We have got a method validated on a Waters Alliance HPLC, the validation was allright, so the method was implemented in the lab, some of our analist are working with some older, less performance HPLC's, in the original chromatograms there was a hughe peak with a lot of fronting at 5 min and the peak of our interest was a symmetric peak at 10 min.

When we use the method on an older HPLC, the first peak has got a lot of tailing, with a maximum at 8 min and back to baseline at 15 min.

We used another (older) HPLC and get the same result. At another Alliance we get a good result.

Has someone has got a good explanation for these troubles ??

Is this a gradient method?

No, its an isocratic method, the first peak is a basic compound, the second is a preservative.

What is the pH of your mobile phase, and how do you mix it?

Its just methanol-water, we have used the same mixture (premixed) at all the HPLC's with a lot of different results.

I suggest you use a buffer This will straighten out the difficulties with the basic peak.

Can you give me a reason about how it comes there are difficulties when I change from HPLC?

First rule:
The HPLC is not the only thing in the world we can deal with.

Second rule:
Never ever use a LPHPLC “less performance HPLC” with methods developed and validated on a Waters Alliance HPLC, if you are not familiar with differences between LPHPLC and HPLC.

Third rule:
If you see “a lot” of fronting at 5 min and, in a while “a lot” of tailing at 8 min… consider the firs rule please.

Presumably you have already checked, but a couple of other possibilities come to mind:

1. Have you tried swapping the columns between the two systems?
2. Is the injection volume the same on both sytems?
3. Is the sample dissolved in the mobile phase or in a different (stronger?) solvent?

The recommendation was to fix the pH. You asked for a reason why there are differences between HPLCs. In order to find this, I need to ask more questions, but they are not going to solve the problem...
What injectors are used on both systems?
What is the sample dissolved in?
Do you know what the bandspreading of either system is?
Also, the questions from the second anonymous on the columns used in both systems and the injection volume are valid. They all point to the same suspicion...

We used always the same column, the same injection volume, the sample is dissolved in the mobile fase, all the HPLC are equiped with an autosampler, I have never heard of a bandspreading, if you can specify your question, maybe I can give you an answer.

I guess I have give an answer to all the questions, is there anyone who can give me a reason about where it goes wrong, we have changed many times from HPLC (with other methods) and never got problems, this is the first time and I have really no idea.

All ideas are welcome.

I still do not yet have a specific reason what is wrong, but I can tell you the fundamental reason: you are running a basic compound without using a buffer. This is a cause of reproducibility problems. If you fix this, the problem will go away.
Other questions to the source of the difference between both instruments:
Is there a pressure difference between both instruments?
What are the temperature settings at all instruments?

We are not interested in the basic component, its all about the preservative, that's the reason why we don't use a buffer. The temperature is ambient, but that won't make the difference.
The pressures are at all the HPLC ± the same.

You have changing chromatography of the basic compound. This is sometimes interfering with the analysis of the preservative. You do not want this to happen, and you asked why this is happening.
How do you know that temperature won't make a difference?

The temperature won't make a difference because we have done the analysis at the same day, so the temperatue variaton is not more than 1 or 2 degrees.

We have done some other test, with a buffer, and we have found the following:

At the alliance, we have found a lot of tailing on our basic compound

At the other ones, we found fronting.

This method was just a quantitative method for the basic compound.

What was the buffer?
Did you dissolve the sample in the buffered mobile phase?
You reported a peak at 8 minutes that lasted to 15 minutes. Was this with the buffer?
How many grams are you injecting?

This method was just like the USP method, we dissolved all samples in the buffered mobile phase, when I reported about the peak at 8 minutes that lasted to 15 minutes, this was with the unbuffered mobile phase, but our question is how it comes there are so many differences between 2 systems (a new Alliance and an older Anonymous HPLC).

The reason I was asking is that one area where instruments can vary significantly is in the volume and mixing characteristics the "transfer" lines between the injection valve and the column. I have seen at least one instance where a method was developed and validated on an "inferior" instrument (relatively large volume) and then failed on a "superior" instrument (much less mixing volume) due to peak shape problems. In that case, the sample was dissolved in a stronger solvent than the mobile phase. On the larger-volume instrument, there was enough mixing between injector and column to "dilute" that difference.

What you described for that 8-15 minute peak sounds suspiciously like a "system peak" (basically and endogenous peak caused by upsetting the equilibrium of the system as a consequence of injection). That's why I asked about buffering, loading, etc.

Two things to try on the Alliance system:

1. Dissolve your sample in the mobile phase (not "something like the mobile phase", but the actual mobile phase) and see if the problem gets better.

2. If it does, then either change the method so as to dissolve sample in mobile phase, or else add a coil of capillary tubing just upstream of the column (maybe 50cm of 0.010"id tubing, tightly coiled) to give you some additional mixing (in effect, making the new Alliance look more like the old Anonymous system).

I guess there is a mistake, the Alliance has got a good separation between the basic compound and the preservative, the old systems give a peak that begin at 8 minutes and back to baseline at 15 minutes, excuse me for that mistake.

There was a lot discussions in a week on this topic and then activities just ceased. Hope someone will come back and read this again.

There were a lot of Q&A exchanges, and the initiator of this posting patiently provided many details. Most questions focused on the method itself (m.p., column, temp., inj. vol., etc). Those are typically defined in validated methods. However, when transferring method between different generations of HPLC systems (i.e., LPHPLC vs. HPLC), what is often overlooked is the effects of hardware on the chromatographic outcome.

In this case, the Alliance gave one fronting peak at 5 min and a symmetrically peak at 10 min (peak of interest). The older HPLC system presented a huge tailing peak starting at 8 min and ending at 15 min. No mentioning if the peak of interest was seen on the older system. If this peak was seen but the RT was longer than 10 min, or this peak was buried under the big tail and could not be detected at all, then the problem may be the injector and the connections from the injector to the column.

The ideal injector will introduce the sample as a plug onto the column with minimum dilution in the process (minimum band spreading). Some older autoinjectors are not very good in doing so. When the injector turns from load to inject position, the m.p. flows through the sample loop and push the sample plug towards the column. Depending on the design of the injection valve, different degree of disturbance to the sample plug occurs and the sample is diluted in the process. The connection between injector and the column is also important. Most likely that the total volume between the injector and the column is larger on the older system, which results in more dilution of the sample before it hits the column. Not all the sample will reach the head of the column at the same time, this results in broader peaks. It will also take longer for it to reach the column (shifts in retention time). The fittings used in the connections are also critical, especially with stainless steel fittings. If the fitting did not sit properly on the head of the column (it may not produce a leak), further disturbance to the sample band occurred. There is no mechanism to control peak tailing in this method (no buffer for a basic compound), which just made the problem more severe. The connections between the column and the detector may also contribute to the observed peak shape but is less critical.

When validating a method, it’s a good idea to try it on very different HPLC systems, preferably a high performance one and a low performance one, to minimize troubles in method transfer do the road.

To expand on what Apr 12 1:23 anon says, I have had to add additional tubing to the Alliance system to make a method work that was developed on an Agilent 1100. Perhaps the Alliance still has less mixing than even older systems, but in my experience it already has far more post-injector/pre-column mixing than the Agilent. Granted it was a poorly developed method w/ 100% ACN as diluent, but it clearly showed the superiority of the Agilent in this area. (Or is the Alliance "better" since it didn't care about the diluent strength?!)

So you needed to add tubing to the Alliance system to make it look like the Agilent 1100.
I do not get how this makes you conclude that the Agilent has less (!!!!) mixing than the Alliance. I think that you just proved the point that the Alliance is the system with the lowest amount of dead space in the machine. Thank you!

Sorry- Total mistake. The tubing was added to the 1100 to make it match the Alliance. Without the approximateky 5' of 0.01" tubing added before the column, the effect of the ACN diluent was clearly ruining the separation on the 1100 whereas the Alliance did not show any of this effect. I haven't worked much on Alliances, but maybe there is some other explaination for this behavior.

Agilent or Alliance isn't the discussion, the point is the poor separation and the differences in separation and how to solve this problem.

Discussing obverved differences in HPLC systems and the steps taken to solve the problems seems like a fine way to help someone who is observing differences between HPLC systems and wants to take steps to solve their problem.