Gradient PQ

[John]

We run a gradient PQ on our low pressure systems. This involves running increasing % B; 0,10,50,90 and 100% where B is 0.3% acetone and A is DI water. The baseline is detected at 254nm. We get inconsistent results from run to run, up to 1% difference and the trace is poor (poor plateaux).
1. The manufacturer suggests that the cause of this could be the removal of acetone from B by the in-line vacuum degasser. Is this likely?
2. Does anyone have a better method?
J

[tom jupille]

1. The manufacturer suggests that the cause of this could be the removal of acetone from B by the in-line vacuum degasser. Is this likely?

I don't think so. However, it's easy enough to confirm/disprove: shut off or bypass the degasser and see if the results improve (i'd bet a pizza that they don't ).

2. Does anyone have a better method?

Not necessarily better, but there is nothing sacred about acetone; all you need is a transparent A solution and a B solution with about 1AU worth of background. A dilute solution of HNO3 combined with a shorter wavelength should also work for B. Or you could use A=MeOH and B=0.1% acetone; that would give you a closer vapor pressure match. By the way, λmax for acetone is 265 nm; the fact that your procedure calls for 254 suggests that it's been around for a while. What kind of results have you had in the past?

As a final comment, I would run the 10% steps all through the range. Proportioning anomalies near the midpoint are not unusual with low-pressure mixing systems.

[John]

As a final comment, I would run the 10% steps all through the range. Proportioning anomalies near the midpoint are not unusual with low-pressure mixing systems.

Thanks for your reply Tom. I find it interesting that anomalies occur at midpoint. Why might this be so? My systems are Shimadzu 10ATvps by the way.
J

[tom jupille]

I find it interesting that anomalies occur at midpoint. Why might this be so?

The problem may have been unique to one brand (not the one you are using, by the way). The explanation had to do with the way the control software handled the proportioning valve "latency" (the few milliseconds during which the valve is switching from A to B). Apparently a different proportioning algorithm was used below and above 50%. John Dolan wrote up the problem in one of his "LC Troubleshooting" columns in LC/GC magazine [LC/GC, 14(4), 294 (1996)].

[DR]

A software induced midpoint issue should be systematic in nature. If you have a "less than flat" plateau over several minutes, you probably have hardware issues (pump check valves, seals) and/or need a length of small bore tubing after your detector to maintain a slight, but steady pressure in your flow cell (back pressure regulator).

I also vote against the degasser stripping acetone from the phase.

[stepluke]

Hello John

When you say "poor plateau", what is wrong with them? can you provide a more detailed description.

There can be issues with the acetone concentration in the mobile phase chnaging if the solvent is left "standing" in the degasser for a long time. The solution to this is a good purge of the "tracer" channel before you start the test.

It is also important to make sure that your plateaus are long enough to allow stabilisation of the solvent composition before the chnage to the next step.

[tom jupille]

need a length of small bore tubing after your detector to maintain a slight, but steady pressure in your flow cell

That's a good point (that I had lost sight of ). Many (most?) of today's instruments need a certain amount of back pressure on the pump in order to function properly. The manufacturer should have a spec on that somewhere in their documentation.

[Mark Tracy]

One of my colleagues suggests the use of 30 mg/L caffeine as a substitute for acetone in this test, precisely because of concerns about volatility. I have tried it, and it does indeed make an workable substitute.