I have been trying to set up a gradient HPLC method developed somewhere else. My modified method (VR) seemed to work when 5 ul of standard mix (5 compounds including the internal std) was injected. But the increased injection volume (10 ul, 20 ul, 40 ul, 60 ul) introduces ghost peaks. When I use the original method (AF) with injection volume of 20 ul I do not observe ghost peaks. The original AF method uses the gradient:
5%B:95 %A for 5 min
5 to 45%B in 20 min linear
45-70%B in 6 min linear
70-5%B in 3 min linear
5%B for 15 min, where A= 10% acetic acid:90 % water
and B= Methanol:ACN:dichloromethane (10:5:1).

My modified VR method uses the gradient:
Start with 20%B:80 %A,
20 to 50%B in 20 min linear
50 to 20%B in 3 min linear
20%B for 5 min, where A=5% acetic acid:95 % water and B=Methanol:ACN:dichloromethane (10:5:1).

I use Waters Alliance HPLC system with 996 PDA detector (use data at 260 nm). The column is Nova Pak C18, 3.9x150 mm, 4 um. The ghost peak appears even with 3.9x300 mm column (20 ul injected).
Ghost peaks do not stem from the previous gradient run, neither from impurities in std as no such peaks appear when all 5 compounds are run separately. The gradient delay is approx. 3 min.

I am attaching ANDI (.cdf) files of following chromatograms:

VR7793.cdf vr7793.cdf (7 k)

VR7801.cdf vr7801.cdf (7 k)

VR7809.cdf vr7809.cdf (7 k)

VR7814.cdf vr7814.cdf (7 k)

VR7819.cdf vr7819.cdf (7 k)

VR7825.cdf vr7825.cdf (14 k)

Good Day,

Without looking at your chromatograms, I would guess that the solvent your standards/samples are made in is stronger than the initial conditions of your gradient. As you increase the injection volume you are seeing micro chromatography affects as the solvent slug works its way through the column giving what appears to be peaks but are really shoulders of your compounds of interest. To test to see if this is your problem, make a set of standards of different concentrations (equivilent in conc. to your various volumes) and inject the same volume (10 or 20 uL)for each. If this is the problem and you really want to inject various volumes, make standards and samples up in mobile phase or a solvent weaker that your initial conditions.

Good Luck

I took a look at your 5-ul and 60-uL chromatograms using the "compare utility" that comes with chrom merge. If I adjust the vertical axis by 12x to compensate for the difference in injection volume, there are really only three significant differences:

- a misshapen peak appears at about 6 min in the 60-uL chromatogram
- the peak at 11 minutes becomes disproportionately large and comes out somewhat earlier in the 60-uL chromatogram
- a very broad peak centered at about 18 minutes appears in the 60-uL chromatogram (but the baseline in this region is no great shakes on the 5-uL run either!).

My first suspect for the fat guy at 18 minutes would have been CRUD from a previous injection, but you've eliminate that already.

What happens if you inject a 60 uL blank (i.e., same dilution solvent, but no standards)?
If you get the same pattern of ghost peaks, then they probably stem from CRUD in the dilution solvent.
If you *don't* get the same pattern, then that points in the direction of solvent effects as suggested in the previous post.

-- Tom Jupille / LC Resources Inc.

Another possibility could arise from late eluting peaks, because you changed the method and finish at 50%B instead as earlier at 70%B.

Are you sure, you need the dichloromethane in your eluent? This could make any kind of trouble: it is volatile and can disappear or even being retarded from the c18 phase under certain conditions. Do you have a method development report to check, why you have it? Or is it according to the recipe of "grandmothers ham" (it means, nobody knows, why it is, like it is)?

I have tried to apply your advice to solve my ghost peak problem.
60 ul of blank did not show any ghost peaks so the solvent strength of my std mix was to be blamed next.
As regards “grandmother´s ham recipe” I tried to exclude dichloromethane from mobile phase B since no method development report is available to me. Evidently dichloromethane was essential for my internal std (last eluting peak) which became retained on the column and even the elution order of earlier eluting peaks was changed.
I compared calculated percentage of organic solvent in my std mix with that of starting gradient mobile phase. Conclusion was that I have around 69% organic solvent in the std mix compared to 20% in the starting gradient MP. I was partly aware of this but the problem was that my individual stock std solutions (around 1mmol/L) were dissolved in 100% methanol. And even if I diluted them in mobile phase as I did it was practically impossible to reach the solvent strength as low as in my mobile phase. In that case I would need more concentrated stock solutions , I guess. But what has confused me and still is a mystery is that the std mix diluted with mobile phase from the original method (AF) has around 63% organic versus 5% in the mobile phase. And I did not observe any ghost peaks with 20 resp 40 ul injected.

So how have I solved the problem? I have decided to change the gradient so that it is closer to the original method but I can keep a short analysis time. My new gradient starts with 10%B: 90%A,
10% to 55%B in 20 min linear
55% to10%B in 3 min linear,
10% B for 7 min.
My analysis time is in total up to 30 min and the most important thing is that there are no ghost peaks present even if I inject 20 resp 40 ul of the same std mix as before. Some small front tailing of only the first peak can be seen with 60 ul injection of the most concentrated std mix. Some hump (a broad peak) can still be observed between the last two peaks at 260 nm but is not that obvious at 280 nm. Hopefully it works.

Once more I appreciate your advice.

I have recently had some problems perhaps related (or not)to ghost peaks. The mobile phase was acetonitrile: methanol: acetic acid 1M (33:7:60). My standards were prepared in methanol and the injection volume was 25 ul. Separation of the peak of interest was fine and also the detectors responses (UV and fluorescence) were linear. My surprise came when I injected 25 ul of methanol or acetonitrile: a small and sharp peak could be detected with both detection systems at the same retention time of my peak. This happens even after repeated injections of methanol or acetonitrile. I have already discarded any contamination of either the injector (has been washed with methanol)or the column (has been flushed with 100% acetonitrile). Could the answer for this peak be also in the fact that the solvent, methanol, is stronger than the mobile phase?