We have a gradient method used to detect impurities in a drug substance. We have found that some unknown peaks and a large gradient shift occasionally occur towards the end of the 50 minute run. Could these extra peaks be from the water we are using? We currently use a Millipore UP system.

The unknown peaks might be from the weak solvent of your gradient. The column will extract impurities from the weak solvent until the solvent strength is high enough to elute them. Try varying the weak-solvent hold time before no-injection blank gradients. If the peaks originate in the weak solvent, the resulting peak heights will be proportional to the weak-solvent hold time.

Gradient baseline shifts can be the result of a zillion factors. What are the run parameters? Solvent, gradient shape, and wavelenght especially?

Steve:
Thanks for your reply and advice. I will carry out the experiment you mentioned.
Some added info: We are detecting at 240nm and use 3 weakly acidified ACN:WATER mobile phases, each at different concentrations(50:50, 65:35 and 85:15.) The gradient is linear with a 1ml/min flow rate. Hope this helps.

Matthew,
A few thoughts:
Your solvent range is 50-85% ACN, yet the extra peaks are occurring late in the run. If these are impurities, they're pretty hydrophobic. And ACN and 240nm shouldn't produce much of a shift at all.
I've seen a similar pattern of gradient artifacts - late peaks and an unsteady baseline - which were traced to a persistent contamination of the pumping system. (In our case it appeared to originate with microbial growth in a phosphate buffer that inoculated the system.) It was eliminated with routine - and harsh - cleaning of the entire solvent delivery system. All lines were washed with water and then put into a 1:20 dilution of hypochlorite laundry bleach solution. 50-100 ml of this were pumped directly to waste. Diluted bleach was followed by water and then the system was returned to solvent.
Systems treated this way have been used for very reliable trace-level gradient determinations - at low wavelength - of impurities in a synthetic drug substance.

Steve
NaPro BioTherapeutics

Steve:
Microbial growth is definetely a possible cause of what we are seeing.(Will a common household bleach do the trick?)
My only concern is that we encounter the baseline shift on more than one system, 2 of which are relatively new. Hence the pumps should not have been contaminated so quickly.
Could it be that the columns are contaminated with a stuborn microbial growth too?
I take it that we flush the LC system, excluding the column, with bleach.

Thanks again
Matthew

Matthew, In a past life, we found signs of persistent bug growth in 6 LCs on a single bench. We traced it to repeated inoculation of the systems from a common helium distribution system. The home-built sparge piping was originally installed without back-flow preventing check valves in the reservoirs, so a sudden pressure vent - as from opening a pressurized reservoir - would cause blow-back from vessels into the tubing. The helium then spread the contamination.

We installed vented unions in the helium lines inside the reservoirs and above the liquid level and check valves in all lines.

We used common household bleach diluted 1:10 or 1:20. It's a good sanitizer as well as scouring oxidant of the system. Don't pump through the detector, since the high pH could attack the cell lenses. Flush well with water.


Steve Bannister
NaPro BioTherapeutics

I have experienced this kind of problem with Gradient chromatography at High% of organic.
Are using Degasser? What type?

If you are using HP it is OK, but if you are utilizing Phhenomenex, Metachem deggqasser ,leaching impurities from the memberane is source of the unknown peak.


Aryo A. Nikopour
PPD Development

We have seen the same problem that mathew has seen...Gradient method for impurities in a drug substance using UV at 220 and HP system. We have noticed some unknown peak eluting lateeee at approx 240min and is reproducible. The organic conc. increases to approx 60% towards the end of the run. Is it possible that the extra peaks are from the degasser membrane!! Gradient uses, phosphate buffer(10mM), ACN and MeOH.

An Easy way to find the contamination is to run your aqueous eluent at 100 % for a while and then run through the gradient. Any contaminants in the Aqueous portion will become concentrated on the column. If the contaminant peak is much larger than normal then the contamination is most likely from the aqueous eluent. Contamination from the degasser membrane itself is highly unlikely unless the pH is ver very low. If it were it would be more likely microbial growth inside the degasser. I have heard some chromatographers using sodium azide to control microbial growth with no effect on detection or elution. I dont recall a concentration but that coupled with the bleach cleaning above would do the trick

We've used sodium azide to control microbial growth at about 0.1%. This works well, and has no effect on detection of any of our drug substances.

Also, are you filtering your ACN? Sometimes there is a polymerization in even the highest quality solvent. This may be a source of contamination on your column, but usually I see this as a clogging of the inlet frit.

If your problem does not appear to be microbial growth, but a buildup of chemical impurities, cleaning your system (without column) with suitable organic solvent (isopropanol, perhaps) and dilute nitric acid frequently works. Check the appropriate cleaning solutions in your manual or with you service engineer from your instrument supplier.

I am experiencing the same problem as Matthew's - gradient elution with phosphate buffer vs ACN, 210nm. The ghost peak is reproducible (elutes late and corresponds to ca. 95% ACN), even when we physically bi-pass the injector. I took apart the pump head, cleaned the parts and changed the frits. That helped to cut down the ghost peak to one-fifth of its original size....but can't improve any further. I've tried the trick to increase the hold time at high %aqueous, the ghost peak is bigger, but not in a linear fashion. Any suggestion?

How strong is the Phos buffer? If it is say, 25 mM, what happens if you drop back to 5 mM? Do the artifact peaks reduce in size by 5 fold?

-brian paasch
Genentech, Inc.

We have experienced reproducible ghost peaks in gradient elution as well and have solved it by replacing the inlet filter on the aqueous solvent with a C18 cartridge. The cartridge traps the non-polar contaminents that usually show up when the non-polar content of the mobile phase is very high.

Just a thought...

The original message (above) mentioned that the mobile phase solvents were weakly acidified...any chance that there was a TFA modification going on? I seem to recall ghost peaks and weird oddities on my chromatogram which stemmed from a problem with the TFA. (Sorry if this isn't that helpful...this was about 7 years ago, and my memory of the problem is a little rusty.)

In response to Peters idea of using a C18 cartridge instead of the inlet filter. Would using this affect your pumping system or the RT's of your compounds?

The C18 cartridge on the reservoir pick-up line will only be a problem if the pressure drop accross it is high - due either to flow resistance or high flow rates. The resulting outgassing ahead of the pump could cause flowrate - and RT - instability. Using a big particle and/or short cartridge bed is a very good idea.

In our gradient system using water : Ethyl acetate : Acetonitrile on a C18 column, after a few runs, ghost peaks start appearing at the middle of the run. How to solve this.