How to get rid of severe matrix effect?

[Uwe Neue]

Well this requires a fair amount of additional homework. SInce you are using a cation exchange material, I assume that your analytes are cations. With the protocol that you have, you are in principle collecting cations, while anions and neutral compounds are washed away.
So you need to make the protocol more specific for your analytes, which hopefully are retained by both hydrophobic interaction and ion exchange. You load your analyte standards, on some 10 cartridges under acidic conditions. Next you wash with a basic solution (ammonia) containing increasing amounts of methanol: 10%, 20%, 30%... and measure the recovery of the analytes. You should get an S-shaped profile. Now you know that you can make additional washes at alkaline pH until the point of the methanol concentration just before the breakthrough of your analytes, and you can elute at alkaline pH with a methanol concentration just above the concentration where all analytes elute.

Now you improve your protocol as follows:
Load at acidic pH.
Wash under acidic conditions with methanol.
Reduce the methanol concentration to just below the breakthrough point.
Elute polar basic impurities now with this methnaol concentration at alkaline pH.
Elute your analytes at alkaline pH with a methanol concentration that is just beyond the point where you achieved 100% recovery.
Leave the rest of the basic interferences on the cartridge.

More work, but the results are likely to be better.

Also, we may need to look at improving your LC protocol, if this does not give you much improvements. However, I am reasonably confident that you will get improved results. The protocol has been proven to work on Oasis cartridges.

[Ruth]

I will try that. Small question: is the amount of ammonia important? I would just add like 1% to water and then add the different percentages methanol, or should the pH exactly be established?

[Uwe Neue]

You need to have sufficient ammonia in all solutions that the eluent will be strongly basic after the SPE cartridge. This depends both on the ammonia concentration and the volume of eluent. If you are OK at 90% methanol, all the other solutions will be OK as well.

[Ruth]

Hello,

I performed the experiments as you suggested.

There was elution of my compounds when using the MeOH-water-NH4OH mixture with 75% MeOH, not at 70%. Therefore, I selected the MeOH-water-NH4OH mixture with 70% MeOH as additional washstep to remove basic interferences.

As final elution solvent you suggested to use the MeOH-water-NH4OH mixture where the recovery was 100%. This was not achieved! Even when using 98% MeOh 2% NH4OH, the recovery was significantly lower than when using CH2Cl2-isopropanol-NH4OH as elution solvent (I guess because of the more apolar character of the latter...?).

So I performed an SPE analysis using the protocol with the additional wash at alkaline pH and elution with CH2Cl2-isopropanol-NH4OH.
There was a severe improvement! There was 40% less matrix effect for the compound with the ion suppression... However, there is still some 60-70% matrix effect (where 100% is the perfect situation). 80% is considered as ok, so I still would like to improve the protocol... All suggestions are welcome!

Thanks for your help, this has really been of great value for me!

[Uwe Neue]

Either the interferences have the exact same properties as your analyte, or I am missing something...

Try one more thing: prewash the cartridge with the solvent mixture used for elution, then with methanol and water as before. This idea is based on the possibility that your elution solvent is washing interferences off the cartridge itself. A prewash with the elution solvent will reduce or eliminate this problem.

If the interferences have the same properties as your analytes, I am running out of ideas...

[Ruth]

When using spiked water on the SPE instead of urine, there is no matrix effect, so I suppose the interferences can't come off the cartridge itself...

Could I try to change porperties of the elution solvent CH2Cl2-isopropanol-NH4OH by increasing the portion of isopropanol (now 20%), and thus making the elution solvent more polar...?

[Uwe Neue]

Good point! This eliminates problems from the cartridge.

I do not know in which direction to go. You can finetune the composition of the elution solvent. Options that we have played with are for example the substitution of ammonia with primary, secondary and tertiary amines. This provides some selectivity, but without knowledge of your interferences, this can make it better or worse. It is worth a try.

[Camisotro]

Ruth, do you know what is the delay volume of your gradient pump? This could be a big issue especially at flow rates such as 250-500 µL/minute. When your first compound elutes, has the gradient actually hit the column yet? Tell us what the typical gradient profile looks like for one of your runs and what time the first peak elutes.

Also, are you using any sort of diverter to keep the matrix peak from going through your ESI probe? If your matrix crud is going through the probe, that may leave contamination behind that is still washing out by the time your first peak elutes. If you are able to divert it away, and then switch back in time to see your first peak, you may get better results and more stability overall.

The ideal would be to set up a method where the matrix washes off first into waste, the gradient hits your column shortly thereafter, and kicks your first peak into the MS. But depending on delay volume in the pump, you may have to time it in a counterintuitive way.

In my case (as documented on these forums) my lab's pump has a rather large delay volume, such that if I initiate a "step gradient" at t=0, there is no sign of the new solvent until at least 1.2 mL has left the pump; and even then, it does not fully equilibrate at the new solvent composition until after a full 3.3 mL. Thus I decided to instead make a "do-it-myself" step gradient using a switching valve and two isocratic pumps. Of course the ideal would be to just have a proper LC pump but the funding for a proper LC stack will probably arrive some time after my graduation.

[Ruth]

The dwell volume of our LC system is around 250 µl. When I look at the pressure, there seem to be minimal delay (I roughly guess 1') between settings of the pump and the effective arrival of solvent in the column.

The gradient is set up as follows:
0': 25%B
2': 90%
3.5': 90%
4': 25%
6': 25%
flow rate is 0.5 ml/min.
First compound elutes at 2.3', second at 3.2'.

I have not used the diverter valve to keep the solventfront from the matrix, but there is one on the system, so I could definetely try this.

[lmh]

conclusion so far:
(1) the problem is in the sample, not the solvents or SPE
(2) the problem is a coeluting suppressor of some sort
(3) sticking to the same analytical column, changing pH and solvent has little effect.

So is it time to reconsider a radically different analytical column? If changing pH and solvent has little effect with the existing RP column, then you need to exploit some other characteristic of your analyte. Your original suggestion of a phenyl column is a possibility if your analyte, or the mysterious co-eluant, contain aromatic groups (run it in MeOH, not ACN, obviously).

[Ruth]

@ ctroster: The use of the diverter valve (so the flow during the first 2 minutes went to the waste instead of the MS) didn't decrease the matrix effect. Results were the same as if no diverter valve was used.

@ Ihm: Changing the column properties was also one of my first thoughts, however Uwe Neue didn't recommend this. My analytes contain aromatic groups, so a phenyl column might be possible... I still have to purchase a new column then, so I would like to have some certainty that a different column will help...

I also will perform some tests with the elution solvent, maybe that an optimization of this will help...

[Uwe Neue]

The common rules of thumb for changing the selectivity of the HPLC separation are as follows, in order of effectiveness:
1. Change the pH.
2. Change the solvent.
3. Change the column.

Solvent changes means going from methanol to acetonitrile, or vice versa.

Effective column changes are going from a C18 to one with an embedded polar group or to one with a phenyl group. Going from one C18 to another, or to a C8, does close to nothing.

[Camisotro]

The dwell volume of our LC system is around 250 µl. When I look at the pressure, there seem to be minimal delay (I roughly guess 1') between settings of the pump and the effective arrival of solvent in the column.

The gradient is set up as follows:
0': 25%B
2': 90%
3.5': 90%
4': 25%
6': 25%
flow rate is 0.5 ml/min.
First compound elutes at 2.3', second at 3.2'.

I have not used the diverter valve to keep the solventfront from the matrix, but there is one on the system, so I could definetely try this.

Okay, so how about starting with an isocratic hold, long enough to elute highly polar matrix components while trapping your analyte on the column; and then kicking up the gradient? If your dwell volume is around 250 µL then we can assume each change in solvent composition hits with only a 30 second delay or so.

Also I notice that you drop the solvent right down to initial conditions again after your last analyte elutes. Might it be helpful to leave it up at 90% B for a little while longer, in case you have some late-eluting compounds to wash off? Otherwise those compounds could remain trapped on the column at 25% and provide some unpredictable carryover.

So let's say:
0': 10% B
1.5': 10% -> this gives you effectively a 2 minute isocratic hold before the gradient
3.5': 90%
6': 90%
6.5': 10%
8.5': 10%

Also if you wanted to visualize exactly how long the matrix effect lasts at a certain isocratic setting, you could do post-column infusion of your analyte while injecting a sample onto the column. This will show you how long of an isocratic hold you would need at that setting to separate out interferences that affect the analyte in question.

[Ruth]

@ctroster: I tried already similar gradients before, but just to be sure I tried the one you recommended... Unfortunately, no change in matrix effect... I am afraid that by changing the mobile phase conditions the ion suppression won't dissappear.

I am now working on the elution solvent. If this doesn't help, I will purchase a phenyl column...

Any other suggestions are always welcome...

[lmh]

Ruth, if this sort of work comes up often in your lab, you could consider getting one of the column development kits that manufacturers offer, where they sell you short columns of 3 different sorts for method development. You might also get a good deal or even a trial column if you approach the column manufacturers with sufficient cheek asking about alternative specificities. Many are extremely keen to establish a customer-base for a new chemistry. But Uwe's right, I hate it when they try to sell me yet another nearly-C18.