We have a method which has been used for five years without any complication. Lately we have been getting a peak showing up randomly through out a run (1 out of 40 injections). The peak comes out at different times in the run. All peaks that would come out after the time of the mysterey peak are missing and there is just a plain baseline. The next injection would be fine. The same thing has happened on different machines with different buffer, acetonitrile and columns. It is a gradient.

Thanks for any help

Does the shape of the baseline remain "normal" after the mystery peak? One possible explanation that I can think of is a malfunction in your solvent delivery system that puts a slug of strong solvent through, eluting all remaining compounds as a single peak (basically, it amounts to a step gradient at that point). I would be surprised to see this happening on multiple instruments (although, if it's random 2.5% of the time, and the instruments are similar and have similar service histories, it's not impossible). If this *is* the cause, I would expect to see an offset of the baseline when it occurs.

It´s almost always quite instructive to collect and reinject phantoms (or analyse them by another technique).
Repeating myself: It would be helpful to be informed of solutions to such problems.

The samples that had the peak were reinjected and did not show the peak on the next injection.

The baseline after the peak remained normal. I had thought there might be a surge of acetonitrile causing the peaks but the baseline did not ramp up as it would in a gradiant also the peaks immediatly preceding the mystry peak were at the correct retention time. The peaks on the following injection appeared to have the typical retention time. I had thought the effect of a surge would linger a little longer.

Also, the odds of two instruments having the same problem, even though they are the same PE model and maintained the same, were too great.

After further investigation we had one injection which the peak appeared at 5 min and 30 injections latter it appeared at 19 min. The peak shape was unique since it had coeluting peaks but both injectioned were identical. The injection following the one with the 19 min peak gave a peak with a 30 sec. shift in retention time (run time 21 min). This does lean toward a surge in acetonitrile. But, could the peaks be large from just a peak of acetonitrile. Since the peaks are identical the theory of the late peaks are all coming out at once is blown.

Perhaps the peak is a collection of impurities on the head of the column? If this was a surge wouldn't effect the peak immediatly before a little. Doesn't it take a little time to change the phase of the column or am I thinking about this wrong?

If this is the case, what is the proabibility of two seperate systems, same manufacturer, having the same problem after years of running without error.

Non polar impurities in your water phase may emigrate slowly through the column with an increasing the AcN in the mobile phase and finally elute some day. Altering the reagent’s supplier, changes in the Milli-Q behavior, or just the glassware’s cleaning procedure, may compromise a gradient method, which has been used for five years without any complication.
After a few normal runs you can increase the percent AcN after the end of the gradient and see what is washing out.

To I.D. the peak is not occuring at the same retention time. This is not like the ghost peaks that show up on a gradiant ramp. This peak will show up at the beginning, middle and the end.

Given that you have observed the problem on multiple systems, it would seem to point to a chromatographic issue. If it is contamination that is heavily retained (and finally eluting sometime in the future), floating retention time for this phantom peak is not unexpected. Try adding a 5 min flush of 100% MeCN at the end of the run, and see if you can get through 40 runs without seeing the ghost peak

Hopefully, we are not wracking our brains over semantics here. Are you saying that after (later RT) the phantom there are never any of the expected peaks coming? Ahead (earlier RT) of the phantom all expected peaks are always present, with the expected areas? Also. are you saying that on reinjection you did not get any peaks at all (except system peaks, of course)?
I am sorry, but your (anonymous) statements of May 16 are uninterpretable for me.

I wouldn’t expect a peak with RT of around 14 hours (40 runs x 21 minutes), plus minimum 40’ between all injections, to elute at the same retention time. Our systems are not ideal at all. Even small deviations in the time between injections (worst case is an integrator, where the integration and reporting times depend on the number of peaks) is more than enough. The peak doesn’t show up on a gradient ramp, it shows up on in 40 gradient ramps. I will not be surprised if you observe a small but permanent drift in peaks’ retention times, cycling 40 injections. If so - Merlin’s suggestion is a good one, and may be even better than washing the column after each 39 runs.
One more thing – I never use commercial HPLC water when running a MeCN gradient method, and I always rewash the glassware supposed to be in touch with the mobile phase.

To HW Mueller, You are correct in your statments including the reinjection.

If the mystery peak seen is just some impurities that have collect at the head of the column. That is not a problem. The peak showing up is just a symptom of what we are now calling an instrument error. The real question is, what's causing the surge of acetonitrile on the machine?? This peak is the effect, what is the cause and why is it random?

I believe that the answer may be in I.D.'s posting of the 16th. If the problem suddenly occurs after 5 years of running and affects multiple instruments, there is almost no probability that it is an instrument issue and a very high probability that it is a laboratory issue. Look for something affecting all instruments that changed around that time, including reagent suppliers, technicians, procedures. I am guessing that you are QCing a product, so don't overlook a change in the process that may be affecting the LCs.

Good point, we have looked into any change. The only one we have are the column are different than what we had used. But, how could a column cause a surge of organic to go through the instrument?
We have just theorized there is a surge since it is the only thing that makes sense.

What about sticking pump check valves or other solvent delivery components sticking? This would be infrequent, cause a surge, and would be most likely after years of instrument use, and would possibly occur on multiple systems of similar age - all situations consistent with yours.

In other words, is all your pump maitenance, etc. up to date?

It would be a lot easier helping you, if you could give a description of your chromatographic system: What columnn do you use, which mobile phase do you use with description of gradient etc.

With the May 16 answer of original anonymous it really becomes clear that Tom was right, though obviously, there must be some strange coincidences occurring. Strangest of all: Why are the missing peaks not showing on reinjection??? Do you reinject without prior modification of the collected fraction (ie, you don´t remove the AcCN?), thus washing these peaks through the column (very early elution)?
Is this due to another power supply problem???

Just by the way; did you consider some sudden changes in the refractive index of the flow? Are you working with an on-line vacuum degasser?

Thank you all for the good sugestions. The instruments are all PE of the IS200 series. We have the instruments on service contract and have regurlar PM work done. The column is a Varain C18 Omnisphere 5 mcn (gives great seperation for the cost). Flow 1.5 mL/min, init 96:4 Buffer:ACN linear 8 min to 90:10 Buffer:ACN, Stays there for 5 min and goes back down to the original concentration in 8 min.

There was some confusion on the reinjection. The mystrey peak did not show on reinjection but all the peaks did show at the expected RT.

A change in the refractive index could be possible. If there were a clog or particle going through the system periodically could it cause a change in the refractive index? I would like more information on this.

We degas solutions by vacuum and put on a helium sparge system.

Are you using separate pumps for the buffer and AN? It almost sounds like you are sporadically losing prime on the buffer pump. You would then get a single peak since AN would be your only solvent but it would be late since the flow would be very low. However, your pressure would also drop and I cannot explain why you would regain prime. Have you changed your degassing procedure?

We are using seperate containers for buffer and ACN but we are using a low preasure mixing system. So there could be some sticking with the propotioning valves. But, we can't seem to prove it. The systems appear to behave most of the time. Usual problems with the proporting valves are evident by continual leaking at the column line with the pump off.

We have not been here when the peak show, so we do not know if the preasure drops.

Are degassing procedure has been the same for over 5 years.

As I stated earlier, we belive it must be a surge do to the evidence. But what is the cause of the surge still can not be determined. The PE Tecnician could not figure it out and has sent it to their speacialist.

Thanks for the continued suggestions.

Don't know how the PE system works, so this may not be worth much. On our Shimadzu systems, the method parameters are downloaded to the system before each injection. If this is the case on your systems, could you have a sporadic software problem where the method parameters are incorrectly or incompletely downloaded to the system? Perhaps you could try a sequence where no sample is injected but the method parameters would be downloaded and check the system during this to see if there is any change in system pressure. Another thought is that the system may have problems with the low flow of AN. Could you try an AN-buffer mix for the strong solvent so you would need a greater amount from that reservoir?

The PE system we employ uses a link box that holds the same method for the entire run.

We have done what you said until we can determine what the surge is. We will have a buffer 96:4 and the other 90:10. That way if we get a surge of one it should not effect the chromatography much.

Repost from Acetonitrile thread below...

By Tom M. on Thursday, May 31, 2001 - 11:54 am:
I do not know if acetonitrile can undergo some type of free radical polymerization. However, I have been told the same thing by Agilent representatives. I know that that Agilent field service engineers here in the Bay area like you to flush the system with hot water before performing the OQ/PV.

We have had 10 HP1100s with quaternary pumps for 4 years. In those four years we have had three instances of the multi-channel gradient valve "sticking". Each time it was an acetonitrile channel sticking open, and each time the problem was fixed by flushing the channel with hot water. I don't know if it is polymerization or a little buffer leaking into the acetonitrile channel and precipating. If anyone has an applicable reference I would greatly appreciate it.

Very interesting. I would like to have further information if anyone has it.