Troubleshooting
Your recovery is low, and it is probably not the instrument
August 3, 2026
At Canadian Analytical Laboratories I worked on vitamin testing for natural health products on a Shimadzu LC-MS. The recovery was low and it varied between batches. The first suspicion was the instrument.
We checked the instrument. The instrument was fine.
The analyte was being lost before it ever reached the column.
That is the common case. When a number comes back low, the instrument is the visible thing in the room, so it gets blamed. The extraction happens upstream. It is done by hand, and it has more places to go wrong.
First, find out whether it is really recovery
The word gets used loosely. Three different things get called recovery, and they need different fixes.
You need three sets of samples at the same concentration:
- A. A neat standard in solvent. No matrix.
- B. Blank matrix extracted first, then spiked. The analyte never went through the extraction.
- C. Blank matrix spiked first, then extracted. The analyte went through everything.
| What you are measuring | Compare | What it tells you |
|---|---|---|
| Recovery | C against B | How much analyte survives the extraction |
| Matrix effect | B against A | How much signal the rest of the sample suppresses or enhances |
| Process efficiency | C against A | The two together. This is what you actually see when you inject |
This is the framework from Matuszewski and co-workers, and it is the one most regulators and reviewers expect to see.
Those three injections are an afternoon of work and they immediately remove half the possible causes. If the pre-extraction spike and the post-extraction spike agree, your extraction is fine and the problem is ionization. If they disagree, the analyte is being lost in the sample prep, and cleaning the source will not recover it.
Do this before you change anything else. I have watched labs re-tune, clean the source and swap a column while the extraction was losing half the analyte.
Where the analyte goes
Six places, and each one has a test that costs an injection.
| Where it goes | How to test it | The fix |
|---|---|---|
| Breakthrough. The bed is too small, you loaded too fast, or the load solvent was too strong, so the analyte goes into the waste on the load step | Collect the load fraction and inject it | A bigger bed, or a weaker load solvent |
| Wrong sorbent chemistry. A reversed-phase sorbent holds by hydrophobicity. A polar or ionized analyte is not retained | Check the pH of the load solution against the pKa of the analyte | Water-wettable polymeric or mixed-mode sorbent. For an ionizable compound the load pH matters more than the brand on the cartridge |
| The wash is too strong. The SOP says wash with 40% methanol and your compound elutes at 45% | Collect the wash and inject it | Weaken the wash, or move the elution further from it |
| The elution is too weak or too small. The analyte is on the sorbent and it stays there | Collect a second elution volume and inject it on its own | More volume, or a stronger elution solvent |
| Loss to surfaces. Some compounds stick to polypropylene, some to glass. Peptides and lipophilic compounds are the usual ones | Change the tube type, or let one sample sit longer before injection, and compare | Different tube material, a carrier or a modifier, shorter standing time |
| It was never released from the matrix. Bound to protein, locked in a tablet excipient, or still inside a capsule or a cell | Try a harsher release step on one aliquot and compare | Protein precipitation, enzymatic digestion, sonication, pH adjustment |
Breakthrough and an over-strong wash account for most low-recovery methods I have seen. The last row is not a cartridge problem. No extraction can recover what was never freed.
That was the vitamin problem.
The checks I run, in order
- Pre-extraction spike, post-extraction spike, neat standard. Extraction or ionization.
- Inject the load fraction and the wash.
- Inject a second elution aliquot.
- Check the pH of the load solution against the pKa of the analyte.
- Check what the sample touched: tube material, filter membrane, how long it sat.
Most of the time the answer is in the first two.
The filter nobody suspects
Syringe filters adsorb analyte. It is real, it is compound dependent, and PVDF, nylon, PTFE and PES all behave differently.
If the analyte turns out to be polar and badly retained as well, choosing a column for a polar analyte is the next thing to read.
If you filter your samples, prepare one filtered and one centrifuged instead, at the same concentration, and compare them. If you are losing 20% on the membrane you will see it immediately. This costs nothing to test and it gets skipped constantly.
If you want a second opinion
Tell me the analyte, the matrix, how you extract it now, and what recovery you are getting. I will tell you which of these checks I would run first and what I expect you will find.
If it turns out your extraction is fine and the problem is ionization, that is the answer you will get. There is no cartridge I can sell you for that one.
If it is bigger than one question, and you want the whole method looked at rather than one symptom, that is the method audit. You send the conditions and a chromatogram, I send back a written diagnosis in 5 working days. CA$300, and if I cannot find a cause worth acting on you do not pay.
Source
The three-set framework for recovery, matrix effect and process efficiency is from B. K. Matuszewski, M. L. Constanzer and C. M. Chavez-Eng, Strategies for the assessment of matrix effect in quantitative bioanalytical methods based on HPLC-MS/MS, Analytical Chemistry 2003, 75, 3019 to 3030. It is the version regulators and reviewers expect to see.