binary solvent noisy with RI detection

[ljc]

I'm having difficulty getting a stable baseline with an 80:20 mix of ACN:Water (premixed of course) on an 1100 quat pump/RI system. Either eluent alone gives extremely stable baselines of around 5-10 nRIUs peak-to-peak noise, but when mixed at 80:20 (and presumably at other compositions also) it has a peak-to-peak noise of about 300 nRIUs. It is not coincident with pump strokes (very good pressure stability), bypassing the degasser doesn't help, it occurs at various flows, and changing the temp of the column compartment and the RI T setting doesn't seem to help either. It also occurs with the column or with a tubing flow restrictor, so it's not impurities or particulates coming off the column. Finally, I have two identical systems and the behavior follows the eluent i.e. it's not a problem with the Agilent hardware. Is it just that some solvents are just more noise-prone than others when using RI detectors ?

[Chris0000]

HI

It might be a mixing problem if you use a low pressure gradient system. Try to add a additional (static or better dynamic) mixer.

regards Chris

[james little]

why not just premix and run out of one reservoir.

We were running a ternary gradient with two organic solvents and aqueous and never got a real stable baseline with UV detector and with dynamic mixer, so finally just mixed the organic which solved the problem

[ljc]

I did premix, it's essential with RI. But thanks

[Chris0000]

... another problem might be a temperature issue, if one solvent is stored in the fridge....

[HW Mueller]

Possibilities?: Inefficient premixing, and/or inefficient or malfunctioning degasser? What does it look like when you degas (to different extend) with vacuum?
Also, I am not at all sure about this yet, but in my UV detector I seem to see strange (nonlinear?) refractive index or scattering phenomena that might be due to different solvent proportions. Has anybody checked whether there is a parallelism between the viscosity change of different H2O/MeOH mixtures and refractive index?

[HIOAg]

Fist off, Hello everyone, first time posting. Great site here.

We currently run a Agilent 1100 system with a RI detector, mp of 100% organic. And getting the instrument to stabilize with organic solvent usually takes some time. Roughly 2 to 4 hours to see a relatively stable baseline. And we always see a jump in the baseline every time we add more mobile phase to the resevior, even the same lot and manufacturer will cause instability in the baseline.
- Have you tried purging your reference cell? We normally purge for 60mins @ 1mL/min
- Or wrapping the column compartement with some aluminum foil, for temperature stability. We have noticed that we can detect the A/C cycles for the lab in the baseline if we don't keep the column compartment stable.

[Rob Burgess]

Is it just inherent that RID with mixed mobile phases are very noisy?

I remember a couple of years back we developed a method for simple sugars using RID with AcN/water mobile phase and we had the same problem with relatively high noise levels compared to single solvents (with just water) alone.

I'm not sure there is an exact fix for this problem although as the previous post has pointed out purging the cell can sometimes help matters a bit.

[ljc]

Thanks to all for your inputs; I have degassed with vacuum in an ultrasound before putting on the inline degasser; tried everything under the sun to stabilize temp; but yet another observation: when you remove the column and just pump straight into the RI inlet you get a pretty good baseline albeit after some equil time...I don't think the column has gunk coming off it, since I get good baselines with ACN going through it too. The noise appears to have a periodic component of about 30 sec to a minute duration superimposed on a random pattern. The only other parameter that differs with and without the column is pressure: it's around 7 bar without and 50 bar with the column (Kromasil C4); I will put a backpressure coil in place of the column and compare the noise at different pressures.. Stay tuned.

[ljc]

I hate to make a thesis out of this problem, but here goes:

The best baseline I've achieved with the 80:20 ACN:H2O is obtained by setting the column oven to 25 deg and the detector at 30 deg or higher, and routing the effluent through one of the heater blocks before the detector. Having the column hotter than the detector is disastrous, about 10x the noise. Conncecting the column outlet directly to the RI doesn't help either.

The Agilent Chemstation permits a readout of the actual temperatures of the left and right column compartment and the RI detector optical unit temperature. You can plot these values along with the RI signal directly on the screen. The period of the ripple seen on the RI signal (about 50 nRIu peak to peak) is almost equal to the T fluctuation in both column compartment readings. I don't think this is a coincidence, even though the column T fluctuation is only +/- 0.1 deg C. The T readout of the RI detector shows that it only varies by about 0.01 deg C. It might be that the actual temp of the solvent going through the flowcell isn't completely stabilized by the RI unit at 1 mL/min. What is still puzzling is why this effect is so pronounced with the binary solvent, and not with straight water or acetonitrile. Put another way, it appears as though the temperature dependency of RI is less in a pure solvent than in a mixture. The answer to this will be in JACS or ANAL CHEM somewhere, as some physical chemist somewhere had to have studied this phenomenon. If so, there might be an optimum mixture that has the smallest temperature dependence. But Murphy will say that this mixture is the WORST for doing the separation !!!

On to Sci Finder....

[ivanvins]

The RI detector is extremely sensitive to any variations in the composition of the mobile phase (0,001% or less change in concentration can be observed). Just a hint to the difference between pure and mixed solvents - the equilibrium of water/acetonitrile on the stationary phase will be dependent on temperature and you may observe the real changes of efluent composition induced by the temperature. Try to switch off the thermostat and insulate the column - the baseline will have increased drift but the noise may decrease.

[ljc]

Your suggestion of simply turning off the column oven did exactly what you said. The +/- 0.1 deg T oscillation with the oven "on" was being seen by the RI detector as a sinusoidal wave having the same frequency. This occurs with the binary solvent, but not with a pure solvent. Turning off the oven eliminated the short term noise. Long term drift is present but with less amplitude than the short term noise I was getting. The fact that this is necessary with the binary but not the pure solvent is still intriguing. I agree that it must be that the T change causes a very small change in the equilibrium composition of the eluent eluting from the column, because the detector should be able to stabilize the temperature. Also, I notice that when you change the flow rate, it takes about a column volume before you see the baseline shift. Thanks again for you suggestion.

[DR]

So this leaves you with 2 options -
Keep your run times short enough that a little drift in the baseline won't ruin your day or buy a good forced air column oven and run far enough RT that it will be on more than it is off. Initial equilibration times may be slower, but I think that a forced air one will minimize the amplitude of the wave you see with the heating block type oven.

[ljc]

DR, It would be ironic if an older 1090 with the forced air heater would outperform the 1100 Peltier type for this application ! It's tempting to try. The 1100 RI "help" menu states that if the room T is constant to within 2 deg C, turning off the RI optical temp control would further enhance baseline stability. The readout from it's T controller shows that it is within 0.01 Deg of the setpoint, so I'm not sure I understand how turning it off would help, unless they're talking about stability way beyonod what I need.