Automated offline µ-SPE cleanup in GC-based multi-residue analysis: overcoming the challenges of fatty acid containing matrices

Abstract

The aim of this work was to investigate and optimize the use of offline automated µ-SPE cleanup for the analysis of GC-amenable pesticides in fatty acid containing matrices, such as wheat and linseeds. The manual d-SPE and the automated µ-SPE procedures were compared in terms of analytical performance. Furthermore, different cartridge compositions, loading volumes and elution rates were tested and compared in order to find the best conditions for the cleanup of highly complex QuEChERS raw extracts. In µ-SPE, the key aspect was the PSA capacity for the removal of fatty acids and prevention of overloading. The novelty of this research lies in the use of customized µ-SPE cartridges containing higher amounts of PSA, which proved to be the most effective strategy for minimizing matrix effects and enhancing the method robustness. The extracts obtained were cleaner compared to d-SPE, and the addition of analyte protectants (shikimic acid) was required to prevent the adsorption of certain analytes in the GC system. The final µ-SPE workflow provided an efficient and reliable cleanup step for testing highly complex extracts rich in fatty acids and improved GC robustness, reducing the need for GC-MS maintenance.

---

1. Introduction

The analysis of pesticide residues in food plays a crucial role in ensuring food safety. Pesticides are applied at different times, during growth, harvesting, transport and storage, to protect crops and food from pests, weeds, and diseases. An excessive use can lead to harmful residues for public health. Hence, the importance of measuring pesticides at very low levels remains the ultimate goal for monitoring control programs. Gas chromatography (GC) and liquid chromatography (LC), mostly coupled with tandem mass spectrometry (MS/MS), are employed to accurately detect and quantify pesticide residues in food. Food is known to be one of the most complex samples to analyse, especially because of the presence of high concentration matrix constituents (e.g., fatty acids).¹ The compounds naturally present in foods can interfere with the detection and quantification of pesticides in several ways, such as interfering with the signal (signal suppression or enhancement), resulting in the so-called ‘matrix effect’.^1,2 Particularly in GC-based methods, co-extractants can also negatively influence the performance of the instrument. Two main options to reduce the adverse effects of the matrix are dilution and clean-up of the final extract prior to the analysis. Diluting the sample with a clean solvent is often used in LC-based analysis, while clean-up strategies based on matrix–sorbent material interactions are typically carried out prior to GC analysis. The main drawback of dilution is the reduction of the in-vial concentration, which for certain analytes may result in limits of quantification (LOQs) not fit for maximum residue limit (MRL) compliance testing or monitoring for risk assessment. The interaction between the sorbent material and some target pesticides is a concern when clean-up strategies are applied for reducing the adverse effects of the co-extracted matrix (e.g., interactions between planar pesticides and graphitized carbon black (GCB)³). Overloading the SPE sorbent with the matrix may be an issue with certain food commodities (e.g. fatty acids to be absorbed on amino-based sorbents like PSA). For these main reasons, both approaches must be well investigated and optimized before being introduced into the final analytical protocol.

Sorbent materials can be used in dispersive mode (d-SPE) or using conventional SPE cartridges. d-SPE is quick, while SPE cartridges can be more effective but involve a more laborious manual procedure.

The first appearance of µ-SPE, dating back to 2015 and initially referred to as automated cartridge-SPE (c-SPE), for extracts’ cleanup was reported by Morris & Schriner,⁴ who used the Instrument Top Sample Preparation (ITSP) mini-cartridges based on zirconia sorbent for fatty acid and pigment removal from avocado, citrus, and buttercup squash, improving extract cleanliness and enhancing reproducibility and throughput for LC-MS/MS pesticide residue analysis of 263 analytes of interest.

The year after, Lehotay et al. introduced automated mini-SPE cleanup combined with low-pressure GC-MS/MS, enabling high-throughput analysis of pesticides and environmental contaminants in foods with reduced manual labour, improved reproducibility, and minimized matrix effects.⁵ The method was validated for the analysis of 54 pesticides and 43 environmental contaminants in 10 different food matrices.

Sapozhnikova et al. reported for the first time the online combination of mini-SPE cleanup coupled to low-pressure GC-MS/MS for several pesticides and environmental contaminants in cattle, swine and poultry muscle tissues⁶ and, later, in catfish.⁷

Other than pesticides, automated mini-SPE was employed for the cleaning of extracts in order to improve the detection of warfare agents in the environment (water and soil) and polycyclic aromatic hydrocarbons in food oils.^8,9

Several scientific manuscripts related to mini-SPE were published in 2021. Hakme and Poulsen evaluated the use of mini-SPE for the clean-up of the analysis of pesticides in cereals, demonstrating high efficiency in removing matrix components and achieving lower LOQs for several pesticides.¹⁰ Goon et al. developed an automated mini-SPE cleanup method for pesticide residue analysis in various spice matrices, including chili powder, turmeric, black pepper, cumin, coriander, and cardamom.¹¹ Noteworthily, Lehotay et al. combined mini-SPE clean-up with a modification of the QuEChERS sample preparation method, called “QuEChERSER”, for the combined detection of pesticides, veterinary drugs, and environmental contaminants in beef, catfish, and animal-derived matrices.^12–15 In the same period, QuEChERSER was again coupled to mini-SPE for the analysis of pesticide residue in hemp and hemp products.¹⁶

A new design of cartridges, defined as septumless μ-SPE mini-cartridges, was described for the first time in 2022 by Michlig & Lehotay.¹⁷ This type of cartridge is characterized by the absence of a septum reducing the chance of leakage and syringe needle break, although more susceptible to moisture uptake by MgSO₄. A different sealing mechanism allows the usage of higher flow rates (>10 μL s⁻¹ compared to the ≤2 μL s⁻¹ for the ITSP cartridges). In addition, the septumless μ-SPE mini-cartridges offer the possibility of a larger sorbent bed, whereas ITSP cartridges are limited to 45 mg. The authors demonstrated that the septumless μ-SPE mini-cartridges enabled a faster, reliable, and highly reproducible automated cleanup, achieving excellent recoveries and low relative standard deviations (RSDs) for over 250 analytes across diverse food matrices, making them a robust solution for high-throughput QuEChERSER analyses. These new μ-SPE cartridges have been used in different applications in the recent past as a clean-up strategy prior to the analysis of pesticide residues, natural toxins and contaminants.^18–22 Recently, a different use of μ-SPE cartridges in combination with a centrifuge, defined as centrifugal μ-SPE, was reported by Michlig and Lehotay, allowing this approach to be used without requiring the complete PAL system although with some limitations.²³

Compared to d-SPE, the volume needed for the μ-SPE cleanup is smaller (≤500 µL), which makes it ideal for miniaturization of QuEChERS sample preparation. Several scientific publications have proven that scaling down is possible without compromising the representativeness of the analytical portion extracted, even without the use of cryo-milling.^24–26

The combination of mini-QuEChERS and µ-SPE can indeed prove to be a highly advantageous approach in terms of environmental impact, cost reduction, and reduced analyst exposure to toxic substances.

The aim of this work is to investigate and optimize the use of offline automated µ-SPE cleanup for the analysis of GC-amenable pesticides in fatty acid containing matrices, such as wheat and linseeds. According to the SANTE guidelines,²⁷ wheat and linseeds can be considered representative commodities of Group 5 (“high starch and/or protein content and low water and fat content”) and Group 3 (“high oil/fat content and very low water content”), respectively. The manual d-SPE and the automated µ-SPE procedure were compared in terms of analytical performance. Furthermore, different cartridge compositions, sorbent amounts, loading volumes and elution rates were tested and compared in order to find the best conditions for the cleanup of highly complex QuEChERS raw extracts.

2. Experimental section

2.1 Chemicals and standards

Pesticide standards were purchased from LGC Standards (Wesel, Germany) and Sigma-Aldrich Chemie BV (Zwijndrecht, The Netherlands). The extraction solvent ACN (LC-MS grade) was purchased from Biosolve (Valkenswaard, the Netherlands). Acetic acid, ammonium formate, magnesium sulfate (MgSO₄), sodium acetate anhydrous (NaOAc), and shikimic acid were obtained from Sigma-Aldrich Chemie BV. (Zwijndrecht, The Netherlands). Water was purified using a Milli-Q system (Millipore, Burlington, MA, USA). An acetate mix conforming to AOAC 2007.01, consisting of 4 g of MgSO₄ and 1 g of NaOAc, was obtained from HPC Standards GmbH (Cunnersdorf, Germany). The d-SPE tubes (containing 150 mg of MgSO₄, 25 mg of C18 and 25 mg of PSA) and Bondesil PSA 40 μm were obtained from Agilent (Santa Clara, CA, USA). One type of commercialized µ-SPE cartridge, namely GCQuE1-45, was purchased from CTC analytics (Zwingen, Switzerland). Three different types of custom-made µ-SPE cartridges were kindly provided by CTC analytics and contained higher amounts of PSA. The composition of the different µ-SPE cartridges is described in Table S1.

2.2 Sample

Wheat flour and linseed samples were purchased in a local organic shop. Wheat flour was not subjected to any pre-treatment. Linseed was homogenized by a conventional milling procedure at ambient temperature. Linseed was stored at −20 °C, whereas wheat was kept at ambient temperature until use.

2.3 Sample preparation

Sample preparation was carried out using the acetate-buffered QuEChERS protocol.²⁸ Briefly, 2.50 ± 0.05 g of each sample was weighed in a 50 mL centrifuge tube, followed by the addition of 7.5 mL of distilled water, and 30 s of hand shaking. Next, 10 mL of ACN (1% AA) was pipetted into the tubes followed by agitation head-over-head for 30 min. A mixture of salt containing magnesium sulphate (4 g) and sodium acetate (1 g) was added, and the tube was shaken for 5 min to induce salting-out and phase separation. The samples were centrifuged at 3500 rpm for 5 min at 10 °C. After the extraction, two different clean-ups were performed as follows:

d-SPE: for this, Agilent d-SPE tubes were used containing 150 mg MgSO₄, 25 mg C18 and 25 mg PSA, to which an additional 225 mg of PSA was added. One mL of raw extract was added to the tube together with 25 µL of PCB-198 (1 µg mL⁻¹), followed by thoroughly shaking and centrifugation for 5 min at 13 000 rpm. Subsequently, 10 µL of ACN (1% AA) or the standard mixture (at different concentrations) were added to 100 µL of the supernatant for the real samples or for the matrix matched standards, respectively.

µ-SPE: for the final µ-SPE clean-up, 300 µL of extract (from a vial containing 1 mL raw extract + 25 µL of 1 µg mL⁻¹ of PCB-198) was cleaned at a speed of 5 µL s⁻¹. Afterwards, 90 µL of cleaned extract were transferred to the final vial, to which 5 µL of ACN (1% AA) or the standard mixture (at different concentrations) were added for the real samples or for the matrix matched standards, respectively. Finally, 5 µL of shikimic acid at a concentration of 4 mg mL⁻¹ (1 µg on column) were added to the vial followed by a final mixing step. All the steps are illustrated in Fig. 1. The final extract (0.25 g mL⁻¹) is then ready for analysis.

Fig. 1 Final µ-SPE workflow optimized for the clean-up of QuEChERS extracts.

2.4 Instrumentation

2.4.1 GC-MS/MS. All the GC-MS/MS experiments were carried out using a GC 7890 B and a G7013 B triple quadrupole MS (Agilent Technologies, Santa Clara, CA, USA). Data were acquired and processed by the use of the MassHunter software (v. B.09.00). The GC was equipped with the 7693 autosampler and a Multi-Mode Injector (MMI). A ultra-inert splitless liner (Agilent 5190–4006) was installed in the injector port. Separation was performed on an Rtx-CLPesticides 30 m × 0.25 mm ID × 0.25 µm d_f (Restek nr. 11123) column. Helium was used as the carrier gas with a flow rate of 1 mL min⁻¹. The injection volume was 5 µL and injection was performed in solvent vent mode at 75 mL min⁻¹ for 0.065 min followed by split at 50 mL min⁻¹ after 4 min. The temperature of the injector was as follows: 60 °C (0.3 min) with 900 °C min⁻¹ to 285 °C (1.5 min) and with 900 °C min⁻¹ to 350 °C (15 min). Temperature oven program: 60 °C (2 min) to 150 °C at 20 °C min⁻¹, to 280 °C at 10 °C min⁻¹, and to 320 °C (5 min) at 25 °C min⁻¹, total run time 26.1 min.

Mass spectrometry conditions: the temperature of the interface was 300 °C; the ion source temperature was 250 °C, with electron ionization at 70 eV. Nitrogen was used as the collision gas with a collision flow of 1.5 mL min⁻¹ and a quadrupole temperature of 150 °C. Two multi-reaction monitoring transitions were measured for each pesticide. A list of all the transitions can be found in Table S2.

2.4.2 GC-Q-orbitrap. All the GC-Q-Orbitrap experiments were carried out on a Q-Exactive system (Thermo Fisher Scientific, Bremen, Germany) consisting of a TRACE 1310 GC and a hybrid Q-Orbitrap mass spectrometer. Data were acquired and processed using TraceFinder software (v. 4.1). The GC was equipped with a TriPlus RSH™ autosampler and a programmed temperature vaporizer injector (PTV). An ultra inert liner (Restek RT-21117-216) was installed in the injector port. Separation was performed on an Rxi-5-SILMS 30 m × 0.25 mm ID × 0.25 µm d_f (Restek nr. 13623) column. Helium was used as the carrier gas with a flow of 1.2 mL min⁻¹. The injection volume was 3 µL and it was performed in solvent vent mode at 50 mL min⁻¹ for 0.10 min followed by split at 50 mL min⁻¹ after 1.2 min. The temperature of the injector was as follows: 70 °C (0.1 min) with 300 °C min⁻¹ to 300 °C (16.5 min). Temperature program: 80 °C (1 min) to 185 °C at 30 °C min⁻¹, to 280 °C (1 min) at 10 °C min⁻¹, and to 320 °C (5.86 min) at 35 °C min⁻¹, total run time 21.5 min.

Mass spectrometry conditions: the temperature of the interface was 300 °C; the ion source temperature was 290 °C, with electron ionization at 70 eV.

Full Scan (FS) acquisition was performed in profile mode using an m/z range of 75–500. The resolving power was set at 60 000 full width half maximum (FWHM) at m/z 200. The automatic gain control (AGC) target was set at 3 × 10⁶ ions, with the maximum ion injection time set to auto. Tuning and mass calibration were carried out before each sequence. Internal mass calibration was automatically performed by the instrument using three background ions from the column bleed as the lock mass (C₅H₁₅O₃Si₃⁺, 207.03235; C₇H₂₁O₄Si₄⁺, 281.05114; and C₉H₂₇O₅Si₅⁺, 355.06994). TraceFinder 4.1 was used to process the data and the exact masses of two ions (quantifier and qualifier-1) for each compound were selected. The list of analytes and the exact masses of the ions are provided in Table S3.

2.5 Methods

2.5.1 Initial clean-up comparison: d-SPE vs. GCQuE1-45 µ-SPE cartridges. In order to compare the robustness of the GC-MS/MS system, the injection of the same QuEChERS wheat extract, spiked at 12.5 µg kg⁻¹ (corresponding to 50 µg kg⁻¹ in the matrix) and cleaned via d-SPE and µ-SPE, was evaluated. Briefly, for d-SPE, multiple aliquots of 1 mL of wheat raw extract were added to commercial d-SPE Eppendorf tubes containing 150 mg of MgSO₄, 25 mg of C18, and 25 mg of PSA, to which an extra 225 mg of PSA was added. A volume of 25 µL of PCB-198 (1 µg mL⁻¹) was added to the tube as the injection internal standard, followed by thoroughly shaking and centrifuging for 5 min at 13 000 rpm. Subsequently, 990 µL was collected from the d-SPE tube and 10 µL of ACN (1% AA) was added to the final vial. The cleaned extracts were merged together and then split in 23 vials and injected subsequently into the GC-MS/MS system.

The µ-SPE clean-up was performed using the GCQuE1-45 cartridge containing 20 mg of MgSO₄, 12 mg of C18, 12 mg of PSA, and 1 mg of GCB. Briefly, 1 mL of extract + 25 µL of 1 µg mL⁻¹ of PCB-198 was added individually to 20 vials and, from each vial 300 µL of extract was cleaned at a speed of 5 µL s⁻¹. Afterwards, 95 µL of the cleaned extract were transferred to the final vial together with 5 µL of ACN (1% AA) for injection into the GC-MS/MS system.

Samples were injected on two different days on the GC-MS/MS system, each time using a new liner and a freshly trimmed front section of the column. Before injecting the spiked samples, the liner was primed with 5 consecutive injections of raw wheat extract.

Besides the robustness comparison described above, the removal of matrix co-extractants was also gravimetrically assessed. For this, after QuEChERS extraction, 2 mL of raw uncleaned extract, 2 mL of d-SPE cleaned extract, and 2 mL of µ-SPE cleaned extract were evaporated using a TurboVap. The difference in vial weight before and after evaporation was calculated in order to obtain the solid residue expressed as mg mL⁻¹. For the d-SPE experiment, the sorbent composition used to clean 9 mL of extract consisted of 600 mg MgSO₄, 360 mg C18, 360 mg PSA, and 30 mg GCB. This composition was selected to maintain the same sorbent-to-extract ratio as in the µ-SPE procedure, enabling a direct comparison of the clean-up efficiency between the two strategies.

2.5.2 Custom-made and commercial µ-SPE cartridges’ comparison. Three different types of custom-made cartridges together with one of the commercially available versions were compared. The composition of all cartridges is described in Table S1. Briefly, the commercially available cartridges contain 12 mg of PSA, while the custom-made ones contain PSA in increasing amounts, 24, 36 and 48 mg.

In order to compare the ability of the different cartridges to remove co-extractants and especially free fatty acids, wheat and linseeds were extracted and then cleaned using the four different cartridges. As for the previous experiment, 1 mL of raw extracts spiked with 12.5 µg kg⁻¹ (corresponding to 50 µg kg⁻¹ in the matrix) of the pesticide mix + 25 µL of PCB-198 (1 µg mL⁻¹) was added to the vials and 300 µL of extracts was cleaned at a speed of 5 µL s⁻¹. The analysis was carried out using a GC-Q-Orbitrap. The total ion current (TIC) was then compared, with particular attention paid to the percentage of fatty acids removed. The cleanup recovery of the pesticides was used to compare the influence of the different cleanup cartridges. For this experiment, shikimic acid was not employed as an analyte protectant. Additional experiments involving dilution with a tomato extract and the use of shikimic acid to improve the signal of selected pyrethroids are included in the corresponding section of the Results and discussion (Section 3.2. Custom-made and commercial µ-SPE cartridges’ comparison).

2.5.3 d-SPE vs. CM-02 GCQuE µ-SPE: clean-up comparison. As for the experiment described in Section 2.5.1, to compare the robustness of the GC-MS/MS system, the injection of the same QuEChERS wheat extract, spiked at 12.5 µg kg⁻¹ (corresponding to 50 µg kg⁻¹ in the matrix) and cleaned via d-SPE and µ-SPE, was evaluated. The d-SPE clean-up is already described in Section 2.5.1. Regarding the µ-SPE clean-up, it was performed using the CM-02 GCQuE µ-SPE, consisting of 20 mg of MgSO₄, 12 mg of C18 and 36 mg of PSA. Briefly, 1 mL of extract + 25 µL of 1 µg mL⁻¹ of PCB-198 was added individually to 20 vials and, from each vial, 300 µL of extract was cleaned at a speed of 5 µL s⁻¹. Afterwards, 90 µL of cleaned extract was added to the final vial and 5 µL of ACN (1% AA) and 5 µL of shikimic acid at a concentration of 4 mg mL⁻¹ (1 µg on the column) were added to the vial and ready to be injected into the GC-MS/MS system.

Samples were injected on two different days on a GC-MS/MS system with a new liner and a freshly trimmed front section of the column. Before injecting the spiked samples, the liner was primed with 5 consecutive injections of raw wheat extract.

3. Results and discussion

3.1 Initial clean-up comparison: d-SPE vs. GCQuE1-45 µ-SPE cartridges

As a first step, the in-house d-SPE method was compared with the method using commercially available GCQuE1-45 cartridges (see Section 2.5.1 for the method details). The aim was to assess whether, despite the relatively small amount of PSA, the conventional µ-SPE cartridges could already be sufficient to obtain an extract clean enough to not compromise the GC-MS/MS performance in a typical overnight sequence. To this extent, the robustness of the GC system was first evaluated by injecting the same wheat extract cleaned by d-SPE and µ-SPE on two different days respectively, using a new liner and a freshly trimmed front section of the column each time. In both experiments, the standard deviation was calculated for the response factor of each pesticide based on 23 injections.

As illustrated in Fig. 2, the results obtained from the robustness test using the d-SPE method (blue bars) can be considered satisfactory (with the exception of dichlorvos). Conversely, µ-SPE using the commercially available GCQuE1-45 cartridges showed poorer results (orange bars), with RSDs exceeding 20% for 12 pesticides, most notably for p,p′-DDE, α-endosulfan, p,p′-DDT, and methoxychlor. To demonstrate that these results are related to a deterioration in instrumental performance, the same extract was diluted sixfold after µ-SPE and re-injected on a different day. The results (yellow bars) show that, after dilution, the majority of pesticides exhibit RSD values below 20%, demonstrating that the system remained stable throughout the entire 23 injection batch.

Fig. 2 Comparison of the standard deviation (RSD%) for the 69 pesticides after 23 consecutive injections of samples cleaned via d-SPE (blue), µ-SPE (orange), and µ-SPE followed by 6 times dilution (yellow).

As further evidence of the inability of the conventional cartridges to efficiently remove matrix components, the solid residue was determined for the raw extract and for the extracts obtained after clean-up by d-SPE and µ-SPE. Moreover, different clean-up flow rates were also evaluated in order to assess whether this parameter could influence the effectiveness of the µ-SPE clean-up and an extra d-SPE tube with the same composition (in terms of sorbent-to-extract ratio) as µ-SPE was included. The experimental details are described in Section 2.5.1.

Table 1 reports the results related to the solid residual obtained after the different experiments. While the in-house d-SPE method efficiently removed the matrix (93%), the results were clearly worse for µ-SPE (55%), thus confirming that a larger amount of matrix-derived material is introduced into the instrument, accelerating instrumental deterioration. Nevertheless, as shown in Table 1, when the same sorbent-to-extract ratio is used (d-SPE with the same ratio as µ-SPE), µ-SPE exhibits a slightly higher removal efficiency compared to d-SPE. The different clean-up flow rates do not play a crucial role, giving similar results in terms of matrix removal.

Table 1 Wheat solid residue and % of matrix component removal

	Residue, mg ml^−1	% of removal
RAW	2.8
d-SPE	0.2	93
d-SPE with the same ratio as µ-SPE	1.7	39
µ-SPE (5 µL s^−1)	1.3	53
µ-SPE (3 µL s^−1)	1.6	44
µ-SPE (1 µL s^−1)	1.3	54

---

The extracts were also injected on the GC-Q-Orbitrap and the TIC was compared in order to evaluate the removal of the GC-amenable co-extractants for the different cleanup options and without any cleanup. As illustrated in Fig. S1, while d-SPE (red TIC) allowed efficient removal of fatty acids (between 9 and 11 min), these were clearly visible in the extracts cleaned by µ-SPE, regardless of the clean-up flow rate applied.

In conclusion, it became clear that the commercially available cartridges did not sufficiently remove co-extractants and, consequently, did not enable the GC-MS/MS system to maintain its performance over the duration of a conventional analytical batch. The presence of the fatty acid peaks indicated that these were insufficiently removed, attributed to overloading of the µ-SPE/insufficient amount of PSA in the cartridge. To address this issue, potential custom-made cartridges characterized by a higher amount of PSA were investigated in order to enhance fatty acid removal. Alternative approaches to address the issue related to the matrix removal, based on the use of commercial µ-SPE cartridges, include dilution of the crude extract prior to clean-up or the cleaning of volumes lower than 300 µL. However, extract dilution may negatively impact the method LOQs. Cleaning a volume substantially lower than 300 µL caused issues with a low eluent fraction/incomplete elution due to the dead volume of the sorbent material (approximately 50 µL).

3.2 Custom-made and commercial µ-SPE cartridges’ comparison

In order to improve fatty acid removal, four different types of cartridges (one commercially available and three custom-made) were used to perform the clean-up of wheat and linseed QuEChERS raw extracts. All these cartridges, as described in detail in Section 2.5.2, are characterized by increasing amounts of PSA.

The PSA removes fatty acids by binding their carboxylic acid groups via acid–base interactions and hydrogen bonding, retaining them on the sorbent while the target pesticides remain in solution.²⁹

Fig. 3 and Fig. S2 show a comparison of the normalized TICs for all the different types of cartridges used for the cleanup of wheat and linseed extracts, respectively. As is clear from both figures, increasing the amount of PSA drastically improves the removal of fatty acids, which elute approximately in the middle of the chromatogram (fatty acid zone in Fig. 3). In contrast, for the commercially available cartridges, it is evident that overloading occurs.

Fig. 3 Comparison of the normalized TICs for all the different types of µ-SPE cartridges used for the cleanup of wheat extract.

In addition to evaluating the matrix removal capability, pesticide recovery after cleanup was also assessed. To this extent, the pesticide mixture was added to the extract before and after the cleanup step, in order to compare the response factors obtained for each individual pesticide.

An overview of the cleanup recoveries obtained for the two different matrices used in this study is reported in Table 2, while the cleanup recoveries obtained for the individual pesticides are reported in Table S4. Briefly, the majority of pesticides exhibited recoveries higher than 70%. This demonstrates that none of the different cartridge types employed caused significant retention, hence losses of any analyte of interest.

Table 2 The number of pesticides with clean-up recoveries of >70%, <70%, and not detected (ND) for the selected matrices using the different µ-SPE cartridges

Cartridge type	Wheat			Linseed
	Rec >70	Rec <70	ND	Rec >70	Rec <70	ND
CM-01 GCQuE	62	0	0	59	2	1
CM-02 GCQuE	55	2	5	47	6	9
CM-03 GCQuE	53	3	6	51	0	11
GCQuE1-45	61	0	1	60	2	0

---

Notably, for the custom 02 and custom 03 cartridges (containing the highest amounts of PSA), no signal was obtained for some pesticides. This included pyrethroids such as cypermethrin. It is well known that cypermethrin and many other pyrethroids tend to be adsorbed by active sites in the liner. The presence of matrix components in the extracts reduces this effect by masking the active sites of the glass and provides better recoveries of these pesticides.³⁰ Given the excellent clean-up performance of cartridges 02 and 03 and considering that the recoveries of these compounds were optimal for all the other cartridges, it was hypothesized that the high clean-up efficiency caused these compounds to be adsorbed by the glass liner. In order to confirm this hypothesis, the extract after clean-up was injected as such and after a 1 : 1 dilution with a tomato extract.

As shown in Fig. S3, depending on whether the linseed sample was diluted with a tomato extract (green box) or injected undiluted (red box), the peaks corresponding to cypermethrin were present or absent, respectively. This behaviour was observed both when fortification was performed before the clean-up (blue rectangle) and when it was carried out after the clean-up (yellow rectangle) in order to exclude any losses of cypermethrin by the cartridges. In addition, it can be noted that the peak areas (and the response factors) corresponding to the extract diluted 1 : 1 with the tomato extract were consistent, regardless of whether fortification occurred before or after the clean-up, proving that clean-up had not a negative impact in the clean-up recovery of this set of four compounds.

So, on one hand, clean extracts are beneficial to avoid issues related to instrumental stability and maintenance. On the other hand, (some) the matrix is needed to shield active sites to ensure proper response and peak shape. In order to obtain both advantages, as demonstrated, it is possible to dilute the extract after clean-up with a not complex matrix extract (such as a tomato) or to rely on the use of compounds able to shield the active sites in the system. Analytical protectants (APs) are compounds intentionally added to the vial to prevent analyte loss and signal instability caused by interactions with active sites in the inlet, liner, column, or detector.³¹ As demonstrated by Rodríguez-Ramos et al., 1 µg of shikimic acid on the column enabled the achievement of optimal results for multi-pesticide methods and organic contaminants.³²

Although the use of a tomato extract can be considered sufficient to prevent liner adsorption of susceptible analytes, the use of APs might be preferable. This is because their use can be controlled and they provide consistent and reproducible results. In contrast, the use of a simple matrix extract requires the availability of a blank and may vary depending on the vegetable variety and the time of year. For this reason, the use of shikimic acid at a concentration of 1 µg on-column was tested in order to prevent the adsorption/degradation issues with more polar and labile pesticides in the GC inlet and column. To demonstrate the effectiveness of shikimic acid, Fig. S4 shows the four cypermethrin peaks obtained from a wheat sample spiked at 12.5 ng mL⁻¹ and injected during a batch of 20 analyses. As can be observed, the peak areas remained stable throughout the sequence, with an RSD of 5.9% for cypermethrin I. Finally, the custom 02 cartridge was considered as the best candidate to be used in all subsequent experiments, as it effectively removed fatty acids, and no improvement was observed for custom 03 with an even higher amount of PSA.

3.3 d-SPE vs. CM-02 GCQuE µ-SPE: clean-up comparison

In order to confirm that these cartridges allow optimal matrix removal in terms of instrumental performance, the same experiment described in Section 3.1 was repeated. As illustrated in Fig. 4, the grey lines show the RSD% results obtained when CM-02 GCQuE µ-SPE cartridges were used for extract clean-up, in comparison with the blue lines corresponding to the in-house d-SPE method.

Fig. 4 Comparison of the standard deviation (RSD%) for the 69 pesticides after 23 consecutive injections of samples cleaned via d-SPE (blue) and µ-SPE with extra PSA (grey).

As is evident, particularly when compared with the commercial µ-SPE cartridges described in Fig. 1, the RSD% values for the majority of pesticides, in a sequence of 23 injections, are now optimal. In comparison with the d-SPE method, most pesticides also show an improvement in RSD%. This demonstrates that these cartridges provide improved matrix removal, resulting in more robust and repeatable instrumental performance.

4. Conclusions

In this study, a comparison between an in-house sample clean-up method based on d-SPE and an automated µ-SPE approach was performed.

The results clearly showed that the presence of additional PSA enabled efficient removal of fatty acids, thereby drastically improving the system stability. Moreover, the use of shikimic acid, as an analyte protectant, proved to be an effective solution to obtain consistent response of pesticides prone to adsorption in the inlet in the case of well cleaned extracts. The clean-up recoveries of the targeted pesticides were evaluated, with the majority exceeding 70%, thereby indicating that no significant retention occurred. Overall, the µ-SPE workflow herein reported represents a promising cleanup approach for highly complex extracts rich in fatty acids, showing the potential to enhance GC robustness and contribute to reduced GC-MS maintenance needs.

Author contributions

Ivan Aloisi: conceptualization, data curation, formal analysis, investigation, methodology, software, validation, visualization, writing – original draft, and writing – review and editing. Lisa Elsinga: formal analysis, methodology, validation, and writing – review and editing. Michel Willemsen: formal analysis, methodology, validation, and writing – review and editing. Hans Mol: funding acquisition, project administration, conceptualization, resources, and writing – review and editing.

Conflicts of interest

The authors declare that they have no known competing financial interest or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

All data are reported in the article and provided in the supplementary information (SI). Supplementary information is available. See DOI: https://doi.org/10.1039/d6an00462h.

References

1. C. F. Poole, Matrix-induced response enhancement in pesticide residue analysis by gas chromatography, 2007, preprint,  DOI:10.1016/j.chroma.2007.01.018.
2. J. Hajšlová and J. Zrostlíková, Matrix effects in (ultra)trace analysis of pesticide residues in food and biotic matrices, Elsevier, 2003, preprint,  DOI:10.1016/S0021-9673(03)00539-9.
3. M.-C. Henniońlaboratoire, in Graphitized carbons for solid-phase extraction, 2000, vol. 885.
4. B. D. Morris and R. B. Schriner, J. Agric. Food Chem., 2015, 63, 5107–5119.
5. S. J. Lehotay, L. Han and Y. Sapozhnikova, Chromatographia, 2016, 79, 1113–1130.
6. Y. Sapozhnikova, J. Chromatogr. A, 2018, 1572, 203–211.
7. L. Han and Y. Sapozhnikova, Food Chem., 2020, 319, 126592.
8. M. A. Althoff, A. Bertsch and M. Metzulat, Separations, 2019, 6, 49.
9. P. Eyring, T. Preiswerk, H. L. Frandsen and L. Duedahl-Olesen, J. Sep. Sci., 2021, 44, 600–608.
10. E. Hakme and M. E. Poulsen, J. Chromatogr. A, 2021, 1652, 462384.
11. A. Goon, R. Shinde, B. Ghosh and K. Banerjee, Application of Automated Mini–Solid-Phase Extraction Cleanup for the Analysis of Pesticides in Complex Spice Matrixes by GC-MS/MS, Oxford University Press, 2021, preprint,  DOI:10.5740/jaoacint.19-0202.
12. S. H. Monteiro, S. J. Lehotay, Y. Sapozhnikova, E. Ninga and A. R. Lightfield, J. Agric. Food Chem., 2021, 69, 1159–1168.
13. E. Ninga, Y. Sapozhnikova, S. J. Lehotay, A. R. Lightfield and S. H. Monteiro, J. Agric. Food Chem., 2021, 69, 1169–1174.
14. E. Ninga, S. J. Lehotay, Y. Sapozhnikova, A. R. Lightfield, G. D. Strahan and S. H. Monteiro, Anal. Methods, 2022, 14, 2761–2770.
15. S. H. Monteiro, S. J. Lehotay, Y. Sapozhnikova, E. Ninga, G. C. R. Moura Andrade and A. R. Lightfield, Food Addit. Contam., Part A: Chem., Anal., Control, Exposure Risk Assess., 2022, 39, 699–709.
16. N. Michlig, S. J. Lehotay, A. R. Lightfield, H. Beldoménico and M. R. Repetti, J. Chromatogr. A, 2021, 1645, 462097.
17. N. Michlig and S. J. Lehotay, J. Chromatogr. A, 2022, 1685, 463596.
18. E. Ninga, E. Hakme and M. E. Poulsen, Anal. Methods, 2024, 16, 3784–3797.
19. N. Michlig and S. J. Lehotay, Comparison of QuEChERS and QuEChERSER using robotic or centrifugal mini-cartridge solid-phase extraction cleanup in fast GC-MS/MS analysis of pesticides and environmental contaminants in nonfatty and fatty food commodities, 2024.
20. M. A. Carrera, J. A. Martinez Martinez, M. D. Hernando and A. R. Fernández-Alba, Heliyon, 2024, 10, e33512.
21. F. Jesús, A. R. García, T. Stecconi, V. Cutillas and A. Rodríguez Fernández-Alba, Anal. Bioanal. Chem., 2024, 416, 675–688.
22. A. Schürmann, C. Crüzer, V. Duss, R. Kämpf, T. Preiswerk and H. J. Huebschmann, Anal. Bioanal. Chem., 2024, 416, 689–700.
23. N. Michlig and S. J. Lehotay, J. Chromatogr. A, 2025, 1749, 465882.
24. Y. Yamasaki, I. Aloisi and H. Mol, Food Addit. Contam., Part A, 2025, 42, 902–913.
25. A. Ferracane, M. Zoccali, F. Cacciola, T. M. G. Salerno, P. Q. Tranchida and L. Mondello, J. Chromatogr. A, 2021, 1645, 462126.
26. N. Casado, R. Perestrelo, C. L. Silva, I. Sierra and J. S. Câmara, Microchem. J., 2018, 139, 110–118.
27. T. Pihlström, A. R. Fernández-Alba, C. Ferrer Amate, M. E. Poulsen, R. Lippold, L. C. Cabrera, P. Pelosi, A. Valverde, H. Unterluggauer, H. Mol, M. Jezussek, R. Štěpán and M. Lambert, Analytical Quality Control and Method Validation Procedures for Pesticide Residues Analysis in Food And Feed Sante 11312/2021 v2026, 2026.
28. AOAC Official Method 2007.01 Pesticide Residues in Foods by Acetonitrile Extraction and Partitioning with Magnesium Sulfate Gas Chromatography/Mass Spectrometry and Liquid Chromatography/Tandem Mass Spectrometry First Action, 2007.
29. O. Shimelis, Y. Yang, K. Stenerson, T. Kaneko and M. Ye, J. Chromatogr. A, 2007, 1165, 18–25.
30. M. Anastassiades, K. Maštovská and S. J. Lehotay, J. Chromatogr. A, 2003, 1015, 163–184.
31. K. Maštovská, S. J. Lehotay and M. Anastassiades, Anal. Chem., 2005, 77, 8129–8137.
32. R. Rodríguez-Ramos, S. J. Lehotay, N. Michlig, B. Socas-Rodríguez and M. Á. Rodríguez-Delgado, J. Chromatogr. A, 2020, 1632, 461596.

---