Trouble with repeatability, please help me diagnose!

[HAZMAT]

I am working to assemble a QC chromatograph to quantify the effects of various hardware (filters, guards, etc) on separation efficiency, tailing, retention time and such.

The Problem: The RT is very stable and peak shape is consistent from run to run, but I’m experiencing trouble achieving repeatability in peak height, peak area and plate counts from run to run. I can run a dozen sample injections literally back-to-back and get upwards of 30% variability in these values. The variation between runs usually follows a rough (but variable) ratio for each peak and parameter. I notice however the ratio changes from peak to peak and from run to run. I can discern no clear trend in the data.

It seems like a sample concentration or sample injection quantity issue. The sample itself isn’t degraded or evaporating between runs.

I’ve rebuilt the injection valve and carefully controlled my injection technique (full loop technique, 3x-5x loading). I’ve replaced seals and checkvalves in the pump to minimize the possibility of uneven flow. I tried changing UV detectors and replacing the lamp. I have tried several sample loops from 1.5ul to 15ul and a range of sensitivity settings on the detector. Also I purged the system exhaustively and have achieved a very smooth baseline.

The rest of the details:
Mobile phase: Isocratic 60/40 ACN/H20
Flow: 1.5ml/min
Column: 4.6mm x 50mm Zorbax 1.8um C18
Sample: Isocratic test mix (diethyl & dimethyl phthalate, biphenyl, terphenyl) in meoh
Pump: Waters 590
Detector: Spectra-Physics Spectra100
Injector: Rheodyne 7125
Data Management: SRI Peaksimple

I'll attempt to post a few example chromatograms ASAP. Also I'll try this http://www.lcresources.com/resources/TSWiz/hs380.htm

[HAZMAT]

Sorry for the large size, it was necessary for the text to be legible

[HAZMAT]

notice how the first two are nearly identical, in fact the plate counts are exactly the same for some of the peaks. Then the values on the last chromatogram are way off.

75 views and no replies so far ... doesn't anyone have a suggestion?

[Uwe Neue]

How about the following: the injector drains your "full loop" partially, so you never or rarely get a full loop injection. Where is the end of your drain pipe that takes away the excess of the sample that you injected? If it is at a very different height than the injection syringe, you may get drainage of the port as you remove the syringe.

[HAZMAT]

Interesting point. My waste tube from the sample loop goes first up several inches and then back down a little, but it is still higher than the injector by a few inches.

My injection technique is to leave the syringe in the injection port until after the valve has been toggled from "load" to "inject". This is to prevent vacuum from withdrawing the needle out of the port from pulling the sample back out of the loop.

Possibly the sample loop is partially siphoning back into the injection port because the waste tube is elevated from the injector. I'll try repositioning the waste tube to the same vertical level as the injector.

Thank you for the input sir!

[Uwe Neue]

OK. leave the syringe in the port. This is the right thing. Also, put the end of the drainage pipe at the same height as the syringe. Let us know if the variability goes away or if it stays.

[Bruce Hamilton]

Firstly, I'd dissolve the sample in the initial mobile phase.
Secondly, I'd try the effect of differing injection speeds, assuming you have manual switching.

I don't know if the 7125 is continuous flow, I don't think it is, so I'd try to control switching speed as fairly fast motion. You may be getting a pressure pulse which affects the valve..

Please keep having fun,

Bruce Hamilton

[tom jupille]

Based on the three chromatograms you showed, the relative standard deviations of the peak areas are around 35%. The relative standard deviations of the area ratios is around 1-2% (still not great, but a whale of a lot better). That says that you definitely have an injection volume problem (which you already knew!).

The 7125 is a "classic" design; it does not involve any type of "make before break" or pressure bypass. All of the suggestions made so far have been good ones. The only other thing I can think of is a bad needle seal in the valve itself allowing aspiration of some air. Rebuilding the valve should have taken care of this, but I've been known to put one together without snugging the seal down tightly enough.

[Consumer Products Guy]

Get your boss to buy an autosampler so you can work on something else !!! Does your boss have pointy hair like Dilbert's ?

[HAZMAT]

Thanks for the great input folks

I'll definitely disassemble the injection valve and take a look at the needle seal. When we rebuilt it, only the rotor seal was replaced. My technique with this valve is to toggle it as quickly and assertively as possible. I still notice a small "pressure bump" at the beginning of each run, though it doesn't show up well at the sensitivity and zoom of the chromatograms above.

I could mix the sample solution I have with 60/40 ACN/H20 ... but it is already dissolved in methanol so I'm going to end up with a very viscous 3-part azeotrope if I do. Is this ok? The sample that I'm working with is a commercially prepared test mix, I've been looking at switching to something that is dissolved in ACN instead. Do you suppose it will make a difference (assuming I remedy this injector problem first)?

I would be delighted if my boss would buy us a complete modular system to work with ... but at least this LC newb is getting some practice at configuring hardware and component troubleshooting. BTW, his hair isn't too pointy

[HW Mueller]

The brochure that comes with Rheodynes has some helpful hints. This is probably from one of them: A leaking needle seal can sometimes be tightened by pushing it a bit into the valve with a blunt stick, or better with this teflon syringe adaptor used for cleaning Rheodynes (one needs to be careful not to damage the outer funnel type section). If liquid comes out while injecting then the seal is bad, and/or one has a blockage. If liquid comes out while inserting the neddle your waste exit is at a wrong level or the seal and rotor are leaking (likely: bad adjustment or bad rotor).
Taking the valve apart doesn´t tell you too much about the needle seal.

[HAZMAT]

very helpful, thanks Mueller

[Hollow]

I could mix the sample solution I have with 60/40 ACN/H20 ... but it is already dissolved in methanol so I'm going to end up with a very viscous 3-part azeotrope if I do. Is this ok? The sample that I'm working with is a commercially prepared test mix, I've been looking at switching to something that is dissolved in ACN instead. Do you suppose it will make a difference (assuming I remedy this injector problem first)?

maybe you can just dilute your test mix with water 50/50, so you get a sample with less solvent strenght than your initial condition, which is then not likely to conduct additional troubles with peak shape.

[Noser222]

You can get a nice used autosampler for $4000.

[thohry]

Simply get a new Rheodyne. That big viariation can probably involve the seal of the valve. An internal leak.