SPE Cartridge Selection and Method Optimization: Sorbent Choice and the Wash-Step Trade-Off

Solid phase extraction (SPE) fails in a specific place. Not at sorbent selection — sorbent selection is largely determined once you know your analyte’s pK_a and log P. It fails at the wash step, where every choice that makes the extract cleaner also makes the recovery worse, and where the two goals are in direct, measurable competition.

This guide covers the decision rules: how retention mechanism follows from analyte chemistry, how to set load and elution pH so retention is deterministic rather than hopeful, how to find the highest wash strength your analyte survives, and how to recognise the case where no wash strength works and the sorbent itself has to change. It closes with a clause-by-clause read of a real regulatory method — EPA Method 537.1 — which is unusual and instructive because it locks its extraction steps: the analyst is forbidden from changing them, so every parameter in it is a settled decision you can read off the page.

This is a method-development guide, not a buying guide. It names sorbent chemistries, not part numbers, because the chemistry is what transfers between suppliers and the part number is not.

Sorbent selection is a retention-mechanism decision, not a product decision

The question a cartridge answers is: by what physical mechanism will this analyte be held while the matrix passes through? There are four mechanisms in routine use, and the analyte’s chemistry usually eliminates three of them.

The four mechanisms

Mechanism	Typical sorbent chemistry	Retains	Load solvent	Elution solvent
Reversed phase	Silica bonded with C18 (octadecyl), C8, phenyl; or polymeric styrene-divinylbenzene (SDVB) and hydrophilic-lipophilic-balanced copolymers	Non-polar to moderately polar analytes, held by hydrophobic (van der Waals) interaction	Aqueous — water, buffer, plasma, urine	Organic — methanol, acetonitrile
Normal phase	Bare silica, alumina, Florisil (magnesium silicate), diol, aminopropyl, cyanopropyl	Polar analytes, held by hydrogen bonding, dipole and pi-pi interaction	Non-polar organic — hexane, dichloromethane	More polar organic — ethyl acetate, methanol
Ion exchange	Sulfonic acid (SCX), carboxylic acid (WCX), quaternary amine (SAX), secondary/tertiary amine (WAX)	Charged analytes, held by electrostatic attraction to an oppositely charged sorbent	Aqueous at a pH that charges the analyte	pH shift that neutralises one partner, or a competing counter-ion at high ionic strength
Mixed mode	A reversed-phase backbone carrying an ion-exchange functional group	Charged analytes, by both hydrophobic and electrostatic interaction simultaneously	Aqueous at a pH that charges the analyte	Organic plus a pH shift, applied together

Reversed phase on silica and reversed phase on a polymer are not interchangeable, and the difference matters operationally rather than chemically:

• Bonded silica depends on a solvated layer of alkyl chains. If the bed runs dry, those chains collapse against the silica surface and retention drops — which is why conditioning discipline is a real failure mode rather than a formality.
• Polymeric sorbents are described in vendor technical literature as water-wettable — tolerant of the bed running dry — and stable across a far wider pH range than bonded silica, whose siloxane bond hydrolyses at high pH and whose silica backbone dissolves. Treat the specific pH stability range as a vendor specification for the product in front of you, not a general property of polymers; check the certificate, not the category.
• Polymeric beds also generally offer higher capacity per unit mass, which matters most for dirty samples, for reasons covered under bed mass below.

The two numbers that pick the mechanism

Almost the whole decision reduces to two properties of the analyte:

• log P (or log D, which is log P corrected for ionisation at a stated pH) describes hydrophobicity, and therefore predicts reversed-phase retention. A high log P analyte will be well retained on C18 from an aqueous load; a log P near or below zero will not be, and reversed phase alone is the wrong choice.
• pK_a determines whether the analyte is charged at a given pH, and therefore whether ion exchange is available at all — and, critically, whether you can switch the charge off to elute it.

Note that log D, not log P, is the number that predicts real behaviour for an ionisable compound, because an ionised molecule is far more water-soluble than its neutral form. Quoting log P for a compound you are loading at a pH where it is ionised will over-predict retention.

The “strong” and “weak” trap

This is the most common misreading in SPE, and it costs methods. In SCX / SAX / WCX / WAX, “strong” and “weak” describe the acid or base strength of the sorbent’s functional group — not how tightly it binds.

• A strong exchanger (SCX, sulfonic acid; SAX, quaternary amine) is permanently charged across the usable pH range. Its charge cannot be switched off.
• A weak exchanger (WCX, carboxylic acid; WAX, secondary or tertiary amine) has a pH-dependent charge. You can neutralise the sorbent itself.

This produces a genuinely determinate selection rule, and it is the opposite of what the names suggest:

Analyte	Can you neutralise the analyte?	Choose	Elute by
Weak base (e.g. a typical amine drug, pKa ~8–10)	Yes — raise pH above pKa + 2	Strong cation exchange / mixed-mode strong cation	Neutralising the analyte with base in organic solvent
Weak acid (e.g. a carboxylic acid, pKa ~4)	Yes — lower pH below pKa − 2	Strong anion exchange / mixed-mode strong anion	Neutralising the analyte with acid in organic solvent
Strong base — quaternary ammonium, permanently cationic; or a base with pKa above ~10	No — it stays charged at any workable pH	Weak cation exchange (WCX)	Neutralising the sorbent — drop the pH below the carboxylate’s pKa and the sorbent lets go
Strong acid — sulfonate, sulfate, phosphonate, or an acid with pKa below ~1	No — it stays charged at any workable pH	Weak anion exchange (WAX)	Neutralising the sorbent — raise the pH above the amine’s pKa and the sorbent lets go

Put plainly: strong exchangers are for analytes you can switch off; weak exchangers are for analytes you cannot. Trying to elute a quaternary amine from a strong cation exchanger means driving it off with high ionic strength — which loads the extract with non-volatile salt, exactly what you do not want going toward an electrospray source.

The pH rule that makes retention deterministic

Every SPE step that depends on ionisation state depends on the Henderson-Hasselbalch relationship, which for an acid gives the ratio of ionised to neutral form as 10^{(pH − pK_a)}. Two consequences follow directly, and they are arithmetic, not heuristic:

• At pH = pK_a, the analyte is 50% ionised. This is the worst possible operating point: half of your analyte is behaving one way and half the other, and small pH drift moves the split.
• At pH = pK_a ± 2, the ratio is 100:1 — approximately 99% of the analyte is in a single form. This is the origin of the widely-taught “two pH units” rule, and it is why the rule is two and not one (90:10 is not good enough for a quantitative method) and not four (you gain a decimal place and risk degrading the analyte or the sorbent).

Applied to each mechanism:

Sorbent	Load at	Wash at	Elute at
Reversed phase, acidic analyte	pH ≤ pKa − 2 (analyte neutral, maximally retained)	Same pH as load — never change pH mid-wash by accident	Organic; pH is irrelevant to elution here
Reversed phase, basic analyte	pH ≥ pKa + 2 (analyte neutral)	Same pH as load	Organic
Cation exchange (analyte a base)	pH ≤ pKa − 2 (analyte charged) — the opposite of reversed phase	Anything that does not disrupt the ionic bond: strong organic is fine	pH ≥ pKa + 2 with base, in organic
Anion exchange (analyte an acid)	pH ≥ pKa + 2 (analyte charged)	Strong organic is fine	pH ≤ pKa − 2 with acid, in organic

The row that surprises people is the third one. On reversed phase you neutralise a base to retain it; on cation exchange you charge the same base to retain it. Load pH is therefore not a property of the analyte — it is a property of the analyte and the mechanism you chose, and a method transferred from an RP cartridge to a mixed-mode one without inverting the load pH will recover nothing and look like a cartridge fault.

The five-step loop, and what each step is actually for

Every SPE method is the same five steps. What differs is what you are willing to lose at each one.

1. Condition

Wet the bed and solvate the stationary phase — typically methanol for a reversed-phase sorbent. On bonded silica this is what stands the alkyl chains up so they are available to the analyte. Failure mode: insufficient conditioning gives low and, worse, irreproducible recovery, because the fraction of the bed that is actually functional varies cartridge to cartridge.

2. Equilibrate

Replace the conditioning solvent with something matching the sample — usually water or the load buffer, at the load pH. Failure mode: residual methanol left in the bed acts as an unintended wash, eluting weakly-retained analytes during the load. If your earliest-eluting analyte has poor recovery and everything else is fine, look here first.

On bonded silica, do not let the bed run dry between conditioning and load. Water-wettable polymeric sorbents are tolerant of this; silica is not.

3. Load

Pass the sample through at a controlled flow rate. Retention is a mass-transfer process and mass transfer takes time, so flow rate is a real method parameter, not a convenience setting. Failure mode: loading too fast produces breakthrough — analyte passing through unretained — which looks identical to low recovery from any other cause.

To distinguish breakthrough from every other recovery loss, run the stacked-cartridge test: put a second, identical cartridge in series below the first, load as normal, then elute and analyse the lower cartridge separately. Analyte found on the lower cartridge is analyte that passed through the upper one. That is breakthrough, and the fix is a slower flow rate, a larger bed, or a weaker load solvent — not a different elution.

4. Wash

Remove matrix while keeping analyte. This is the whole game, and it has its own section below.

5. Elute

Displace the analyte in the smallest volume that recovers it. Two practical rules:

• Two small aliquots beat one large one. Elution is an equilibrium process, and a fresh aliquot of solvent re-establishes a favourable partition that a continuing single aliquot has already exhausted. EPA Method 537.1 does exactly this — two separate 4 mL methanol aliquots rather than one 8 mL pass.
• Slow down. Method 537.1 specifies a low enough vacuum that “the solvent exits the cartridge in a dropwise fashion.” Contact time is doing real work.

If the eluate is going on to be evaporated, or if you are eluting with a solvent that is not miscible with water, dry the bed under vacuum first. Method 537.1 draws air or nitrogen through the cartridge for 5 minutes at 10–15 in. Hg before elution.

The wash step: recovery versus cleanliness

This is where methods are made or ruined, and it is worth stating the problem precisely rather than as an aphorism.

The wash step is a competition on a single axis. On a reversed-phase sorbent, that axis is elutropic strength — in practice, percent organic. Every interferent and every analyte has some organic percentage at which it starts coming off the bed. The wash works if, and only if, the analyte’s threshold sits above the interferent’s, with enough gap between them to place a wash in the middle.

That gives a real decision rule:

Choose the strongest wash whose strength is still strictly below the analyte’s breakthrough threshold. If the interferents you need to remove come off at or above the analyte’s threshold, no wash strength solves the problem — the fix is a different retention mechanism, not a different wash.

Finding the ceiling: the wash-strength ladder

Do not guess this number. Measure it, once, and reuse it for the life of the method:

1. Prepare a set of identical fortified samples in real matrix — not in clean solvent, because matrix competes for the sorbent and shifts the thresholds.
2. Load each identically.
3. Wash each with a different organic strength — for example 0%, 5%, 10%, 20%, 30%, 40% methanol in water.
4. Elute all of them identically, and measure both analyte recovery and a cleanliness metric (see below) at every rung.

Recovery against wash strength has a characteristic shape: flat, then a knee, then a collapse. Cleanliness improves monotonically. Take the rung immediately below the knee, and then back off one further step as your operating point, so that normal run-to-run variation in pH, temperature and matrix load cannot push you over the edge mid-batch.

An illustrative shape — these are not universal numbers, they depend entirely on the analyte, the matrix and the sorbent, and yours will differ:

Wash (% methanol in water)	Analyte recovery	Extract cleanliness	Read
0%	High	Poor	Leaves matrix on the bed that will co-elute
10%	High	Better	Still on the flat part of the curve
20%	High	Good	The knee has not been reached — operating point
30%	Beginning to fall	Very good	The knee. Tempting, and fragile
40%	Collapsed	Excellent	You have washed the analyte down the drain

The 30% row is the trap. It looks like the best trade — nearly all the cleanliness, only a little recovery lost — and a validation run at a single time point will pass on it. It fails later, intermittently, when a dirtier-than-usual sample or a slightly warm lab shifts the knee left by a few percent and a whole batch quietly under-recovers.

Why mixed-mode changes the trade-off rather than improving it

The reason mixed-mode sorbents dominate difficult matrices is not that they retain better. It is that they make the wash orthogonal to the retention mechanism.

On a mixed-mode cartridge holding a basic analyte by cation exchange, the analyte is anchored electrostatically. Organic solvent does not break an ionic bond. So you can wash with 100% methanol — which strips essentially all neutral hydrophobic matrix, lipids, phospholipids and pigment — without touching the analyte at all. You can additionally wash with dilute acid to remove anything anionic. Neither wash sits on the same axis as the retention, so neither one costs recovery.

The trade-off has not been optimised; it has been dissolved. That is the actual argument for mixed-mode, and it is why a mixed-mode method for a plasma or soil extract will usually beat a well-optimised plain reversed-phase method rather than merely matching it. The cost is that mixed-mode methods have more parameters to get wrong — load pH, wash pH, wash organic strength, elution pH and elution organic strength all interact, and inverting the load pH by habit (see above) recovers nothing.

Cleanliness has to be measured, not assumed

“Cleaner” is not a visual judgement about a less yellow extract. For anything ending at a mass spectrometer, the operative definition of clean is how much the co-eluting matrix suppresses or enhances ionisation, and that is measured by post-extraction addition.

Run three sets at the same nominal concentration:

Set	Preparation	Response
A — neat standard	Analyte in clean solvent, no extraction	A
B — post-extraction spike	Blank matrix extracted first, then spiked into the finished extract	B
C — pre-extraction spike	Blank matrix spiked before extraction, then extracted	C

Three distinct quantities fall out, and conflating them is the most common reporting error in SPE method development:

• Matrix effect = B / A. Below 1 is ion suppression; above 1 is enhancement. This measures cleanliness. It is entirely unaffected by how much analyte you lost.
• Extraction recovery = C / B. This measures how much analyte the cartridge actually gave back. It is entirely unaffected by ion suppression.
• Process efficiency = C / A. The overall number, which is the product of the other two.

A method reporting only “recovery” as C/A cannot tell you which knob to turn. A process efficiency of 50% caused by 50% ion suppression needs a stronger wash; the identical 50% caused by analyte washing off the bed needs a weaker one. Without the three-set experiment, the two are indistinguishable and you have a 50% chance of making it worse.

For regulated bioanalysis, ICH M10 (Bioanalytical Method Validation and Study Sample Analysis, adopted 2022) sets the expectation formally: matrix effect is evaluated using at least three replicates of low and high quality-control samples prepared in matrix from at least six different sources or lots, and for each individual lot the accuracy must fall within ±15% of nominal with a coefficient of variation not greater than 15%. The six-lot requirement is the part most often skipped, and it exists precisely because matrix effect is not a constant — one donor’s plasma or one field site’s soil will suppress differently from the next, and a method validated against a single lot has not been validated against the variability that will actually break it.

When the wash step cannot win

Some problems are not wash problems. Recognising them early saves a week of ladder experiments that converge on nothing.

Symptom	Likely cause	Fix
Cleanliness improves and recovery falls together across every wash strength, with no flat region	Analyte and interferent have overlapping elution windows on this mechanism	Change mechanism — move to mixed-mode so the wash becomes orthogonal
Recovery is high but ion suppression persists no matter how hard you wash	The suppressing species co-elutes with the analyte on the analytical column, not just the cartridge — phospholipids are the classic case	A dedicated phospholipid-removal step, or shift the chromatography so the suppression region no longer overlaps the peak — see ESI source tuning
Recovery is poor and unaffected by wash strength at all	Not a wash problem — breakthrough on load, or incomplete elution	Stacked-cartridge test to distinguish the two
Recovery varies wildly between cartridges from the same lot	Conditioning inconsistency, or the bed ran dry	Fix conditioning discipline before changing anything else
Recovery falls only for the most polar analytes in a multi-analyte panel	The wash is tuned for the mid-panel and is too strong for the early-eluting end	Set the wash by the weakest-retained analyte in the panel, not the average

That last row is the multi-analyte constraint, and it is unforgiving: in a panel, the wash strength is set by whichever analyte comes off first, so adding one polar compound to an established panel can force the wash down and degrade the cleanliness for everything else. This is a genuine reason to split a panel into two methods rather than a failure of optimisation.

Worked example: EPA Method 537.1, clause by clause

Most SPE guidance is advisory. EPA Method 537.1 (Revision 2.0, Determination of Selected Per- and Polyfluorinated Alkyl Substances in Drinking Water by Solid Phase Extraction and Liquid Chromatography/Tandem Mass Spectrometry) is the opposite: it fixes the extraction and forbids you from changing it, which makes it a rare chance to read a fully-committed set of SPE decisions and see the reasoning behind each.

What the method fixes

Clause	What it specifies	The method-development point
§6.9.1	0.5 g, 6 mL SPE cartridges containing styrene-divinylbenzene (SDVB) polymeric sorbent. “The sorbent may not be modified with monomers other than SDVB.”	A polymeric reversed phase, not bonded silica. PFAS analytes span a wide chain-length range; the polymer’s higher capacity and pH tolerance matter, and the explicit ban on other monomers rules out hydrophilic-modified copolymers whose selectivity differs.
§11.4.1	Rinse with 15 mL methanol, then 18 mL reagent water, “without allowing the water to drop below the top edge of the packing.” If the cartridge goes dry, “the conditioning must be started over.”	The strongest possible statement of the conditioning-discipline rule: a dry bed is not recoverable by carrying on, it is a restart.
§11.4.3	Wash: two 7.5 mL aliquots of reagent water — no organic at all. Then draw air or nitrogen 5 minutes at 10–15 in. Hg.	The wash-step trade-off, resolved at its most conservative setting. See below.
§11.4.4	Elute with 4 mL methanol, then repeat with a second 4 mL aliquot, at low vacuum “such that the solvent exits the cartridge in a dropwise fashion.”	Two aliquots, not one; contact time deliberately extended.
§11.4.1 and §11.4.4 notes	“If low recoveries are observed for PFBS and PFHxA during the IDC, recoveries may be improved by allowing a one- or two-minute soak time after each addition.”	The method names which analytes fail first — and they are the short-chain, least hydrophobic ones. That is the wash-and-elution ceiling being set by the weakest-retained analyte in the panel, written into a federal method.
§1.6	The analyst may modify evaporation technique, LC column, mobile phase, LC conditions and MS/MS conditions. “Changes may not be made to sample collection and preservation (Sect. 8), the sample extraction steps (Sect. 11.4), or to the quality control requirements (Sect. 9).”	The regulator’s judgement about where method variability actually bites: chromatography is negotiable, sample preparation is not.

Why the wash is water only

Method 537.1 washes with reagent water and nothing else. Read against the trade-off above, that is a deliberate choice at the far conservative end of the ladder, and the method’s own notes explain why: the analytes at risk are PFBS and PFHxA — the shortest-chain, least hydrophobic compounds in the panel. On a reversed-phase polymer, those are the first analytes off the bed as organic strength rises. Any organic in the wash buys cleanliness at their expense, and the method’s list is fixed, so they cannot be dropped from the panel to allow a stronger wash.

The method then recovers the lost cleanliness elsewhere, which is the part worth copying: it uses a 250 mL sample concentrated to a 1 mL final volume — a 250-fold concentration factor — and isotopically labelled surrogates added to every field and QC sample “to monitor the extraction efficiency of the method analytes.” Rather than clean the extract harder, it accepts a dirtier extract and instruments the method so that suppression and recovery loss are measured on every single sample instead of assumed. §9.1.1 makes the same point from the other direction, requiring that any LC modification “still produce conditions such that co-elution of the method analytes is minimized to reduce the probability of suppression/enhancement effects.”

That is a transferable pattern. When the wash-versus-recovery trade-off has no good answer — because the panel’s weakest analyte pins the wash at the bottom of the ladder — the remaining moves are to instrument the loss with labelled internal standards, and to separate the suppression chromatographically rather than removing it during preparation.

One correction worth flagging: Method 537.1’s SPE sorbent is SDVB, not C18. C18 appears in the method as the LC column (§11 scope: “a 10 µL injection is made into an LC equipped with a C18 column”). The two are easy to conflate when reading a method summary, and conflating them leads to specifying the wrong cartridge — which §6.9.1 explicitly does not permit.

Bed mass, format and capacity

Bed mass is sized by total load on the sorbent, and the mistake is to size it by analyte mass. In a trace method the analyte is a vanishing fraction of what the cartridge is actually holding; the capacity is consumed by matrix. A urine or soil extract will exhaust a small bed at concentrations where the analyte itself is nowhere near saturating anything, and the symptom is breakthrough that scales with sample dirtiness rather than with analyte concentration.

Vendor technical literature commonly quotes a working capacity for bonded-silica SPE sorbents on the order of a few percent of the bed mass, with polymeric sorbents higher — but this figure varies by sorbent chemistry, by supplier and by what else is competing for the surface, so treat any quoted percentage as a starting point for an experiment rather than a specification. The number that governs your method is the breakthrough volume in your actual matrix, and the stacked-cartridge test above measures it directly in one run.

Format follows throughput and sample volume, not chemistry:

• Cartridges — the default. Bed masses from tens of milligrams to grams; sample volumes from under a millilitre to hundreds of millilitres.
• 96-well plates — same chemistry, small beds, designed for parallel processing of small biological samples. The constraint is bed mass, so a plate suits clean-ish, low-volume samples and not large environmental ones.
• Disks — a thin, wide bed. Large cross-section means high flow without the pressure drop of a deep bed, which is why they appear in high-volume water methods where a cartridge would take impractically long to load.

Solvent volume is also a real cost and a real hazard, not only a convenience: the conditioning and elution steps of a routine SPE method consume more organic solvent than the analysis itself, which is why sample preparation is usually the largest single line in a lab’s solvent budget — see green chemistry in the lab for solvent-selection guidance that applies directly here.

SPE in the wider workflow

SPE is one of several ways to get a sample ready, and choosing between them is a separate decision from optimising the one you chose:

• Protein precipitation is faster and cheaper than SPE and removes protein well, but removes almost nothing else — phospholipids in particular survive it, which is why precipitation-only bioanalytical methods so often show suppression that no amount of source tuning fixes.
• Liquid-liquid extraction partitions on a single axis (solvent polarity) and offers no equivalent of the wash step. It can be excellent for a clean, well-behaved analyte and offers no route to selectivity when it is not.
• QuEChERS (Quick, Easy, Cheap, Effective, Rugged and Safe) is a buffered salt-out extraction followed by dispersive SPE — the sorbent is mixed into the extract rather than packed in a bed, and it is standardised in pesticide-residue work, where pesticide residue testing methods use it routinely. Dispersive cleanup trades the wash step’s fine control for speed: you cannot run a wash-strength ladder on a sorbent that is stirred in and centrifuged out.
• Cartridge SPE is what you use when you need the wash step — that is, when selectivity has to be engineered rather than hoped for.

The reason this matters downstream is quantitative. Sample cleanliness is the single largest determinant of ion suppression at an electrospray source, because co-eluting matrix competes for charge and for space at the droplet surface. Suppression that originates in a poorly-washed extract cannot be tuned out at the source and cannot be corrected by MRM transition optimisation — a cleaner transition monitors a suppressed ion just as faithfully as a dirty one. It shows up as a depressed and drifting response, degraded limits of detection and quantitation, and a calibration curve that is linear in solvent and curved in matrix. The wash step is where that is decided, several hours upstream of the instrument that gets blamed for it.

For the separations context behind sorbent behaviour, column chromatography covers the same adsorption and elutropic-strength principles at preparative scale, and thin-layer chromatography is a fast, cheap way to scout the solvent strength that will move your compound off a silica surface before you commit a cartridge to it. How the cleaned extract is then separated and detected is covered in LC-MS coupling and, for the instrument choice itself, in HPLC vs UPLC vs UHPLC. This guide sits in the lab equipment cluster.

A method-development sequence that converges

1. Look up pK_a and log D at the pH you intend to load at. If you have several analytes, do this for all of them and note the extremes.
2. Pick the mechanism from the table above. If any analyte is permanently charged, that decides it — weak exchanger.
3. Set load pH two units from the pK_a, in the direction the mechanism requires. Confirm the direction against the mechanism, not against habit.
4. Establish elution first, not last. Load a clean fortified sample, skip the wash entirely, and confirm you can get the analyte back. A method that cannot recover analyte from clean solvent will never recover it from matrix, and diagnosing that after you have started optimising the wash wastes the whole ladder.
5. Confirm no breakthrough with the stacked-cartridge test in real matrix at the intended load volume and flow rate.
6. Run the wash-strength ladder in real matrix, measuring recovery and matrix effect at every rung.
7. Take the rung below the knee, then back off one more.
8. Re-measure matrix effect and recovery separately with the three-set experiment at the chosen operating point, across multiple matrix lots — six, if the method is headed for regulated bioanalysis.
9. Add labelled internal standards that co-elute with the analytes, so that whatever suppression remains is corrected per-sample rather than assumed away.

Step 4 is the one most often done out of order, and it is the cheapest to get right.

Frequently asked questions

How do I choose an SPE cartridge for a new analyte?

Look up its pK_a and log D first. If it is neutral across the workable pH range and reasonably hydrophobic, reversed phase. If it is ionisable and you can neutralise it, a strong ion exchanger or mixed-mode strong exchanger. If it is permanently charged — a quaternary amine, a sulfonate — use a weak exchanger, because you will elute by neutralising the sorbent rather than the analyte. Sorbent format and bed mass come after the mechanism, not before.

What is the difference between C18 and polymeric SPE cartridges?

Both retain by reversed-phase interaction. Bonded C18 silica depends on a solvated alkyl layer, so it loses retention if the bed runs dry and its usable pH range is limited by the stability of the silica and the siloxane bond. Polymeric sorbents are described in vendor literature as water-wettable and are stable across a wider pH range, generally with higher capacity per unit mass. Check the specific pH and capacity claims on the product’s own documentation rather than assuming them from the category.

What does “strong” mean in strong cation exchange SPE?

It describes the sorbent’s functional group, not the strength of the binding. A strong cation exchanger carries a sulfonic acid group that stays negatively charged at any usable pH; a weak cation exchanger carries a carboxylic acid whose charge you can switch off by lowering the pH. This is why a permanently-charged analyte belongs on a weak exchanger — it is the only one of the pair you can persuade to let go without resorting to high ionic strength.

How strong should the SPE wash solvent be?

As strong as possible while staying strictly below the point at which your weakest-retained analyte starts coming off, with a margin for run-to-run variation. Determine that point empirically with a wash-strength ladder in real matrix, measuring recovery at each rung, and take the rung below the knee rather than the knee itself. There is no transferable default percentage; it depends on the analyte, the matrix and the sorbent.

Why is my SPE recovery low?

Four causes, distinguishable by different tests. Breakthrough during load — confirmed with a stacked second cartridge. Analyte lost in the wash — confirmed by collecting and analysing the wash fraction. Incomplete elution — confirmed by a second elution aliquot that still contains analyte. Poor conditioning or a dried bed — indicated by high variability between nominally identical cartridges. Test in that order, because breakthrough and wash loss are the two most common and the cheapest to check.

Does a cleaner extract always mean a better method?

No. Cleanliness is worth having only up to the point where buying more of it costs analyte. Measure the two separately using post-extraction addition — matrix effect and extraction recovery are different quantities with different fixes, and a single combined “recovery” number cannot tell you which one is failing.

Can I substitute an equivalent cartridge in a regulatory method?

It depends on the method’s own flexibility clause, which you must read rather than assume. EPA Method 537.1 §6.9.1 specifies 0.5 g, 6 mL SDVB cartridges and states that the sorbent may not be modified with monomers other than SDVB; §1.6 permits changes to LC and MS conditions but explicitly not to the sample extraction steps in §11.4. Other methods are more permissive. Where a method allows substitution, it will normally still require you to repeat an initial demonstration of capability.

Is SPE always better than protein precipitation or liquid-liquid extraction?

No — it is more selective and more controllable, and correspondingly slower and more expensive. Protein precipitation is appropriate when the remaining matrix does not suppress your analyte, which you should verify by measurement rather than assume. SPE earns its cost specifically when you need the wash step, because that is the only one of the three techniques that gives you an independent selectivity control between capture and release.

Sources

• U.S. Environmental Protection Agency, Method 537.1: Determination of Selected Per- and Polyfluorinated Alkyl Substances in Drinking Water by Solid Phase Extraction and Liquid Chromatography/Tandem Mass Spectrometry, Revision 2.0. Sections 1.6, 6.9.1, 9.1.1, 11.4.1, 11.4.3 and 11.4.4 as cited above.
• ICH, M10 Bioanalytical Method Validation and Study Sample Analysis (Step 4, 2022), matrix-effect evaluation requirements.
• Sorbent chemistry, water-wettability and pH-stability characteristics as described in the published technical literature of SPE sorbent manufacturers. These are vendor performance claims, presented here as vendor data rather than independent validation — confirm any specific figure against the documentation for the product you are actually using.

Recovery figures in this guide are illustrative and stated with their conditions. Sorbent chemistry behaviour generalises; specific recovery percentages do not — they are a property of a particular analyte in a particular matrix on a particular sorbent, and must be established in your own laboratory.