Hello everybody,
recently, I have problems with durability of my HPLC columns. Earlier, it lasted at least 3 months, sometimes up to six (that good at our department, often the columns used by others lasted much shorter time, but nobody here has any realy HPLC training). However, several last columns deteriorated very quickly and I have no idea whether I'm doing something wrong or it can be just coincindence of combination of old columns (long time unused since purchased) with old used column that just happened to collapse right now.
The problem also is, that I'm trying to detect new compounds, so I'm changing conditions and thus I'm not really able to recognize, whether the change in peak shape is due to wrong pH (close to some imaginery pKa) or because the column was fouled.
I'll try to list as many things as possible and I'd like if you could tell me, whether in your opinions any of these things can lead to such deteriorations.
I standardly use ZORBAX Eclipse PLus C18 column from Agilent. In the past I used pH 4.0, but because of some dificulties I switched to phosphate pH 7.0. During my playing, I was changeing the pH in the range about 5 - 7.5, but I don't think I would ever go higher. I operate at 0.4 ml/min and 40°C, which should be well below the limits for the column. After a batch is finished, the column is cooled down to 25°C and flow rate adjusted to 0.05 ml/min with 50%B (we were told this is better for the pumps then shutting it down). When the machine is to be turned off, I put the column into H2O/MeOH and wash usually at 0.2 - 0.3 ml/min with 5% MeOH first and then with 95% MeOH (I make 5% increments).
The mobile phase is filtered. We have in-line filter, I have changed it several times recently, but it always looked rather clean (in comparison to previous occasions). I have two types of samples - enzymatic reactions and plant extracts. The enzymatic reactions are usually stopped with double volume of methanol, centrifuged and filtered. However, because I was measuring some low concentrations recently and needed low concentration of methanol, I stopped them with heating, centrifuged and filtered. This could lead to higher load of salts, on the column, but on the other hand, the starting concentration of methanol was very low, so it should wash out, shouldn't it? The other samples are extracts from plants into organic solvents, thus they contain plenty of compounds, but those (hopefully) elute.
The first column where I noticed the quick deterioration I managed to partially dry out. I ran out of mobile phase A, but B was still there, so it was like 50% methanol, 50% air. But as I read on the Internet that shouldn't be such big concern anymore for the modern columns.
The other thing is, that I had to decrease the methanol concentration as much as I ended up with 0% in the beginning of the gradient. However, can this lead to peak spliting?
The second column was core-shell Kinetex 2.6u column we got long time ago (maybe year or two) to try it. This one lasted also only few weeks.To my surprise the main problem (at least in the beginning) was with the low flow rate, where the pressure quickly rose (between days, but also when I was washing the column with H2O/MeOH). However, when I started my analysis with the high flowrate, the pressure was about the same as the first time. I tried backflushing this column as it is possible, but that didn't help.
And last week I took ZORBAX column that was used previously many times. I checked and it worked fine. First I loaded some extracts and it worked OK (although it's hard to say with such low and so many peaks, but on fluorescence detector it seemed OK). Yesterday I run some activity measurements. The very first sample had (almost) split peaks (on attached picture). But later it actually improved, the peaks were narrower (although still wide in comparison with normal runs 0.15 vs 0.07) and tailing was 1.1, but that's rather because it's front tailing as you can see on the second picture. So it seems rather like in the beginning there was more of the first variant, while it slowly shifted to the second variant, but not enough to split the peaks. However, even when there was less of the first compound, it did not split into two, but rather remained one wide peak (see the third picture). Nevertheless, in any case the very first peak seems to be split. Also, while the backpressure with the low florrate on this column was around 30 MPa in the previous days, now it's only around 15-20MPa, if that may be of any help.
The problem is, that most of the things I have done also previously, when the columns lasted for longer, so the main changes are the air in the column (only in the first though), old unused column (only the second though); low methanol concentration in the gradient and possibly high salt in the sample (both only in the first and second columns though).
Hi
You did not exactly specify your buffer system, maybe you find additional infos in the
Product Data Sheet, here is the link:
https://www.agilent.com/cs/library/specifications/Public/820114-002.pdf
There are a few tips for usage, think they can be helpful:
....
Eclipse Plus C18 can be used with basic, neutral or acidic compounds. Ionizable compounds (basic, acidic) generally are best separated at about pH 3 with this column. How-ever, Eclipse Plus C18 is especially suited for separating basic compounds when an intermediate pH (4–8) must be used to maintain compound stability or to obtain desired band spacing (selectivity). For optimum results and long-term reproducibility, the use of 10–50 mM buffers is always recommended when separating ionizable compounds.
.....
NOTE: Eclipse Plus columns are designed for high sta-bility over a wide pH range. However, all silica-based packings have some solubility in pH >6 aqueous mobile phases. Therefore, when using silica-based columns under conditions of pH >6, maximum column lifetime is obtained by operation at low temperatures (6 is also enhanced by avoiding phosphate and carbonate buffers [ref.: H. A. Claessens, M. A. van Straten and J. J. Kirkland, J. Chromatogr. (A), 728 (1996) 259].
Columns should not be maintained at a neutral or elevated pH, or at elevated temperature, when not in use
....
Think the pH is a key for your stabilty, RP-Columns work best in low pH-ranges...we use 10mM TFA for many applications and have a good stability.
Call some represantatives from local analytical suppliers RP-Chromatography is a wide field and there are a lot of alternative systems, sometimes suppliers can propose you methods or better analytical systems.
Good luck
Hi Tomas
usually, RP columns should have a good chemical stability. I cannot see any harsh conditions in your methods, maybe the pH is a problem..looks like a little like an insolubility, maybe your fowling is supported by conditions like buffer ion and pH.
We have TFA as standard aqueous RP-Phase, it guarantees a good solubility for a wide range of molecules and you can even evaporate it if you need to change to mass selsective systems.
If your columns have a mechanical problem you can change the column "principal" from adsorbent "particles" to sintered Columns like the Chromolith system of Merck, they withstand very high pump and flow rates even if you perform highspeed sampling with 5ml/min plus, they keep there structure, while usual adsorpents get compressed in the direction of flow and you loose permeability and have increasing backpressure.
Do you use pre-columns? Often, they help prevent damages in the more expensive main column area...avoid to run your columns dry, usually the pumps are more sensitive than the coumns, but it is not an optimum if you want to perform standard analytics. if you dont use an in-line degaser, degas your solvents before usage.
good luck
Thanks for the answer, I appreciate it.
Of course I'm trying not to let air in, I had to calculate wrongly or not check the amount of mobile phase. We have in-line degaser.
We do not have pre-column, only the in-line filter. Do you think it makes big difference?
I was also trying formic acid, but that has pH close to 2, that is the limit of the column. And since I want to separate some phosphates, I'd have to go below like 1.2 to be away from the pKa.
Hi Tomas,
what are actually your solvents A and B? Maybe I missed it in your text, although I read through several times. Did you checked wether those two solvents are miscible in each other/no precipitate occuring when increasing % of B? Whe had the problem once that when using PBS as A and some water/organic solvent as B, the salt precipitates. Even if this happens in a low percentage you will notice from the pressure. Maybe it unmixes within the column (because heated to 40°C). Did you tried to mix that solvents at 40°C? Did you try to run the analysis at room temperature (e.g. column at 21°). How does this affect the separation?
Pre-colums with replacable filters are a must for frequent HPLC usage in my opinion. I can tell you which ones we use frequently if you like.
pH below 1.2 will destroy most columns for sure! Even if the supplier tells that pH 2 is fine, there is still the posibility to destroy the column with time. There are some special columns out there for that case, I can tell you which I know if necessary.
You should not measure the column lifein months but in the amount of samples you run with it. That would be my first point.
Second pH 7,5 is not the optimal to the column though when you mix the sample with the eluent it should go to an optimal one, but you might have a lot of precipitation, and this will damage the guard-column and the column itself. I would advise you to grab one sample and mix it with eluent A and see if there is precipitation.
I have the same problem, with the same columns but I know I am using non optimal conditions and my samples are quite complex so I am expecting it.
Also you can and should contact Agilent about this as well as your HPLC brand for some insight.
Oh also, pH changes peak shape!
Marcus Binner usually solvent A is water base and B Methanol.
Hi Vera,
good points you mentioned. To the mobile phase: in our group, we use water/acetonitrile mixture with 0.1 % formic acid as solvent A and B... I'm not very experienced with that but I guess phosphate buffer and methanol are not suitable if MeOH content is too high during the gardient program.
Actually MeOH helps cleaning the column, in the case of ZORBAX Eclipse PLus C18 as it is described in the column manual the gradient goes fomr 2% inicial concentration to 100% and then back.
In my case I use a buffer without phosphate salts (they precipitate easily), formic acid is used (to my understanding) to remove lipids which I really hope my samples don't have (they shouldn't at least).
In my case the columns are fine and durable, just now using Milk and yogurt samples I was expecting this to happen (also I went testing different extraction techniques and so on).
ensure that you are filtering your samples properly. often samples can precipitate in the vials... injecting such samples will quickly clog your column, especially kinetex columns deteriorate easily because of the small particle sizes.
Hi everybody,
thanks for your answers.
The mobile phases are:
A used to be ammonium formate pH 4.0 (shifted to 4.5), now I use 20 mM ammonium phosphate pH 6-7 (mostly 7.0 though). I tried once 100mM phosphate, but that was only once, so again, it should affect only one column.
B 100% methanol
The HPLC is Nexera from Shimadzu.
I'm trying to always flush the injection loop by injecting 50ul of H2O/MeOH and also there is some washing system with 50% methanol, so that should leave the bubbles away.
I know that the number of injections is of course more important, but I cannot really imagine how to measure that. And what I meant with the previous columns was, that in the spring I had a column on which I was measuring regularly (although they were probably mostly/only the enzymatic reactions) and that column lasted for several months.
Your pH range should be fine, it stands pH 2-9.
I have a log book where I write down, my name, date, the amount of samples I am running, the initial pressure and leave a observation column to put if the pressure rise a lot or when I do maintenance on it I always write down what I did and when and the result of that (like pressure went back to normal after washing or guard-column changed, etc).
I would contact either Agilent or Shimadzu to try to figure why is that happening...
Hi
You did not exactly specify your buffer system, maybe you find additional infos in the
Product Data Sheet, here is the link:
https://www.agilent.com/cs/library/specifications/Public/820114-002.pdf
There are a few tips for usage, think they can be helpful:
....
Eclipse Plus C18 can be used with basic, neutral or acidic compounds. Ionizable compounds (basic, acidic) generally are best separated at about pH 3 with this column. How-ever, Eclipse Plus C18 is especially suited for separating basic compounds when an intermediate pH (4–8) must be used to maintain compound stability or to obtain desired band spacing (selectivity). For optimum results and long-term reproducibility, the use of 10–50 mM buffers is always recommended when separating ionizable compounds.
.....
NOTE: Eclipse Plus columns are designed for high sta-bility over a wide pH range. However, all silica-based packings have some solubility in pH >6 aqueous mobile phases. Therefore, when using silica-based columns under conditions of pH >6, maximum column lifetime is obtained by operation at low temperatures (6 is also enhanced by avoiding phosphate and carbonate buffers [ref.: H. A. Claessens, M. A. van Straten and J. J. Kirkland, J. Chromatogr. (A), 728 (1996) 259].
Columns should not be maintained at a neutral or elevated pH, or at elevated temperature, when not in use
....
Think the pH is a key for your stabilty, RP-Columns work best in low pH-ranges...we use 10mM TFA for many applications and have a good stability.
Call some represantatives from local analytical suppliers RP-Chromatography is a wide field and there are a lot of alternative systems, sometimes suppliers can propose you methods or better analytical systems.
Good luck
As Frank mentioned, you should try to go lower with the temperature as well, this seem to affect the speed of degradation of the column material. Phenomenex recently introduced their "evo" columns with ethane cross-linking between the silicium atoms which provides pH stability between 1-12 (as they describe, I personally would not go lower then 2 and not higher then 10). Maybe there is something similar avaliable from Agilent, fitting better for your application. This should not be an advertisement, just a thing which came in my mind since we used this columns recently.
I would like to add 3-4 points.
1. Whether temperature is essential for the HPLC analysis ? If yes, combination of temperature and pH may be effecting your column life.
2. Whether guard column is used in your system ? Addition of guard column will improve the durability of your column. More over it is always preferred to have solvent with high purity which may be purchased from reputed farm like Sigma, Merck and others. Manual filtration is not well appreciated.
3. Column regeneration could be done as follows ..
Following is the column flushing procedure for reversed phase column.
Regeneration
· To regenerate a reversed phase column:
1. First invert the column
2. Rinse the column at ± 40% of the optimum flow rate, for about 45 - 60 min with solvent in the following order:
· Flush the column with 20 column volumes Water
· Flush the column with 20 column volumes Methanol
· Flush the column with 20 column volumes Acetonitrile
· Flush the column with 5 column volumes Isopropanol
· Flush the column with 20 column volumes Dichloromethanol
· Flush the column with 5 column volumes Isopropanol
· Flush the column with 20 column volumes Acetonitrile
3. Invert the column to the original position and equilibrate with the analytical eluent.
Note: When using another rinsing method: always start with water to remove buffers and assure that consecutive eluents are mixable.
HOPEFULLY THE WHOLE THING WILL HELP YOU TO OVERCOME THE PROBLEM.
Good Luck
Hi Tomáš: Besides the notation about pH, I do not see an explanation of what your mobile phase is in your post. Are you running 100% aqueous buffer? Isocratic or gradient? What concentration(s)? You mention a wash with methanol and water too. We need your actual HPLC method conditions.
When asking for help with chromatography questions, please try and provide the basic details. These include injection volume, flow rate, column type, column name & model and most importantly, the column's dimensions (L x ID x particle size). Without this basic info, we can not help. Additionally, we need details regarding your actual mobile phase solutions used. This should include all additives, solutions, concentrations, grade of materials used, molarity or % and composition time changes for gradient analysis or mixture ratio for isocratic analysis. Additionally, we need detector type (wavelength and bandwidth), sample concentration and injection solvent used.
*Based on the info so far, I would ask you if you are regularly washing the column with a stronger solvent than water/buffer after each run? Are your samples soluble in the mobile phase? Are you filtering the samples through a 0.45 micron filter first? For RP applications, you must use a stronger solvent (e.g. methanol and/or acetonitrile) at some point during or after each analysis to wash off any late eluters or accumulated material. If you are running isocratic, then you must use a gradient after to wash the column. Failure to develop your method to initially retain, then elute ALL of the material injected will result in a fouled column and poor scientific technique/method (data not valid). Everything must be soluble at all times. Do not worry about guard columns at this point. I think your primary problems are related to a lack of basic training and your column is simply being fouled through the use of the wrong HPLC methods. Develop better methods and column washes and many of the problems you are experiencing will go away.
In the meantime, please seek advice from an experienced and local chromatographer and also read some of the books on chromatography so you can learn the fundamentals of method development (critical to using an HPLC system). Take training classes (professional paid ones) if possible too. Always change one thing at a time, observe and discuss your findings with someone local to get help with this.
In addition to all Nice advices, mentioned above, I am using Zorbax Eclipse C18, Zorbax C18, C8 Extend columns since about 7 years ago. ALL those columns still working very well, using pH 2 - 10, at 25 to 55 C. This may be due to low flow rate (0.1 to 0.6 mL/min). In case of forked peaks, install the column in revert position, heat at 50 C, inject 100 uL (x 5 times) of DEMSO:CHCl3 (1:1), acetone. Then Bring back to original flow direction, leave the column over 100% acetonitrile.
Be sure, that you have proper fittings.
Hi
I will discuss about the samples… the first sample is the result of an enzymatic reaction and I think, you are using low methanol concentration to see the analytes at low concentrations and low wavelengths (near of the cut off of the methanol). If you are analyzing at wavelengths lower than 240, you will have detection problems, particularly if they do not absorb light strongly… if that is the problem, you should use another detector (or derivatization process)…
For the extract of plants, if you extract with organic solvents, you have some problems: you extract very non polar compounds that cannot elute in the C18 at low concentrations of methanol or acetonitrile; you extract acidic and alkaline compounds and the change in the pH of the mobile phase is not effective for all of them. You should divide the extract on acidic, neutral or alkaline compounds, so, you can use the adequate buffer to separate the compounds. With the fluorescence detector you will see only the fluorescent compounds and they usually are neutral (fusioned rings) and they are not affected by the pH.
good luck..
My mobile phases:
A: 10 mM H3PO4 (HPLC grade), 15 mM triethylamine (HPLC grade) in MilliQ water, filtered through 0.22 um filter
B: 100% methanol (gradient grade)
my run
for 2.1x50mm Kinetex column 2,7um (core-shell) - 0,4ml/min, 40°C
(time/%B) 0/0; 1/0; 2/1; 2,5/5; 3/60; 4/60; 4,5/0; 7/end
for 2x150mm Polaris column 3um - 0,2ml/min, 35°C
0/5; 3/5; 7/13; 8/18; 9/60; 11/60; 12/60; 17/end
my samples contain 100mM phosphate, 50 mM citric acid, 20 mM MgCl2, some proteins
after reaction, they are heated, so some of the ions precipitate, centrifuged 10 min at 20 000g. Diluted 10-20-times with A, filtered through 0.22 filters.
Injection 10-20 ul (sometimes up to 50ul)
The injection volume seems quite large for such small columns and low flow rates. Do you need to inject that much to get decent signal-to-noise?
Hi Tomáš,
are you sure that for the mixtures of your mobile phase A with 60% B no precipitation occurs? Matthew is right, the columns are rather small, but the particle size is rather small too, this may cause that high pressures. E.g. we go for 5µm for the analytical (~80-90 bar @ 0.5 ml/min) and 10µm for the prep. HPLC (~ 30 bar @ 20 ml/min), respectively.
@Matthew what injection volume would you recommend? I'm trying to keep it at 10ul, but as I'm currently varying the concentration, I must vary the volume.
Do you think it would be better to dilute less (and keep higher salt concentration) with smaller injections than this?
@Marcus well, I haven't tried it, but phosphate should be soluble and since triethylamine is organic, I'd assume it's soluble in methanol as well.
So I tried to mix A and B in 1:2 and 2:1 ratio and centrifuged at 20 000g for 10 minutes and I haven't seen any pelet.
too bad I cannot actually be there to see, sounds like a sticky wicket.
volume- as low as is reasonable. Looking again at the chromatograms you posted, the signal-to-noise looks reasonable. The phosphate buffers will precipitate if you exceed 60% organic, but I would not think the MgCl2 would like that much organic. Could that be what is fouling the columns? Marcus is right- test the solvent mix and see if you can make it precipitate. Could it precipitate during mixing in the instrument? Have you tried reprogramming the gradient so that B line contains some buffer mixed in with the organic (and then prefiltered) rather than running the B line at 100% organic?
Never assume- there are multiple proverbs for that.
"Theory guides, but experience decides." is a tame version.
Can the fouled columns be brought back to life with flush instructions as above but with 60 degree C DI water and the column heater set to 60 for the water step? -trying to redissolve the inorganics and get them to pass on through.
I thought the magnesium is washed away in the beginning.
I saw a table in some book showing maximum concentration of phosphates (sodium, potassium, ammonium) soluble in organics. I assumed it's in 100%. And there was something like 50 or so mM ammonium phosphate. So shall I go maximally to 30%? And then wash after the analyses with 95% H2O and then with 95% MeOH?
With the backflushing, I must wash it with water/methanol and keep the flowrate low, right? I tried the backflushing recently, but only with the water and methanol and low temperature (I think I didn't go above 35°C). So I can try that next time. I can try the acetonitrile and isopropanol. I don't think we have anything stronger here.
I'm just wondering, whether it could create such peak shapes as I'm observing (see the attached picture).
The problem is, that I've been using this reaction stup for long time. Although I used to stop the reaction with double volume of methanol, while now I switched to heating. So it could be, that it used to precipitate, while now it stayes in solution and precipitates on the column.
Dear Tomas
I am sorry you are experiencing such hassle. I am surmising this problem is not what your research is about, but you want to overcome the column fouling so that you can pursue your research. When encountering such problems, it is very tempting for me to change more than one variable at a time (i.e. method of stopping a rxn, method of column flush, etc.) However the most difficult problems might only be solved by a classical reductionist approach, by which you try to recreate a point at which it worked (you got good chromatography) and then only change one variable at a time for each experiment, giving yourself time to evaluate, ponder, discuss with colleagues after each experiment.
I hope you have a very Merry Christmas!
I tried the regeneration with the last column as follows
column ZORBAX Polaris 3um 2.0x150mm in reverse direction
flowrate 0.05 ml/min; 60°C
first I washed with 95% water, 5% methanol for ~1hr
then stepwise switched to 100% methanol and washed for ~1hr, left overnight (without pumping and heating)
next day started with 100% methanol, then swithed stepwise to 100% acetonitrile for ~1hr, then 100% isopropanol for few hours (I had some unscheduled appoinment), switched back to acetonitrile and to methanol; kept in methanol overnight (pumping, not sure about the temperature now, but probably 60°C).
Today I switched stepwise to 5% methanol, 95% water and left for some more time. Then I switched the column to normal direction, washed some more time with 5% methanol/95% water, then changed to buffer and ran some standards and you can see the result in the attached picture.
There is obviously some improvement. The three peeks shifted back to front, although not as much as they were on the new column (the shifted back during the fouling) and because of the front-tailing the tailing factor is between 0.8 and 0.95 (before it didn't even calculate the tailing factor). For the later peak (there are actually two with similar effects), again shifted slightly to front (although less, as there is the gradient), but mainly tailing factor before 2.45 and 2.5, now 1.07 and 1.33 and mostly nothing like what I posted last time! So there is definitely improvement!
I've run one real sample in the meantime and all the tailing factors are in the range 1.1 to 1.25 (even the first peaks, that do not front-tail(?)).
Although that kinda destroys my current hypothesis :( I talked to one our HPLC guru and he told me, that the mobile phase can change pH upon mixing with organics. So I tried that and it did indeed rise to 7.8 with 50% methanol.
So my next plan is, as I got only one more column to destroy, before my project will be terminated, that I will stop the reactions with methanol again (I changed it mainly because of NADH that was oxidised and because I started the run with 2% methanol and low conc. of substrate, so I couldn't dilute it as much) and I will prepare B as you suggested as A mixed with 50% B and thus I can control the pH to something like 6.0 and prevent silica hydrolysis.
However, I have one more question, I got after the discussion with you. I have always run some blank runs in between samples, with injection of 100% methanol, usually 10ul, but occasionaly up to 50ul, kind of washing of the loop. Now that I've been doing the gradients starting at less than 5%, I noticed there is always pressure rise with the methanol injection and thus I started using 100% water instead.
Do you think, that injecting of 50ul of either water or methanol could damage the column in long-term? (but again, at least with the methanol I've been doing it since ever). Shuld I switch to at least 5% methanol, or rather decrease the volume?
Thanks to everybody for your help!
- Badges
- Science method