Solid Phase Microextraction: Fibre Selection

Most SPME methods that fail in validation did not fail on sensitivity. They failed because the analyst picked a fibre from whatever was in the drawer, got beautiful chromatograms on standards in clean water, and then found that the calibration curve bent over in real matrix, or that a high-level sample suppressed everything around it. Both of those are symptoms of one decision made without thinking about it: whether the coating extracts by absorption or by adsorption.

That is the first of the two choices that decide an SPME method. The second is sampling mode — headspace, direct immersion, or membrane-protected — which is far more a question of fibre lifetime and matrix tolerance than of raw sensitivity. Everything else (time, temperature, salt, pH, agitation) is optimisation around those two.

Absorbent versus adsorbent coatings: the distinction that sets your linear range

Commercial SPME coatings split cleanly into two mechanistic families, and Supelco’s own SPME training material draws the line explicitly:

Absorbent (liquid film) coatings. Polydimethylsiloxane (PDMS) at 7, 30 and 100 µm — non-polar; polyacrylate (PA) — polar; polyethylene glycol (PEG/Carbowax) — polar. Analytes partition into the bulk of a liquid polymer film. Capacity scales with film thickness and analytes do not compete for sites.
Adsorbent (particle) coatings. Carboxen/PDMS, PDMS/DVB, DVB/Carboxen/PDMS, Carbowax/DVB, and Carbowax/templated resin (an HPLC-only phase). Analytes are trapped on the surface and in the pores of suspended porous particles. Surface area is enormous, but the number of sites is finite, so analytes can and do compete.

The practical consequences follow directly. Supelco’s summary is blunt: adsorbent fibres “physically trap” analytes on a porous, high-surface-area material and “analytes may compete for sites,” giving fibres “limited capacity”; absorbent fibres extract by partitioning, “analytes do not compete for sites,” and the fibres “can have high capacity.” The peer-reviewed literature says the same thing in kinetic terms: absorption is non-competitive and therefore linear over a broad concentration range, while adsorption saturates when high-affinity compounds are abundant, and higher-affinity molecules then displace lower-affinity ones already on the surface.

So the choice is a genuine trade-off, not a ranking:

Trace-level target list, concentrations that stay within a narrow band → adsorbent. You will get detection limits an absorbent film cannot reach.
Wide or unpredictable concentration range, screening work, samples that sometimes come in hot → absorbent. You give up sensitivity and buy back linearity and freedom from displacement.
Non-targeted profiling of a complex matrix → adsorbent will find more compounds, but treat the peak areas as semi-quantitative unless you can demonstrate that the abundant matrix volatiles are not displacing the minor ones.

If you are stuck with an adsorbent fibre and hitting saturation, the standard fix is to shorten the extraction time. Moving further from equilibrium loads less onto the coating and extends the usable concentration range, at the cost of higher detection limits and a much stronger dependence on timing precision.

Matching the coating to molecular weight and volatility

Supelco ranks the commercial phases along an approximate analyte molecular weight axis running roughly 0–450, from weakest to strongest retention: templated resin, 7 µm PDMS, 30 µm PDMS, 100 µm PDMS, PDMS/DVB, DVB/Carboxen, Carboxen. Read it as a rule of thumb: the more strongly retentive the phase, the smaller and more volatile the analyte it suits.

The low end: why volatile gases need Carboxen

This is the case where the wrong fibre gives you literally nothing. In Supelco’s comparison of small hydrocarbons at 1 ppm in air with a 10-minute extraction, a 100 µm PDMS fibre returned zero response for ethane, propane and butane, and only 230 and 460 area counts for pentane and hexane. A PDMS/DVB fibre gave zero for ethane and propane, 340 for butane, 2,150 for pentane and 9,280 for hexane. Carboxen/PDMS gave 750, 20,000, 72,100, 108,000 and 105,000 respectively.

The reason is structural. Carboxen is a microporous carbon: Supelco characterises its pore distribution as micropores (<20 Å), mesopores (20–500 Å) and macropores (>500 Å), with a Carboxen-1006 particle carrying roughly 0.29 mL/g of micropore, 0.23 mL/g of mesopore and 0.26 mL/g of macropore volume. Micropores are the right size to trap C2–C5 molecules that simply pass through a liquid PDMS film.

The high end: where Carboxen stops helping

The same micropores that make Carboxen brilliant for gases make it a poor choice for large, planar molecules. Across an n-alkane series, Carboxen/PDMS response climbs steeply into the C14–C16 region and then collapses by more than an order of magnitude by C18, while DVB-containing phases hold up much better at the top of the range. For polycyclic aromatic hydrocarbons the failure mode is desorption, not extraction: the analyte goes on and does not come off cleanly, so you see poor recovery and carryover rather than no peak.

The triple-phase fibre exists to bridge the gap. A DVB/Carboxen/PDMS fibre is physically layered — roughly 60 µm of DVB outside 25 µm of Carboxen-1006 on a 110 µm fused silica core — so larger analytes are retained in the outer mesoporous DVB layer while small volatiles migrate through to the inner micropores.

Polarity, and the coating’s limits

Polarity matching still matters, but less than analysts expect. Polyacrylate and Carbowax phases outperform PDMS on polar semivolatiles — phenols, nitro compounds, aromatic acids and amines — but polyacrylate also extracts non-polar aromatics reasonably well, so treat “polar coating for polar analyte” as a starting hypothesis rather than a law. If your analytes are so polar or so involatile that no coating gives useful headspace partitioning, the answer is usually chemical rather than physical: derivatise. SPME supports both on-fibre derivatisation (dope the fibre with reagent, then sample) and post-extraction derivatisation, and the same reagent chemistry described in our guide to derivatization for GC-MS applies.

If you have to pick one fibre to start with

Published head-to-head comparisons converge on the triple phase. In a PLOS ONE study that screened eight commercial fibres (65 µm DVB/PDMS, 85 µm PA, 75 and 85 µm Carboxen/PDMS, 100 and 7 µm PDMS, 60 µm PEG, and 50/30 µm DVB/Carboxen/PDMS) against the human faecal volatile metabolome, the 50/30 µm DVB/Carboxen/PDMS fibre identified the most compounds (47), just ahead of 85 µm Carboxen/PDMS (46) and 75 µm Carboxen/PDMS (45) — though for raw peak count without identification the 65 µm DVB/PDMS fibre led with 122 peaks. A separate optimisation study on a cooked soup matrix compared 85 µm Carboxen/PDMS, 65 µm PDMS/DVB, 100 µm PDMS, 50/30 µm DVB/Carboxen/PDMS and 85 µm PA, and again selected the 50/30 µm three-layer coating.

Note also what that first study implies: a five-fibre set was needed to reach roughly 90% metabolite coverage. No single fibre gives you the whole volatilome. If your work is non-targeted, plan for a small panel, not a favourite.

Headspace, direct immersion, or membrane-protected

ASTM D6520-18, the active standard practice for SPME of water, explicitly covers extraction from “water and its headspace” — the choice is yours to justify.

Headspace (HS-SPME)

Default to headspace whenever the analytes will partition into the vapour phase at your working temperature. The fibre never touches the matrix, so proteins, lipids, particulates and salts cannot foul or abrade the coating, and the coating lasts far longer. For volatile analytes headspace is also usually no slower than immersion, because mass transfer through a gas is fast compared with diffusion across the stagnant aqueous boundary layer that forms around an immersed fibre.

Reviews of SPME in environmental and food analysis recommend headspace specifically when the sample contains high-molecular-weight, non-volatile constituents such as proteins, when the matrix is greasy or oily, and when there is a risk of physically damaging the fibre.

Direct immersion (DI-SPME)

Use direct immersion when the analyte is too involatile to reach the headspace in useful quantity — most drugs, pesticides, phthalates, and higher-boiling semivolatiles. The cost is fibre life: irreversible adsorption of macromolecules from the matrix shortens coating lifetime and, on adsorbent phases, occupies the very sites you need. Immersion also demands agitation to break down the depletion zone around the fibre.

Overcoated fibres exist for exactly this problem; a review of commercially available SPME technologies reports that Supelco’s overcoated fibres extend fibre life by 75–100% by reducing matrix build-up on the coating.

Membrane-protected

A selective membrane between fibre and sample blocks interferences and protects the coating. It benefits analytes that are too involatile for headspace but that would foul a bare fibre. The penalty is kinetic: extraction slows because analytes must diffuse through the membrane, partly offset by thin membranes and elevated temperature.

The conditions that move response more than the fibre does

Salt and pH

For ionisable analytes this is usually the single biggest lever. In Supelco’s data for phenols at 50 ppb on an 85 µm polyacrylate fibre, pentachlorophenol gave about 2,305 area counts with no salt at neutral pH and about 143,905 with salt at pH 2 — roughly a 62-fold gain. 2,4-Dinitrophenol was not detected at all at neutral pH without salt, and only appeared once salt and acid were added.

But the same table contains the warning: 4-nitrophenol gave its highest response with salt at neutral pH (about 11,458) and a lower response with salt at pH 2 (about 6,536). Salting-out and acidification are not universally beneficial. Optimise them against your own analyte list, and if you are running a mixed list, expect to compromise.

Temperature: two effects pulling opposite ways

Raising temperature drives more analyte into the headspace and speeds equilibration — but sorption onto the fibre is exothermic, so the fibre/sample distribution constant falls as the fibre gets hotter. There is therefore an optimum, and it is analyte- and matrix-specific. In the soup optimisation study, extraction was tested at 50, 60, 65, 70 and 80 °C and peaked at 65 °C, with response declining above it precisely because the distribution coefficient falls. Extraction time in that study was optimised across 20–60 minutes and settled at 30 minutes.

Agitation and sample volume

Agitation affects kinetics, not equilibrium, but in a pre-equilibrium method that makes it a quantitative variable. Reported gains range from a modest improvement at 500 rpm to factors of 1.4× (ethyl decanoate) to 3.1× (linalool) at 1,000 rpm. Whatever you choose, it must be identical for standards and samples.

Sample volume behaves counter-intuitively. The equilibrium amount extracted is n = KV_fC₀V_s / (KV_f + V_s). When the sample is large relative to the coating (V_s » KV_f) this collapses to n = KV_fC₀ — the amount extracted becomes independent of sample volume. That is why field and in-vivo SPME sampling works at all, and why worrying about a few millilitres of vial fill is usually misplaced. What you must keep constant is the headspace ratio, because in a sealed vial the phase ratio determines how much analyte is available in the vapour.

Quantitation: what SPME is actually measuring

SPME is a non-exhaustive method. In a matrix with binding sites — plasma proteins, dissolved organic matter, sediment — the fibre samples the free (unbound) fraction, not the total concentration. That is a feature for bioavailability work and a trap for anyone comparing SPME numbers against an exhaustive solvent extraction and expecting agreement.

Non-depletive conditions have a working definition: extraction of less than 5% of the unbound concentration is generally considered negligible depletion. Below that threshold the fibre does not perturb the matrix binding equilibrium, and repeated sampling from the same aliquot becomes possible.

Three calibration strategies are in general use:

Equilibrium calibration with matrix-matched standards. Simplest, but you must actually reach equilibrium and the surrogate matrix must have the same binding behaviour.
Pre-equilibrium (fixed-time) calibration. Extract for a controlled short time on the linear part of the uptake curve. This is where most routine methods live, and it is why an autosampler with reproducible timing is worth more to SPME precision than almost any other purchase.
Kinetic calibration / on-fibre standardisation. Pre-load an isotope-labelled standard onto the coating; during extraction it desorbs into the sample while the analyte is extracted onto the fibre, and the loss percentage of the standard equals the gain percentage of the analyte. This is the method of choice for in-vivo and field sampling where you cannot control agitation or reach equilibrium.

Expect real fibre-to-fibre variation. One high-throughput non-depletive study reported intra-fibre RSDs of 2.7–18.1% and inter-fibre RSDs of 1.2–17.9% across concentration levels. Qualify each new fibre against a check standard before you put it into a validated run, and prefer an internal standard over raw area.

For regulated work, note that ASTM D6520-18 states typical SPME detection limits with GC-FID, GC-ECD or GC-MS range from mg/L down to µg/L, and that the detection limit, linear range and sensitivity for any specific compound depend on the aqueous matrix, the fibre phase, sample temperature, sample volume, sample mixing and the determinative technique. ISO 17943:2016 (Water quality — Determination of volatile organic compounds in water — Method using headspace solid-phase micro-extraction (HS-SPME) followed by gas chromatography-mass spectrometry) is the corresponding international water-quality method. Where a regulator specifies a purge-and-trap method instead, SPME is often still useful as a pre-screen; ASTM D6520-18 notes it can be used to check whether dilution is needed before purge and trap to avoid overloading the trap. Our guide to EPA Methods 8270, 8260 and 624 covers where those regulated VOC and SVOC methods sit.

Desorption, carryover and the inlet

Desorption is where good extractions get thrown away.

Use a narrow-bore liner. A conventional 2 mm ID splitless liner lets the desorbed band expand and broaden. Supelco’s gaseous-VOC comparison at 50 ppb shows visibly sharper early-eluting peaks on a 0.75 mm ID SPME liner. This is the most common single reason a working SPME method gives ugly front-end peaks.
Set desorption temperature by coating, not by habit. Published methods span a wide range: the faecal metabolome study used inlet temperatures from 240 °C for the PEG fibre up to 320 °C for 7 µm PDMS. Exceeding a coating’s rating strips phase and raises the blank; undercutting it causes carryover.
Assume carryover on adsorbent phases until proven otherwise. Carboxen retains high-mass analytes tightly enough that desorption can be incomplete. Run a second, blank desorption after every high-level sample during method development and quantify what comes off.
Condition new fibres and bake between runs as the manufacturer specifies, and carry a fibre blank in every sequence.

Two hardware details that cause avoidable grief: piercing needles come in a standard 24-gauge and a wider-bore 23-gauge for septum-free inlets, so confirm the gauge matches your inlet before ordering; and fibre cores come as fused silica, StableFlex and metal alloy, with fused silica by far the easiest to snap. Vial septa matter too — a septum that cores or bleeds will contaminate a headspace method, and the selection logic in our vial and septa compatibility guide transfers directly. Detection-side choices are covered in our introduction to mass spectrometers.

A short method-development path

List your analytes with molecular weight, boiling point and pK_a. That table decides most of what follows.
Choose sampling mode first: headspace unless the analytes will not volatilise, then direct immersion, then membrane-protected if the matrix will destroy a bare fibre.
Choose a mechanism: adsorbent for trace targets in a narrow concentration band, absorbent for wide or unknown ranges.
Screen two or three coatings across the molecular weight range you actually need — not one. Start from 50/30 µm DVB/Carboxen/PDMS if you have no prior.
Optimise pH and salt before temperature and time. They usually move response further.
Optimise temperature by building a real curve; expect a maximum, not a monotonic rise.
Build an extraction-time profile and decide deliberately whether you are running at equilibrium or at a fixed pre-equilibrium time.
Prove desorption is complete with a blank second desorption, and check linearity in matrix — specifically at the top of your range, where an adsorbent phase will bend first.

Frequently asked questions

Which SPME fibre should I start with if I do not know my analytes?

A 50/30 µm DVB/Carboxen/PDMS fibre. It is the layered phase designed to span a wide molecular weight range, and it won head-to-head comparisons in both the faecal volatilome study and the cooked-soup optimisation cited below. Treat it as a starting point, not an answer — for non-targeted work a panel of fibres will always find more than any one of them.

Headspace or direct immersion?

Headspace whenever the analytes will partition into the vapour phase. It keeps the coating out of contact with proteins, lipids and particulates, which is the main thing that kills fibres. Switch to direct immersion only when the analytes are too involatile for headspace, and then budget for shorter fibre life or use an overcoated fibre.

Why do my peak areas drop when I raise the extraction temperature?

Because sorption onto the coating is exothermic. Heating drives analyte into the headspace, which helps, but simultaneously lowers the fibre/sample distribution constant, which hurts. Above the crossover point the second effect wins. Build a temperature curve rather than assuming hotter is better; one published HS-SPME optimisation found the optimum at 65 °C, with lower response at 70 and 80 °C.

Do I have to let the extraction reach equilibrium?

No. Fixed-time pre-equilibrium sampling is entirely legitimate and is what most routine methods do, because full equilibration can take far longer than the analysis. The condition is that everything affecting kinetics — time, temperature, agitation rate, vial fill, headspace ratio — is held identical between standards and samples. That constraint is why reproducible SPME is much easier on an autosampler than by hand.

What is analyte displacement, and how do I know it is happening?

On adsorbent coatings the number of sorption sites is finite, so when a high-affinity compound is abundant it can push a lower-affinity compound off the surface. The signature is a calibration curve that flattens at the top, or an analyte whose response falls when an unrelated co-extracted compound rises. Absorbent films do not do this, because partitioning into a liquid phase is non-competitive. Shortening the extraction time, diluting the sample, or moving to an absorbent phase all mitigate it.

Why do two fibres of the same part number give different results?

Coating thickness and particle loading vary between fibres, and coatings age. A non-depletive SPME study reported inter-fibre RSDs of 1.2–17.9%, on a par with or worse than intra-fibre variation of 2.7–18.1%. Use an internal standard, qualify each fibre against a check standard before use, and avoid swapping fibres mid-sequence.

Does adding salt always improve extraction?

No. Salting-out helps most polar and ionisable analytes substantially — in Supelco’s phenol data, salt plus acidification raised pentachlorophenol response roughly 62-fold and made 2,4-dinitrophenol detectable at all — but the same data set shows 4-nitrophenol responding better with salt at neutral pH than with salt at pH 2. High salt also accelerates coating degradation on immersed fibres. Optimise it, do not assume it.

Is there a standard method for SPME?

Yes, for water. ASTM D6520-18 is the active standard practice for SPME of water and its headspace for volatile and semi-volatile organics, and ISO 17943:2016 specifies determination of VOCs in water by HS-SPME followed by GC-MS. Both are practices rather than compound-specific methods: they require you to demonstrate that your chosen coating actually extracts your analytes, and to validate the application yourself.

Why are my early-eluting peaks broad even though the extraction looks fine?

Almost always the inlet liner. A standard 2 mm ID splitless liner is too wide for SPME desorption; the desorbed band expands and the front of the chromatogram smears. Fit a narrow-bore SPME liner — Supelco’s comparison used 0.75 mm ID — and keep the split closed for the desorption period.