Help on ghost peak please

[moonchips]

28 min ghost peak:

I got a ghost peak in my chromatogram.
The gradient is:

Time (min) % Water % ACN Flow (mL/min)
0 50 50 1
11 0 100 1
25 0 100 1
30 50 50 1
35 50 50 1

It has fixed retention time of 28 min, and very nice peak shape. Its peak area (more than 4000) is independent on sample injection volume, sample concentration and sample solvent species. Its peak area is dependent on the purity of my acetonitrile mobile phase.
(HPLC grade, Peak area 4000, gradient grade, Peak area 400)

What are ghost peak origins?
1. Sample: Injecting different samples, different solvents should get rid of it. So NO.
2. pH electrode : NO, I did not adjust pH.
3. Water: NO, I used different water
4. Mobile phase?: should retention time be random? ACN mobile phase purity does affect its peak area. (most likely)
5. Column?: Peak area should change with washing, and its not observed that peak area is dependent on washing. I only have one column and it is hard to tell.
6. Injector?: by pass column, it should be there. I tried using a connector and it is there, but the area decreases. I tried 0 mL injection and it is still there.

[Bruce Hamilton]

Well, you have covered several of the possible sources.

Because it appears as you go from the 100% acetonitrile back to 50%, and presumably are using HPLC grade solvents, I have a couple of suggestions you could try.

It would be nice to know the HPLC system, column type and temperature, sample solvents ( should be similar to mobile phase ) and the wavelength you are monitoring. If you have a DAD or variable wavelength detector, does the peak's UV spectrum match any of your analytes?. I assume that the system is in good condition, and piston seals aren't leaking, solvents are of suitable grades, and samples always give a clear solution in solvent and mobile phase.

Have you allowed the system to condition at 50% for a while, then performed a few rapid injections. Does the peak area get smaller after the short equilibrium times. If so, it's likely to be from your mobile phase, mobile phase filters, or degasser.

Clean or replace the solvent and pump filters, using ultrasonication in warm water forllowed by acetonitrile.

Flush the whole system ( minus the column ), including both the acetonitrile and water solvent systems, especially the degasser ( if your system has one ) with water for about 30 minutes ( more if you have used use ion pair modifiers - including TFA - in your mobile phase ), and also flush the column with 95% water, heating it slightly if possible.

Then perform some large volume injections of 95:5 water/acetonitrile to rinse the injector. See if the peak gets smaller as you perform multiple injections.

If using a ternary or quaternary pumping system, ensure your gradient mixing valve isn't leaking one of the other mobile phases - just change them to water to test.

Please keep keep having fun,

Bruce Hamilton

[tom jupille]

You have done a very thorough job of checking out problems!

One last test: does the size of the peak vary with equilibration time before the gradient? If it does, then your ghost peak is coming from the A reservoir. Because you are seeing a variation with ACN purity, I would agree with you that the ACN is the most likely source.

This is not an unusual situation. Look at the this thread (and the other threads referenced there) for more information and discussion: http://www.sepsci.com/chromforum/viewtopic.php?t=4689

[moonchips]

Agilent 1100 HPLC system
column: ODS Hypersil 150mm*4.6mm*5um
temperature: 60 C
sample solvent: 70/30 isopropanol/water
wavelength: 200 nm

I used another Agilent 1100 HPLC system and 28 min peak is still there, which can rule out injector, HPLC system and this 28 min peak should be coming from either the column or the ACN.

I do not understand. Its peak area is huge with HPLC grade ACN(400-6000 depending on the grade of ACN). What could it be in HPLC grade ACN? Could this be due to my gradient? is it posible it is column bleeding?(I worked with GC before and not sure if this is the right term for HPLC)

Thanks

[tom jupille]

It just struck me that if you're new to HPLC, you may be misinterpreting what you are seeing.

First of all, how high is the peak (in absorbance units)?

Second, how wide is the peak (in seconds)?

Baseline drift and noise are very common (in fact, difficult to avoid) in gradients.

Assuming that it is a peak (rather than the classic "mid gradient hump"), with detection at 200 nm you could be seeing almost anything that has a double bond. I've seen instances of contamination by vapor from acetone being used in the same room, and had reports of contamination from residual detergent left on incompletely rinsed glassware.

Here's a link to the appropriate section in The HPLC Troubleshooting Wizard on our web site: http://www.lcresources.com/resources/TSWiz/hs400.htm

[moonchips]

Thanks a lot! Tom. The link is very useful.
I will do the equilibration time test.

Peak high is 60 (gradient grade ACN) to 800 (HPLC grade ACN) mAU
Peak area is 400 (gradient grade ACN) to 6000 (HPLC grade ACN)
base line width is about 30 seconds, half width is about 3 second.

[Bruce Hamilton]

I don't understand why the peak would appear at that point of the gradient when the column wasn't there if it was impurities in the acetonitrile, so I'm wondering if you are observing a thermal or refractive index effect in the detector from the relatively high column temperature.

You also might like to consider some runs with the column at 50, 40, 30C, and determining if the peak changes in area or retention. Do you know the history of the column?.

At 200 nm, the obvious issue would be the grade of acetonitrile, or any contamination you or the instrument may be introducing. If not already doing this, try using the acetonitrile direct from the manufacter's bottle - no filtering or transfer through intermediate dispensing devices.

If you have a UV spectrophotometer handy you can check the UV absorption of samples from your HPLC mobile phase pump outlet and the suppliers container against specification, and also at 200nm.

If your results are higher than the manufacturer's specification or certificate, you may have to He purge or degas samples before retesting to ensured dissolved gases are removed. As asked earlier, have you looked at the UV spectrumof the peak?.

In earlier threads, I've recommended having HPLC solvents from two different known mnufacturers ( rather than repackagers ), as it's likely that products grades from the same manufacturer will have the same impurities, just at different concentrations.

Bruce Hamilton

[moonchips]

Hi Bruce.
I actually ruled out mobile phase solvent when I got fixed Rt.
Should Rt be random if the peak is due to impurities in mobile phase?

Then I thought maybe something in the mobile phase is so sticky and it is cumulated at the column head. Only my gradient can flush it out. I modified my gradient.

old gradient:
Time (min) % Water % ACN Flow (mL/min)
0 50 50 1
11 0 100 1
25 0 100 1
30 50 50 1
35 50 50 1

new gradient:
Time (min) % Water % ACN Flow (mL/min)
0 50 50 1
11 0 100 1
35 0 100 1
(no peak aat 28 min)
40 50 50 1
(two peaks showed up around 38 mins)

So it is not impurities. It is gradient.

I was told that it is possible that changing from 100 ACN to 50/50 water from 25 min to 30 min (within such a short time), heat can be released and absorbance can change. Is this thermal effect? If it is, how come purer solvent has less effect?

I am going to try temperature, wavelength and UV absorbance as you suggested.
I did not get spectrum for this peak yet.

Tom, I tried equilibration time.
Peak areas are 6400, 6400, 6800. It is higher when equilibration time is 15 min (6800) than 5 min (6400, 6400).

Thank you all

[tom jupille]

It is higher when equilibration time is 15 min (6800) than 5 min (6400, 6400).

If the peak were coming from contaminated A solvent, the area should triple, so that eliminates that possibility.

[Bruce Hamilton]

I'd expect a retained solvent impurity peak to have consistent retention in the gradient.

The problem is that your peak appears 3 minutes on both gradients, after the start of the gradient return from 100% to 50% B over 5 minutes, suggesting it's not a solvent impurity, which Tom has also confirmed from the peak areas.

I'd suggest trying to change the rate of decline from 10%B/min to 5%B/minute ( ie change from 5 to 10 minutes ), and consider the effect on the peak, in both retention time and area. Note that 10% change/min. is not extreme, and should have no effect at typical wavelengths, but 200 nm may have some surprises, especially if your D2 lamp is nearing the end of it's life.

Another possible item to consider is whether you are using a reference wavelengthon the DAD, and what your spectral bandwidth setting is.
These may help explain the different results from the two grades of acetonitrile.

I suspect that the peak is a consequence of the column being at 60C, rather than the 10% gradient alone, but some of the suggested simple experiments should reveal all.

Please keep having fun,

Bruce Hamilton

[moonchips]

Hi Bruce
This peak is gradient-related.

I tried 10% B/min, 5% B/min, 1% B/min, then come back to 10% B/min.
Peak area 3400-4100-3700-4800
Peak height 294-266-140-463

I tried 60, 50, 30 C column temperature
Peak area 3400-3700-3900
Peak height 294-283-251

I tried to change Column #1 to a connector then to Column #2
Column #1, Thermo ODS Hypersil, 150mm*4.6mm*5um
Column #2, Agilent SB-C18, 150mm*4.6mm*5um
The peak is there and peak area (Column #1>Column #2>connector)

Could you please explain thermal effect briefly? How is refractive index change affecting VWD detector?
Is it harmful to my column? How to get rid of it? So far, purer ACN can change peak area 20 times (from 6000 to 300)

I am using VWD. Please elaborate: “Another possible item to consider is whether you are using a reference wavelengthon the DAD, and what your spectral bandwidth setting is.
These may help explain the different results from the two grades of acetonitrile.â€

[Bruce Hamilton]

OK, I'll have to check my 1100 VWD manual tomorrow, as I don't use it often enough to know if it has a reference wavelength setting, as I was assuming a DAD.

[ Update - From a quick look at the manual, there doesn't seem to be an option to define the reference wavelength on the VWD, it's automatically subtracted from the sigmal in the sefault mode, and users only have choices of recording the signal alone or reference alone. ]

The main purpose of the reference wavelength is to compensate for the changing refractive index of the mobile phase on the detector during a gradient. The RI can cause baseline drift.

The latest results are interesting, did the peak retention time and width change with the gradient, the peak heights suggest they may have ?.

The peak area increase with decreasing column temperature suggest that perhaps the peak may be partially caused by dissolved gases becoming undissolved as the gradient progresses.

That could also explain why your gradient grade Acetonitrile gives a lower peak area - it may just have lower quantity of dissolved gases than the HPLC grade.

How different is the purity of your acetonitrile grades?. Most HPLC grades are usually high purity, and the gradient grades tend to have less impurities that affect a gradient, but the difference in absorbance usually isn't great, eg 1 versus 3 mAU at 210nm, or 0.2 versus 0.5 mAU at 254nm, during a test gradient with water.

I would try using supplemental degassing of both components of the mobile phase ( helium sparging, ultrasonication, or vacuum ), and also try preblending the 50:50 mobile phase, degassing it, and putting that as component A, and changing the gradient programme to compensate - so that you still have the same gradient.

If the peak get smaller, then it's likely that dissolved gases or refractive index changes are having an effect.

It is very unlikely the peak will harm your instrument or column, and you may find that investigating the causes is more trouble than it's worth.

Please keep having fun,

Bruce Hamilton

[luna_leo]

An interesting thing is: If we change Acetonitrile to Methanol in the gradient, the ghhost peak disappears. we know that mixing of methanol with water is exothermic while mixing of acetonitrile with water is endothermic. how to explain this phenominen?

[Bruce Hamilton]

The mixing should all occur at the pump, and your column is at 60C, which is close to the boiling point of methanol. That would be more of a concern to me than the thermal effects of pump mixing, as the temperature changes could affect dissolved gas solubility as well as refractive index.

The problem with using methanol at 200 nm is that it usually absorbs most of the energy, so assessing the ghost's behaviour may be very difficult, and the gradient baseline changes should be significantly larger.

However, if using methanol solves your problems, that's good.

Please keep having fun,

Bruce Hamilton