SPME without an isotope-labelled internal standard is a guess

An SPME fibre can give beautiful, linear calibration curves in clean water and still give the wrong answer for a real sample.

The reason is that SPME is non-exhaustive. The fibre takes up a fraction of the analyte, and that fraction moves with extraction time, temperature, agitation, salt, and anything in the sample that holds the analyte back. Change any of those between the standards and the samples, and the calibration no longer applies.

The fix is old and simple: an internal standard that goes through the same extraction as the analyte. My Ph.D. was on SPME in Janusz Pawliszyn’s lab, and if I had to give one piece of advice to a lab starting SPME, it would be this one.

What EPA Method 8272 does

The clearest regulatory example is US EPA SW-846 Method 8272, for parent and alkyl PAHs in sediment pore water by SPME and GC-MS. It is guidance rather than a mandatory method, but it is a complete, multi-laboratory SPME method, and its design choices are worth copying (EPA 8272, Revision 0, December 2007).

Eight deuterated PAHs are added to the sample before extraction. The method gives the reason: they adsorb to the nonpolar PDMS coating “at equivalent rates” to the target PAHs, which compensates for variations in equilibrium partitioning and kinetics (pp. 3–4). Because the labelled standards track each analyte through the whole extraction, the method treats surrogates as unnecessary (p. 10).

The rest of the method protects that ratio:

• An autosampler is “much preferred”, with lower and more reproducible blanks than manual extraction (p. 3).
• A fixed fibre and fixed conditions: 7 µm PDMS, 30-minute extraction, splitless desorption at 320 °C (pp. 5–7).
• A carryover blank between every new sample. If it exceeds a third of the lower limit of quantitation, analysis stops until the fibre is clean (p. 17).
• Fibre cleaning for one hour at 320 °C under helium at the start of each set and after dirty samples (p. 18).
• Calibration on area ratios, with r² above 0.99 and continuing calibration checks within ±25% for heavy PAHs and ±20% for light ones (pp. 11–12).

One small inconsistency in the method, for anyone reading it closely: it says the calibration curve “must not be forced through the origin” (p. 11), but quantitation uses a mean response factor, which is a line through the origin (pp. 14–15). Decide which you will use and write it down.

Why the ratio works, and when it does not

The internal standard works because, for SPME, what matters is not whether the analyte and standard are similar on the column, but whether they move into the coating the same way. A deuterated or carbon-13 version of the analyte has almost the same distribution constant and diffusion coefficient. Whatever slows or speeds the analyte’s uptake does the same to the standard.

A review of SPME calibration methods lists the internal standard as the method that corrects for sample matrix effects, and isotope-labelled standards as compensating for matrix effects, losses during sample preparation and injection variability. Its downside is cost and limited availability (Merkle et al. 2015, section 3.5.1 and Table 2).

Two conditions have to hold:

1. Add it to the sample, before extraction. A standard added at injection corrects the instrument only. Most SPME variation happens in the vial.
2. Let it equilibrate with the matrix first. In a sample with proteins or other binding components, the analyte is partly bound and only the free fraction extracts. A standard that has not had time to bind like the analyte will not track it. In my 2015 Analytical Chemistry paper I showed that binding can make the equilibrium time shorter or longer depending on the binding strength and the dissociation rate, which is one more reason to let the standard behave like the analyte before the fibre goes in (Alam et al. 2015). I explain those effects in why the matrix changes how fast SPME reaches equilibrium.

When there is no labelled standard

For many analytes there is no affordable isotope-labelled version. The options, roughly in order:

• A close structural analogue with similar polarity and molecular weight, added before extraction. It will not track perfectly, so measure how much variation remains during validation, across real samples, not only in water.
• Matrix-matched calibration with blank matrix, extracted exactly like the samples. It needs a truly blank matrix and a very stable procedure.
• Standard addition, which the same review recommends for heterogeneous samples (section 3.5.1). Slow, but each sample calibrates itself.
• Kinetic calibration, where a standard is loaded onto the fibre before sampling and its loss corrects for the extraction conditions. This is the approach I modelled in my 2016 paper, and it suits on-site and in vivo work where nothing can be added to the sample (Alam & Pawliszyn 2016). I covered it in how SPME is calibrated.

What I would not do is calibrate in clean solvent, extract real samples, and report the result as if the two extractions were the same.

Sources

• US EPA. SW-846 Method 8272: Parent and alkyl polycyclic aromatics in sediment pore water by solid-phase microextraction and GC/MS in selected ion monitoring mode. Revision 0, December 2007.
• Merkle S, Kleeberg KK, Fritsche J. Recent developments and applications of SPME in food and environmental analysis. Chromatography 2015; 2:293–381.
• Alam MN, Ricardez-Sandoval L, Pawliszyn J. Numerical modeling of solid-phase microextraction: binding matrix effect on equilibrium time. Anal Chem 2015; 87:9846–9854.
• Alam MN, Pawliszyn J. Numerical simulation and experimental validation of calibrant-loaded extraction phase standardization approach. Anal Chem 2016; 88:8632–8639.

If you are planning an SPME method and are not sure an internal standard exists for your analytes, tell me the analytes on LinkedIn and I will tell you what I would use.