A Green and Sensitive Electromembrane Extraction (EME) Method for the Determination of Dialkylphosphates in Maternal Urine and Amniotic Fluid

A green and sensitive electromembrane extraction (EME) method for the determination of dialkylphosphates in maternal urine and amniotic fluid Juan Manuel ´ Avila-García a , Rut Fern´ andez-Torres a , Rocío S´ anchez-Ruíz b , Isabel María Moreno b , Noemí Aranda-Merino a,* , María Ramos-Pay´ an a,* a Department of Analytical Chemistry, Faculty of Chemistry, Universidad de Sevilla, c/Prof. García Gonz´ alez s/n, 41012 Sevilla, Spain b Area of Toxicology, Department of Nutrition and Bromatology, Toxicology and Legal Medicine, Faculty of Pharmacy, Universidad de Sevilla, c/Prof. García Gonz´ alez s/ n, 41012 Sevilla, Spain ARTICLE INFO Keywords: Electromembrane microextraction Dialkylphosphates Amniotic fluid Maternal urine Mass spectrometry ABSTRACT Organophosphates compounds are extensively used pesticides, whose metabolites, dialkyl phosphates, are considered as bioindicators of human exposure. Recently, research has indicated that they can affect thyroid endocrine function. Pregnant women are a particularly vulnerable cohort to these effects due to concerns about prenatal exposure to environmental pollutants. For the first time an electromembrane microextraction method is proposed for the simultaneous determination of seven dialkylphosphate metabolites in maternal urine and amniotic fluid, which is compatible with mass spectrometry detection and achieves highly sensitive levels of quantitation. Dialkylphosphates were extracted from a 10 mL donor solution at pH 7 into a 50 µL acceptor solution at pH 12 using 1-octanol as the supported liquid membrane and applying 60 V for 10 min at 600 rpm. The obtained extracts were analysed by ion pair liquid chromatography coupled to a triple quadrupole mass spectrometer, using an Acquity UPLC® BEH C18 column (100 mm ×2.1 mm i.d., 1.7 µm particle size) at 30 ◦C. Under optimal extraction conditions, the preconcentration factors were achieved were up to 178-fold, allowing for high sensitivity levels, in the order of ng mL −1 . The limits of quantitation were within 0.075 ng mL −1 (diethyl thiophosphate) and 1.67 ng mL −1 (dimethyl phosphate) in maternal urine samples and within 0.015 ng mL −1 (diethyl thiophosphate) and 0.5 ng mL −1 (dimethyl dithiophosphate) in amniotic fluid samples. The comparative study revealed a significant improvement in terms of analysis time, sensitivity, greenness, and practicality. Finally, the applicability of the method for the determination of selected dialkylphosphates was demonstrated in paired maternal urine and amniotic fluid samples. These results suggest potential placental transfer to the offspring. 1. Introduction Organophosphorus (OPs) insecticides have been the predominant class of insecticides globally for an extended period. These chemicals are commonly used in commercial agriculture to control pests on vegetable and fruit crops, as well as in home gardens for flea control on pets and in some no-pest strips. However, due to their potent anti-cholinergic properties, continuous exposure to these compounds can be detrimental to human health, resulting in acute neurotoxicity. The most common route of exposure for humans to these OPs is dermal adsorption, which can occur through occupational exposure, inhalation, or even ingestion of food contaminated or treated with such compounds [1–3]. OPs are commonly metabolised in the liver, resulting in the production of dialkylphosphates (DAPs), which are non-specific hydrolysis products. Most of DAPs metabolites are excreted in urine, rendering them suitable for use as biomarkers to assess exposure to organophosphates. The most frequently analysed non-specific hydrolysis products are dimethyl phosphate (DMP), diethylphosphate (DEP), dimethyl thiophosphate (DMTP), diethyl thiophosphate (DETP), and dimethyl and diethyl ditiophosphates (DMDTP, DEDTP). Furthermore, dibutyl phosphate (DBP) was also analysed in this study. The presence of these metabolites indicates exposure to specific OPs and their breakdown products [4]. The analysis of organophosphorus metabolite levels in urine can provide valuable insights into the extent of exposure to organophosphates. It is important to note that the specific metabolites identified may vary depending on the organophosphate compound and * Corresponding authors. E-mail addresses: [email protected] (N. Aranda-Merino), [email protected] (M. Ramos-Pay´ an). Contents lists available at ScienceDirect Microchemical Journal journal homepage: www.elsevier.com/locate/microc https://doi.org/10.1016/j.microc.2024.111715 Received 19 June 2024; Received in revised form 17 September 2024; Accepted 18 September 2024 Microchemical Journal 207 (2024) 111715 Available online 21 September 2024 0026-265X/© 2024 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC license ( http://creativecommons.org/licenses/bync/4.0/ ). the metabolic processes involved in its breakdown within the body. Consequently, the measurement of DAPs in urine is a reliable indicator for the assessment of OPs exposure, particularly in the context of acute exposure events such as occupational exposure. Pregnant women represent a particularly vulnerable demographic within the general population due to numerous factors. These include placental transfer, high sensitivity to chemicals, and hormonal changes. During pregnancy, the body undergoes significant physiological changes to facilitate foetal development, which can affect the way in which it processes and eliminates certain harmful substances. The growing concern over exposure to harmful substances during the prenatal stages and their potential impact on infants is underscored by this heightened vulnerability. Moreover, research has indicated possible adverse effects on neurodevelopmental linked to prenatal exposure to organophosphate pesticides [5]. In this context, it is necessary to develop effective analytical methods with a high degree of sensitivity for the biomonitoring of these compounds in biological samples. A considerable number of studies have been published on urine samples to date, primarily because OPs compounds undergo hydrolysis and are excreted through this pathway. Nevertheless, there have been little developed of methodologies for other types of biological samples, including amniotic fluid, which could be of great interest for studies of placental transfer. The most commonly employed techniques for the analysis of DAPs in biological samples include traditional extraction methods such as liquid–liquid extraction (LLE) or solid-phase extraction (SPE), azeotropic distillation, or lyophilization extraction, and subsequent analysis by liquid chromatography (LC) and gas chromatography (GC) coupled to mass spectrometry detection (MS) [6–8]. However, the sample treatment involved in GC methods can be cumbersome due to the need for a derivatisation step, resulting in prolonged analysis times. In contrast, methods based on liquid chromatography (LC) exhibit lower limits of quantification (LOQ), typically in the order of ng mL −1 . The analysis of biological samples usually requires a preliminary step before the extraction of compounds. For solid biological samples, such as hair or nails, acid digestion is the most commonly used sample preparation technique. For liquid biological samples, such as urine or plasma, the extraction can be performed directly. However, in most cases, a pre-treatment step to remove lipids and proteins is recommended. In both cases, the resulting mixture is subsequently extracted by either SPE or LLE [8]. Although LLE is a simpler and more straightforward approach, SPE is usually preferred due to its practicality, potential for automation and superior selectivity. On the other hand, azeotropic distillation and lyophilization extraction have been proposed for determining DAPs in urine. Azeotropic distillation is a sample dehydration process that employs cryodehydration, which has the advantage of preventing microbial growth and enzymatic reactions that could chemically alter the sample [9]. Lyophilisation extraction involves the addition of an external component to form an azeotropic mixture, which modifies the relative volatility of the analytes, facilitating their extraction and separation [10,11]. To the best of our knowledge, approaches utilising microextraction sample treatment methods, such as solid-phase microextraction (SPME), hollow fiber liquid-phase microextraction (HF-LPME), electromembrane extraction (EME), dispersive liquid–liquid microextraction (DLLME), or microfluidic devices, have not been addressed yet for DAPs analysis in maternal urine (MU) or amniotic fluid (AF) samples. These microextraction techniques are appealing because they offer excellent advantages including low cost, shorter analysis times, environmentfriendly, high selectivity, sensitivity, and extraction recovery. Furthermore, they allow for a reduction in the consumption of organic solvents and sample volumes. Thus, in this work, we optimised an analytical procedure using EME as a greener approach to convectional sample preparation techniques for the extraction of seven DAPs in urine and amniotic fluid from pregnant women, reducing the extraction time, the consumption of organic solvent to the order of µL, and increasing the sensitivity. EME is a straightforward liquid–liquid microextraction technique that employs an applied potential difference (between two platinum electrodes) to facilitate the electromigration of target analytes from the sample solution to the acceptor phase across a supported liquid membrane (SLM). In comparison to alternative microextraction techniques, EME offers effective sample clean-up, high analyte recovery, and outstanding selectivity within a brief timeframe [12]. Finally, the optimised method was validated in a convenient manner [13] and applied to the analysis of paired MU and AF samples. The outcomes of this comparison demonstrate that our method is more straightforward, less timeconsuming, and enables the determination of seven DAPs involving two different types of matrices. 2. Experimental 2.1. Chemicals, standard solutions and materials DEP (99 %) and DBP (97 %) were acquired from Acros Organics (Fisher Scientific S.L., Madrid, Spain), DMP (98 %) from AccuStandard (New Haven, USA), DEDTP (90 %) and DETP (98 %) were acquired from Sigma-Aldrich (Madrid, Spain), and DMTP (95 %) and DMDTP (99 %) were acquired from Toronto Research Chemicals (Toronto, Canada). Table S1 in the Supplementary Material provides comprehensive information on the chemical structure and physicochemical characteristics of the selected DAPs. Individual stock solutions of each analyte (650 mg L −1 of DMP, 550 mg L −1 of DMTP, 1310 mg L −1 of DMDTP, 1190 mg L −1 of DEP, 560 mg L −1 of DETP, 820 mg L −1 of DEDTP, and 620 mg L −1 of DBP) were prepared in acetonitrile (UltraChromasolv TM , Honeywell, Riedel de Ha¨ en, LC-MS grade) and stored at −80 ◦C. Stock solutions of sodium hydroxide (NaOH, 1 M) and hydrochloric acid (HCl, 1 M) were prepared in Milli-Q Plus water (Millipore, Bedford, MA, USA) from NaOH pellets (≥99 %, Merck, Darmstadt, Germany) and HCl 37 % (Scharlab, Barcelona, Spain), respectively. Subsequently, daily working solutions (donor solutions, acceptor solutions, and analyte standard solutions for UPLC separation control) were prepared in ultrapure water by diluting all the aforementioned stock solutions as required. The organic solvents tested as SLM, including 1-heptanol (≥98 %), 1-nona- nol (≥98 %), 1-octanol (≥99 %), 1-decanol (≥97.5 %) and 1-undecanol (≥99 %) were acquired from Sigma-Aldrich (Madrid, Spain). Methanol (UltraChromasolv TM , Honeywell, Riedel de Ha¨ en, LC-MS grade) and formic acid (FA) (Suprapur®,98–100 %) were acquired from Merck (Darmstadt, Germany). Tripropylamine (TPA 98 %) was acquired from Sigma-Aldrich (Madrid, Spain). Ion pair trypropylammonium formate (TPAF) (1 M) was prepared by dissolving 1.43 g of TPA and 0.96 g of FA in a 10 mL volumetric flask with Milli-Q Plus water. The ion pair additive TPAF 40 mM was prepared by an adequate dilution of TPAF 1 M with pure water. Polypropylene (PP) hollow fibers (S6/2 Accurel® with 800 μ m i.d., 450 μ m wall thickness, and 0.2 μ m pore size) were acquired from Membrana (Wuppertal, Germany). Platinum wires with a diameter of 0.5 mm (Premion®, 99,997 % (metal basis) ≈4.21 g/m), were acquired from Alfa Aesar® (Karlsruhe, Germany). 2.2. Laboratory apparatus and equipment All standards and working solutions were prepared in Milli-Q Plus water obtained from a Milli-Q® Reference A + system (Millipore, Bedford, MA, USA). The system combines deionisation, UV technology and final polishing to produce ultrapure water with a resistivity greater than 18 MΩ⋅cm at 25 ◦C (Type I ultrapure water; ≤5 ppb total organic carbon (TOC)). The pH of the acceptor and donor solutions was monitored using a GLP 21 Crison pH meter (Hach® Lange Spain, Barcelona, Spain) with selectable resolution between 0.1, 0.01 and 0.001 pH. The system offers different measurement modes (by stability, continuous or by time) and calibration modes. The measurements were carried out in stability mode and the system was calibrated by recognition of 3 families of buffers (pH 4.01, 7.00 and 10.90). The extractions were conducted using a TP-2603 DC power supply (Twintex, USA) with programmable voltage within the range of 1–120 V and output current in the range 0–3 A. The power J.M. ´ Avila-García et al. Microchemical Journal 207 (2024) 111715 2 supply consists of two regulated blocks that can operate in constant voltage or constant current mode and as a serial, parallel or independent power supply units 60 V each. Additionally, the equipment offers simultaneous voltage and current readout and voltage and current adjustment. The output current during the extraction was monitored using a professional digital multimeter with True RMS measurement for AC voltage and current, maximum / minimum value storage (MAX-MIN) as well as peak value recording and a high resolution with a maximum display of 22,000 counts (PeakTech® 3430, Ahrensburg, Germany). All measurements were simultaneously recorded to a PC through an integrated USB interface for further analysis and evaluation using the free software “PeakTech DMM-Tool”. All analytical measurements were performed in a triple quadrupole mass spectrometer (Waters Acquity H- Class coupled to a Xevo TQ-XS, Milford, MA, USA) equipped with an ultra-performance liquid chromatography (UPLC) system comprising a Quaternary Solvent Manager (QSM), a 10-tray Sample Manager, a thermostatised Column Manager and an electrospray (ESI) ionisation source. 2.3. Biological samples MU and AF samples were obtained from pregnant women upon delivery. The applied procedures were conducted in accordance with the ethical standards set forth by the Coordinating Committee of Ethics of the Biomedical Investigation of Andalucía (protocol number: 1479-N- 20). Furthermore, each participant was informed of the nature of the study and their informed consent was duly secured. MU samples (125 mL) were gathered immediately post-delivery at the Department of Gynaecology and Obstetrics in two Hospitals in Seville, Spain. During the delivery, AF samples (15 mL) were collected from women with a dilated cervix greater than 6 cm and an intact amniotic sac in order to mitigate any potential risks associated with an accelerated labour. The samples were promptly frozen at −20 ◦C in sterile polyethylene containers. Subsequently, the biological material was transported to the laboratory under proper refrigeration conditions (4 ◦C) and immediately frozen and stored at −80 ◦C upon arrival until extraction. 2.4. Sample electromembrane extraction Prior to EME, MU and AF samples were microfiltered (0.22 µm) and diluted 500-fold and 100-fold, respectively, with ultrapure water. EME was carried out using a classical set-up with polypropylene hollow fibers (HF) as support for the liquid membrane and for housing the acceptor phase. The extraction procedure was accomplished according to the previously reported methodology by Aranda-Merino et al. [14]. Briefly, 30 mm hollow fiber pieces were cleaned before use by washing with acetone. Subsequently, the fibers were thermally sealed and impregnated with 1-octanol for 5 s to form the SLM and to ensure that the pores of the fiber were fully saturated. The excess of the organic solvent was wiped off with a lint-free soft tissue. The hollow fiber was filled with the acceptor phase (50 µL NaOH 10 mM, pH 12) using an HPLC syringe and finally immersed in the sample solution (10 mL, pH 7, not adjusted). Two platinum wires were placed in the donor and the acceptor solutions as electrodes to carry out the extraction process. Efficient migration of target analytes was achieved by applying 60 V (DC power supply, with programmable voltage 1–120 V, TP-2603, TWINTEX, USA) for 10 min at 600 rpm. The electrodes in the donor solution and acceptor solutions were connected to the negative and the positive poles, respectively. The system’s current was monitored using a digital multimeter connected to a computer via an integrated USB interface (PeakTech® 3430, Ahrensburg, Germany). The current readings were taken at 5-second intervals. Following extraction, 10 μ L of the acceptor solution were analysed by LC-MS. 2.5. Chromatographic and mass spectrometry analysis The extracts were analysed by ultrahigh-performance liquid chromatography coupled to a triple quadrupole mass spectrometer (Waters Acquity H-Class coupled to a Xevo TQ-XS, Milford, MA, USA) (UPLCMS/MS) and using electrospray ionisation (ESI). Negative ESI mode and multiple reaction monitoring (MRM) were employed for analyte detection. The quantitation and confirmation of each analyte were achieved by identifying the two most abundant fragment ions. The chromatographic separation was carried out at 40 ◦C using an Acquity® BEH C18 column (50 mm ×2.1 mm i.d., 1.7 μ m particle size) was connected to the ultra-performance liquid chromatography (UPLC) system. Prior to analysis, the samples were maintained at 10 ◦C in the conditioned autosampler. 2.6. Calculations The enrichment factor (EF) was selected as the parameter to assess the efficiency of the extraction process. The EF indicates the preconcentration level of the target analytes under optimal operational conditions. The mathematical expression for this parameter is represented by Eq. (1). The numerator represents the concentration of the analyte in the acceptor solution, C a, final , once the extraction is complete, while the denominator corresponds to the initial concentration of the analyte in the donor solution, C d, initial , prior to the extraction. Enrichment factor (EF) = Ca,final Cd,initial (1) 3. Results and discussion 3.1. Operational parameters optimisation for EME The procedure described in Section 2.4 was employed to optimise the operational parameters in EME. These parameters include the pH of the donor and acceptor phases, voltage, solvents used as SLM, and extraction time. The optimisation was carried out in water samples using a step-by-step method, wherein four variables were fixed, and one was modified in each case. The results were evaluated based on the enrichment factors calculated according to Section 2.6, always seeking their maximum value for the selected compounds. Given the acid nature of the analytes, determined by their pKa values (see Table S1), preliminary tests were conducted by fixing the following experimental parameters at: donor phase pH: 7 (aqueous donor solution 10 ppb), acceptor phase pH: 5 (acidified water), 50 V, 15 min, 600 rpm, and 1-octanol as the solvent used as a SLM. Under these donor and acceptor phase compositions, the analytes of interest are negatively charged. 3.1.1. Organic solvent selection for supported liquid membrane In EME, selecting an adequate SLM is crucial for the efficient extraction of target compounds. In this regard, a wide variety of SLMs have been employed in EME procedures and with different geometric devices [15], meeting the following criteria: (i) low water solubility; (ii) low or null volatility; (iii) sufficient mass transfer permeability; and (iv) adequate electrical conductivity to avoid power failures. On the other hand, the selection of the organic solvent also depends on the chemical nature of the analytes, the sample matrix, and the experimental conditions. As can be seen in Table S1, the studied DAPs, with pKa values in the range 1.30 <pKa <2.90, are acidic, and, therefore, they migrate towards the acceptor solution and through the SLM as anionic species. According to the literature, the Kamlet and Taft solvatochromic parameters (hydrogen bond acidity ( α ), hydrogen bond basicity (β), and dipolarity-polarizability ( π *)) provide valuable information for selecting the organic solvent that best matches the polarity requirements of the target analytes [16,17]. For the extraction of acidic compounds, solvents with high α , low β, and moderate π * values have been found to be the J.M. ´ Avila-García et al. Microchemical Journal 207 (2024) 111715 3 most successful SLMs, with aliphatic alcohols of 8 carbons and upwards showing successful molecular interactions with acidic compounds [18]. Thus, 1-heptanol, 1-octanol, 1-nonanol, 1-decanol, and 1-undecanol were selected for evaluation of extraction efficiency and selectivity. As anions, acidic substances exhibit pronounced hydrogen bond acceptor properties, leading to interactions through hydrogen bonding with the alcohols due to their hydrogen bond donor properties. As depicted in Fig. 1a and b, the general trend observed from 1-octanol to 1-undecanol is a progressive decrease in the enrichment factors for all compounds. Hansen and Pedersen-Bjergaard [19] found that solvents with lower intrinsic hydrophobicity facilitate the mass transfer of medium polar analytes. Therefore, based on the octanol–water partition coefficient (log P) for each solvent (1-octanol (2.97), 1-nonanol (3.77), 1-decanol (4.57), and 1-undecanol (4.72)), it can be concluded that 1-octanol exhibits superior extraction efficiency. This behaviour is also consistent with the viscosity of the solvents used (1-octanol (7.3 cp (25 ◦C)), 1-non- anol (9.12 cp (25 ◦C)), 1-decanol (10.9 cp (25 ◦C)), and 1-undecanol (14.73 cp (25 ◦C))) [20,21]. An increase in the aliphatic chain increases the viscosity, which results in higher electrical resistance of the SLM and, therefore, in a lower mass transfer. In the case of 1-heptanol, this solvent exhibits lower viscosity (5.8 cp (25 ◦C)) and log P (2.62) values. However, it also has a higher water solubility, providing higher electric currents and, consequently, a more unstable liquid membrane compared to 1-octanol. This resulted in lower enrichment factors for most of the analytes, except for DMP and DEP. However, DEP and DMP exhibit higher sensitivity in MS detection than the other DAPs. Therefore, the choice of 1-octanol as SLM appears to be the most appropriate option since it is less water-soluble and provides EME systems with reasonable stability, resulting in higher enrichment factors. 3.1.2. pH of donor and acceptor solutions According to the fundamentals of EME, the diffusion of analytes across the SLM is influenced by the ionic balance, χ . This parameter is defined as the ratio of the total ionic concentration in the donor and acceptor phases, which in turn depends on the pH value of these two solutions. Thus, it is essential to select an optimal pH for both solutions to maximise the ionisation of the analytes and ensure sufficient mass transfer across the liquid membrane when the electrical field is applied [22]. The main role of the donor solution is to facilitate the ionisation of target analytes while minimising or preventing the ionisation of potential sample matrix interfering substances, thus avoiding the coextraction of non-desirable components. To evaluate the influence of the donor solution composition, the pH of the acceptor phase was fixed at 5, while the pH of the donor solution was investigated within 2–12. The results in Fig. 2 show that most of the analytes achieve their maximum extraction efficiency at pH 7, except for DMP and DEP, whose maximum Fig. 1. A and b. Organic solvent selection. HF-EME conditions: pH donor solution: 7; pH acceptor solution: 5; Voltage: 50 V; Extraction time: 15 min; Agitation speed: 600 rpm; n =3. J.M. ´ Avila-García et al. Microchemical Journal 207 (2024) 111715 4 extraction efficiency is observed at pH 4 and 6, respectively. As shown in Table S1, the selected DAPs have pKa values below 3 and the compounds are completely deprotonated and negatively charged at pH values above 4, facilitating their mobility across the liquid membrane. The optimal pH value for the donor phase was selected by establishing a compromise to achieve the highest sensitivity for all target analytes, considering the level of sensitivity obtained for each analyte with mass spectrometric detection. Therefore, pH 7 (analyte aqueous solution, 10 ppb) was selected as the optimal pH donor solution, as it (a) maximises the extraction efficiency for most of the analytes, and (b) ensures sufficient preconcentration for DMP and DEP, enabling the required level of sensitivity for quantification after the extraction, besides not being at their optimum value. Given that the pKa values of the target compounds are below 3, the analytes will be predominantly ionised at pH 7, thus ensuring sufficient electrokinetic mobility. For optimisation the acceptor phase composition, the donor phase was fixed at 7, and the pH was studied between 2 and 13. As illustrated in Fig. 3a and b, different trends can be observed for each analyte, with the highest enrichments observed at pH 12 for most of the compounds. Once more, DMP and DEP exhibit different optimal values, reaching their maximum extraction efficiency at pH 6 and 8, respectively. Considering the theoretical EME model for acidic compounds [23], the pH of the acceptor phase should be higher than that of the donor phase. Furthermore, a strong alkaline environment in the acceptor phase is necessary to achieve an efficient release from the SLM and maintain the analytes in their ionised form, thus avoiding back diffusion into the donor solution in their molecular form. Therefore, pH 12 (NaOH, 10 mM) was established as the optimal acceptor phase for further analysis. The chromatograms showed an absence of evidence suggestive of compound degradation with this acceptor phase composition. No distorted or deteriorated chromatographic peaks were observed, and peak resolution was also unimpaired. Moreover, the extracts were stable under these strong alkaline conditions for a minimum of two weeks when stored frozen at −80 ◦C. 3.1.3. Applied voltage optimization In EME, the total mass transfer is significantly affected by the magnitude of the electric field generated between the two electrodes. Consequently, the applied voltage represents another crucial operational parameter that must be carefully evaluated, as it represents the primary driving force for the mobility of analytes from donor to acceptor solution [24]. In general, the extraction efficiency increases with increasing voltage, as high voltages facilitate the flow of analytes from the donor to the acceptor solution [25]. However, excessive voltages can generate high electrical currents, which, in turn, can cause membrane instability and lead to a decrease in extraction efficiency. Hence, it is important to carefully consider and optimise the applied voltage in order to achieve an efficient and selective extraction of analytes without compromising the stability of the system. A range of voltages between 10 and 100 V was evaluated. Fig. 4 summarises the results obtained. The plot indicates that the enrichment factor increases gradually from 10 V to 60 V for most analytes. No further improvement was observed at voltages higher than 60 V, except for DBP and DMTP, which reached maximum efficiency at 100 V and 80 V, respectively. On the one hand, the enrichment factors for DBP and DMTP at 60 V are high enough for adequate quantification after their extraction. On the other hand, the EFs values at 60 V for these two compounds (DBP and DMTP) are only 20 % lower than their optimal values. Therefore, the extraction voltage was selected by establishing a compromise, and 60 V was considered the optimal voltage. 3.1.4. Extraction time Extraction efficiency is influenced by the interdependent relationship between extraction time and voltage. Thus, prolonged exposure to the electric field may compromise the integrity of the liquid membrane [26,27]. Therefore, the extraction time must be carefully optimised in order to achieve a balance between the efficient analyte transfer and the stability of the SLM. As can be seen in Fig. 5, all the analytes reach their maximum enrichment value at 10 min, except for DBP, which reaches its maximum at 30 min. Further analysis of the results indicates that the increase in the enrichment factor for DBP from 10 to 30 min is less than 20 %. Considering the high sensitivity of this compound to mass spectrometric detection, it seems reasonable to select an extraction time of 10 min, which significantly reduces the analysis time and provides the highest enrichment factors for most of the analytes. Thus, 10 min was established as optimal extraction time for the remainder of the study. Therefore, the optimal extraction conditions can be summarised as follows: donor solution: 10 µg L −1 aqueous solution containing seven DAPs (10 mL; pH 7), SLM: 1-octanol, and acceptor phase: 50 µL alkaline aqueous solution (pH 12, 10 mM NaOH). The extractions were conducted at 60 V for 10 min at 600 rpm. Fig. 2. Optimisation of donor solution pH. HF-EME conditions: SLM: 1-octanol; pH acceptor solution: 5; Voltage: 50 V; Extraction time: 15 min; Agitation speed: 600 rpm; n =3. J.M. ´ Avila-García et al. Microchemical Journal 207 (2024) 111715 5 3.2. Optimization of MS/MS parameters and chromatographic conditions For MS/MS optimisation, 50 μ g L −1 individual standard aqueous of each DAP were directly infused into the mass spectrometer. The analytes were ionised in negative mode using an ESI source and data acquisitions were carried out in full-scan mode using different source voltages and temperatures. The sensitivity of each analyte was quantified, yielding a valid signal with a value exceeding 10 5 counts per second. Once the initial conditions were established, successive injections were performed by modifying the instrumental conditions of the mass spectrometer in order to obtain the most favourable detection conditions. Table S2 provides a summary of the optimal detection parameters, including the MRM transition ions, collision energies and cone voltages. The most intense signal (Q1) was used for quantification and the second one (Q3) was used for confirmation purposes. Argon was employed as the collision gas (99.995 %, Praxair) while nitrogen was used as the desolvation gas (600 L/h) and the cone gas (150 L/h). The desolvation and source temperatures were set to 550 ◦C and 150 ◦C, respectively. The capillary voltage was established at −1.0 kV. Chromatographic separation was optimised using standard mixtures of the analytes and based on the method previously described by Cequier et al. [28], with minor modifications. The composition of the mobile phase and the detection conditions were evaluated in order to achieve symmetric peaks and well-defined signals as well as good sensitivity and selectivity in a reasonable chromatographic analysis time. A mobile phase comprising 0.5 mM formic acid/trifluoroacetic acid in aqueous solution (A) and acetonitrile was employed at a flow rate of 0.4 mL min −1 for 12 min. The following gradient elution was applied: 0 min 99 % A, 9.0 min 90 % A, 9.5 min 70 % A, 11.5 min 70 % A, returning to initial conditions in another 0.5 min. A two-minute waiting period was set between injections to re-equilibrate the chromatographic column. 4. Validation of the proposed methodology The analytical methodology proposed in this research was validated following the EURACHEM (European Analytical Chemistry) guide principles [13]. In order to ascertain the suitability of the method the following quality parameters were assessed: linearity, sensitivity, precision, and accuracy. The figures of merit are presented in Table 1. The linearity was evaluated with a ten-point calibration curve constructed by subjecting aqueous standard mixtures at different concentrations (within 0.0001–10 ng mL −1 ) to the EME procedure described in Section Fig. 3. A and b. Optimisation of acceptor phase pH. HF-EME conditions: SLM: 1-octanol; pH donor solution: 7 (aqueous donor solution); Voltage: 50 V; Extraction time: 15 min; Agitation speed: 600 rpm; n =3. J.M. ´ Avila-García et al. Microchemical Journal 207 (2024) 111715 6 2.4 This was performed in triplicate. The calibration curve was obtained by least squares linear regression analysis of the data set obtained. As seen in Table 1, good linearity was obtained for all analytes, with determination coefficients R 2 >0.9988 in all cases. The sensitivity was determined in diluted sample blanks for each matrix, in terms of method limits of detection (MLOD) and quantitation (MLOQ), calculated from a signal-to-noise (S/N) ratio of 3 and 10, respectively. Given the complex matrix of urine and AF (high salinity), dilution was required to avoid or minimise potential matrix effects that may cause electrolysis phenomena and current instabilities during EME performance. The dilution ratios for MU and AF were 1:500 (v/v) and 1:100 (v/v), respectively, to completely avoid the matrix effect. As seen in Table 1, exceptional sensitivity was achieved, with MLOD between 22.1–500 pg mL −1 in urine matrix and 4.41–100 pg mL −1 in AF. This high sensitivity observed can be attributed to both, the EME procedure and the analytical instrumentation employed (UPLC-MS/MS). To assess the precision of the method, repeatability (REP) and intermediate precision (IP) were evaluated. Spiked samples (free of analytes) at three concentration levels (2, 50, and 500 ng mL −1 ) were properly diluted and subjected (in triplicate) to the entire analytical procedure and measured on a single day (REP) and one day per week over a month (IP). The relative standard deviation percentages (%RSD) for REP were found to be within the range of 1.7–8.4 % while those for IP were within 4.4–9.9 %. The EFs shown in Table 1 were calculated using the external (0.1–50 ng mL −1 ) and extracted (0.0001–10 ng mL −1 ) calibration curves. The EFs ranged from 14 to 178, observing the higher values for DMTP, DEDTP and DMDTP, followed by DETP, DBP with EFs of 44 and 32, respectively. The lower EFs values were for DEP and DMP, which hardly showed any preconcentration. The latter can be attributed to the careful selection of extraction conditions, which were optimised to maximise the detection capabilities while maintaining a balance between sample enrichment and instrumental detection. The extraction conditions were favourably selected for those compounds that exhibited lower sensitivity in mass spectrometry detection. In the case of DEP and DMP, the enrichment Fig. 4. Optimisation of applied voltage. HF-EME conditions: SLM: 1-octanol; pH donor solution: 7 (aqueous donor solution); pH acceptor phase: 12 (NaOH, 10 mM); Extraction time: 15 min; Agitation speed: 600 rpm; n =3. Fig. 5. Optimisation of extraction time. HF-EME conditions: SLM: 1-octanol; pH donor solution: 7 (aqueous donor solution); pH acceptor phase: 12 (NaOH, 10 mM); Voltage: 60 V; Agitation speed: 600 rpm; n =3. J.M. ´ Avila-García et al. Microchemical Journal 207 (2024) 111715 7 factor was deliberately sacrificed in favour of heightened sensitivity during the detection stage. Before the analysis of real samples, a recovery assay was conducted in MU and AF to assess the matrix effect and potential interferences present in the sample that could affect the signal. First, selected MU and AF samples were analysed to check the possible presence of any target analytes. Once confirming the samples did not contain any DAPs, they were employed to assess the accuracy of the method. The samples were spiked at three concentration levels (2, 50, and 500 ng mL −1 ), microfiltered (0.22 µm), diluted, and subjected to the extraction procedure described in Section 2.4. The fortification levels were selected considering the linear range of the method. According to the obtained results (Table 2), the relative recovery percentages (RR%) were not significantly different between the three spiked concentration levels analysed nor between the compounds. As can be seen, RR% were consistently above 80 % with %RSD below 9 %. Therefore, it can be stated that the proposed EME procedure is adequate for the preconcentration and selective extraction of DAPs in MU and AF samples. A representative chromatogram of a spiked urine at 50 ng mL −1 is presented in Fig. S1. 5. Comparison with previous methods Table 3a. details a comparative evaluation of the procedure proposed in this study with other reported procedures described in the literature. The evaluation is focused on three key areas: sample volume, chromatographic separation time, and figures of merit. Most techniques described in the literature employ classical microextraction approaches such us, liquid–liquid extraction (LLE) or solid-phase extraction (SPE), QuEChERS, or direct injection. The proposed microextraction method in this study achieves high preconcentration for most analytes, thereby enabling high method sensitivity and the determination of the target compounds at low concentrations in samples. As observed, our method requires a very small sample volume (20 µL for urine and 100 µL for AF) compared to other methods previously reported. This feature is particularly advantageous when dealing with very small sample volumes, such as a baby’s urine. Therefore, our method is highly beneficial for extrapolating the analysis of these dialkylphosphates to infants. Our proposal can be considered a highly efficient alternative for rapid sample analysis with minimal chromatographic and extraction times in comparison to other methods, including those proposed by Cequier et al. [28] and Odetokun et al. [29]. Regarding the recovery percentages, the proposed method yields highly satisfactory results for most of the compounds analysed in both matrices. These recoveries are slightly superior to those suggested by Oya et al. [30] and Koutroulakis et al. [31] for those DAPs with comparable or better sensitivity levels. Only the method proposed by Sinha et al. [32] shows superior recoveries and a better level of sensitivity. However, the analysis time and sample processing time are considerably longer than those required by our method. In terms of sensitivity, our method is as one of the most sensitive of the presented alternatives, with the exception of DMP. Of particular note are DBP and DETP, which are the most sensitive compounds in the method. It is important to acknowledge that previously reported methods with comparable or superior sensitivities involve much longer sample treatment procedures, including multiple stages as exemplified by the approaches proposed by Q. Lu et al. [33], Oya et al. [30] (SPE), and Sinha et al. [32] (lyophilisation +LLE), or derivatisation steps as in the case of the procedure proposed by Koutroulakis et al. [31]. Therefore, our proposal can be considered an improvement over existing procedures. On the other hand, the greenness assessment was evaluated using the free open-access software AGREEprep [34] to compare the different sample procedures reported in the literature. This greenness metric tool was developed in 2022 to improve upon its previous version, AGREE, and was specifically designed to evaluate the Table 1 Figures of merit of the method. Analyte Linear range (ng mL ¡1 ) (R 2 ) MLOD (pg mL ¡1 ) MLOQ (pg mL ¡1 ) EF REP/IP (%RSD) U AF U AF 2* 50* 500* DMP MLOQ-5000(U)/1000 (AF) 0.9993 (U) 0.9995 (AF) 500 100 1667 333 14 5.1/8.4 2.5/5.9 1.7/3.9 DMTP MLOQ-5000(U)/1000 (AF) 0.9998 (U) 0.9997 (AF) 92.3 18.5 308 61.5 178 5.3/7.5 3.1/4.4 4.5/7.2 DMDTP MLOQ-5000(U)/1000 (AF) 0.9995 (U) 0.9996 (AF) 150 30.0 500 100 152 6.7/7.1 5.6/6.3 5.2/6.9 DEP MLOQ-5000(U)/1000 (AF) 0.9988 (U) 0.9992 (AF) 125 25.0 417 83.2 22 3.9/8.1 3.4/5.0 6.3/5.9 DETP MLOQ-5000(U)/1000 (AF) 0.9996 (U) 0.9997 (AF) 22.1 4.41 75.5 14.7 44 8.4/9.9 5.5/5.2 6.0/6.9 DEDTP MLOQ-5000(U)/1000 (AF) 0.9994 (U) 0.9995 (AF) 147 29.4 490 98.0 167 4.3/4.4 4.6/6.8 5.5/7.5 DBP MLOQ-5000(U)/1000 (AF) 0.9997 (U) 0.9994 (AF) 25.1 5.05 83.3 16.6 32 4.1/6.8 2.2/3.1 4.7/6.4 *Data were calculated with three replicates. Table 2 Relative recoveries (%) in MU and AF samples spiked at three concentration levels. Maternal urine RR* (% RSD) Amniotic fluid RR* (% RSD) Analyte Fortified level (ng mL ¡1 ) Fortified level (ng mL ¡1 ) 2 50 500 2 50 500 DMP 85 (5) 93 (2) 90 (2) 95 (4) 90 (3) 90 (3) DMTP 84 (5) 80 (4) 74 (8) 80 (5) 84 (7) 64 (9) DMDTP 93 (6) 82 (5) 70 (7) 93 (5) 99 (5) 104 (5) DEP 91 (5) 89 (3) 79 (3) 90 (8) 95 (1) 94 (1) DEDTP 89 (6) 87 (6) 76 (7) 82 (5) 98 (1) 100 (1) DEDTP 89 (4) 90 (4) 80 (4) 95 (6) 93 (2) 83 (2) DBP 92 (7) 99 (1) 82 (1) 89 (6) 95 (2) 97 (2) *Average of three determinations (% RSD). J.M. ´ Avila-García et al. Microchemical Journal 207 (2024) 111715 8 sample preparation stage [35]. This tool provides a score calculated according to the ten principles of green sample preparation [36]. The scores for each criterion are assigned weights and combined to generate an overall score, which ranges from 0 to 1, with 1 representing optimal performance. Table 3b. summarises the results for the greenness assessment of the selected articles. As can be seen, the proposed method in this work not only achieves the best overall score in terms of green assessment compared to other previously published methods, but it also demonstrates its greatest strengths in reducing the use of solvents and hazardous substances, sample consumption, sustainability, renewability, reusability of materials, and energy consumption. Therefore, it can be considered to be environmentally sustainable overall. However, its major weakness lies in its automation capability. It is important to note that most of the previously reported procedures also suffer from this weakness, except for the method proposed by Odetokun et al. [29], which integrates a certain degree of automation in the sample preparation stage. Compared to previously published methods that exhibited similar or superior sensitivities for some compounds [30–33], our proposal demonstrated a more environmentally sustainable profile with a considerably higher overall score, in addition to the aforementioned advantages. Only the method proposed by Babina et al. [37] shows a superior green assessment profile, although not in the overall score, but in the analysis of the ten criteria individually. However, this procedure involves the direct injection of the sample, which results in much lower sensitivities, and potential matrix effects in the measurement stage. Finally, the practicality of the proposed method was also evaluated. White Analytical Chemistry (WAC), a concept introduced in 2021 by Nowak et al. [38], emphasises the importance of considering not only the analytical performance of the method but also its practical aspects of implementation and environmental sustainability, integrating ecological considerations, analytical precision, and practical applicability. This comprehensive approach aimed to address the environmental impact of analytical methods while maintaining the rigour and functionality required in scientific analyses. In 2023, Manousi et al. introduced a notable enhancement to WAC, especially focusing on the practical aspects of its implementation [39]. This recent complement, known as the Blue Applicability Grade Index (BAGI), emphasises the importance of practical applicability, ensuring that WAC can be easily integrated into laboratory workflows and everyday analytical practices. The BAGI index is a simple metric tool employed to derive a score that assesses the method’s applicability. The BAGI index generates two correlated outcomes: a numerical score and an asteroid-shaped pictogram displaying the overall evaluation result at its centre. Scores ranging from 25 to 100 are assigned, with a higher score indicating greater practicality of the method. A score of 60 or above is indicative of the method’s applicability. The strengths and weaknesses of the method are visually represented using a pictogram, thereby providing a comprehensive depiction of its characteristics. The methods were evaluated using BAGI 0.9.2 free software [40]. The software considers various attributes, including the type of analysis, the number of analytes that can be measured simultaneously, the analytical technique, the required instrumentation, the number of samples that can be analysed at once, the sample treatment, the amount of samples that can be processed per hour, the reagents and materials employed, the need for preconcentration, the level of automation, and the sample volume required. Table 3b. illustrates the resulting BAGI index pictograms for all the compared methods. The proposed method in this study achieved a BAGI score of 70, thereby confirming its applicability and resulting in the highest score among all the selected methods in this comparison. The proposed method in this study demonstrates significant strengths in preconcentration and the generation of precise and confirmatory results. However, it has limitations in terms of automation, the ability to process multiple samples simultaneously, and the reliance on sophisticated instrumental techniques. Only the methods proposed by Q. Lu et al. [33] and Babina et al. [37] achieved a score over 60, thereby demonstrating their practical applicability. Regarding the other previously reported methods, they Table 3a Comparison of the method main characteristics. Ref. Sample Volume (mL) Extraction Method Instrumental Technique Run time (min) %R LOQs (ng mL ¡1 ) Matrix DMP DEP DBP DMTP DETP DMDTP DEDTP This study 0.02 0.1 EME LC-MS/MS 12 64–104 1.67 0.33 0.42 0.08 0.08 0.02 0.31 0.06 0.08 0.01 0.5 0.1 0.5 0.1 U AF [3] 4 LLE+salting-out LC-MS/MS 20 82–117 0.5 0.5 NA 0.5 0.5 0.5 0.5 U [28] 0.3 X-AW96-well plates UPLC-MS 11 39–112 4 1.3 NA 0.66 1 1.9 0.5 U [29] 0.6 SPE LC-MS/MS 10 99.99–100 1.4 0.13 NA 0.2 0.33 0.22 4.65 U [30] 5 SPE UPLC-MS/MS −54–101 1.53 1.77 NA 0.09 0.02 0.14 0.07 U [31] 5 LLE+salting-out +Derivatisation CG-MS 32.2 59–114 0.13 0.07 NA 0.05 0.05 NA 0.04 AF [32] 10 Lyophilize +LLE LC-MS/MS 35 93–107 0.07 0.03 NA 0.04 0.49 0.03 0.07 U [33] 0.2 SPE LC-MS/MS 13 80–127 1.13 0.04 NA 0.05 0.03 0.03 0.03 U [37] 1 Direct inject LC-MS/MS 17 89–103 1* 1* NA 1* 1* 1* 1* U [41] 5 QuEChERS Orbitrap LC-MS 20 69–127 10 10 NA 1.6 3.2 1.6 3.2 U [42] 1 Double SPE LC-MS/MS 12 64–105 1.2 0.6 NA 1.2 0.3 NA NA U [43] 3 LLE LC-MS/MS 20.6 100–122 NA 2* NA NA 1* NA 0.6 U [44] 5 LLE LC-MS/MS 22 13.2–99.8 2 2 NA 2 2 2 2 U [45] 4 ND CG-MS/MS −91–96 0.10 0.20* 0.20* NA 0.10 0.20* NA NA NA U AF* [46] 2 QuEChERS LC-MS/MS 18 ND 0.15* 0.10* NA 0.15* 0.10* 0.15* 0.10* U *LOD; ND: not described; NA: not analysed. J.M. ´ Avila-García et al. Microchemical Journal 207 (2024) 111715 9