thermal desorption unit - transfer line - ultimate union - column (the transfer line connects to the column inside the GC oven) - MSD. There are a lot of points for potential leaks.
I have not been managing to use the system because I was getting about 20-30% of 28/69 when doing autotune and I could not manage to find any leaks. I don't have an 'air duster' spray or He detector and I was doubting about my ability to make the CFT connections for the ultimate union.
So an Agilent Crosslab engineer came to the rescue. He found some small leaks only with Snoop and left the system with an acceptable 28/69 percentage. The next day, that value was off again.
Before I reach to him again, here is was I have been doing:
- load a maintenance method where the heated zone of the MSD are at 100ºC (cool.U),
- fast vent the MSD until Turbo Pump Speed is >40%,
- remove the column from the MSD and cap the transfer line (quickly before pump speed is at zero),
- Pump Down and check for leaks (tune evaluation or manual tune),
- reduce the entire system pressure to 10psi at the main He outlet and set the pressure reducing valve on full open (not reducing anything),
- connect the MSD side of the column to the GC injector (sounds weird but I wanted to use the EPC of the GC as a pressure transducer),
- place the ultimate union in isopropanol and check for bubbles,
- switch off carrier gas at main He outlet and let pressure stabilize,
- record pressure of GC inlet over time.
So far, I have a pressure drop of 0.12psi/hour for the entire system. I don't know if it is a lot and/or enough to be a problem for the MSD air leak detected.
I'll switch off valves to try to isolate where the leak could be.
I still haven't discarded the column as the problem. Crazy idea: can I put the column in an appropriate (?) liquid to check for bubbles? How often is the column the culprit for a leak?