AnChem1904013Shelepchikov fm ISSN 1061 9348, Journal of Analytical Chemistry, 2019, Vol 74, No 6, pp 574–583 © Pleiades Publishing, Ltd , 2019 Russian Text © The Author(s), 2019, published in Zhurnal.
ISSN 1061-9348, Journal of Analytical Chemistry, 2019, Vol 74, No 6, pp 574–583 © Pleiades Publishing, Ltd., 2019 Russian Text © The Author(s), 2019, published in Zhurnal Analiticheskoi Khimii, 2019, Vol 74, No 6, pp 426–236 ARTICLES A New Method for Purifying Fat-Containing Extracts in the Determination of Polybrominated Diphenyl Ethers A A Shelepchikova, b, *, V V Ovcharenkoa, A I Kozhushkevicha, E S Brodskiib, A A Komarova, K A Turbabinaa, and A M Kalantaenkoa a The Russian State Center for Animal Feed and Drug Standartization and Quality, Moscow, 123022 Russia bSevertsov Institute of Ecology and Evolution, Russian Academy of Sciences, Moscow, 119791 Russia *e-mail: dioxin@mail.ru Received November 13, 2017; revised July 1, 2018; accepted July 1, 2018 Abstract—We developed a sample preparation method for the determination of polybrominated diphenyl ethers (PBDEs) with from one to ten bromine atoms in samples of feed and food products containing approximately 0.5 g of animal fat or vegetable oil The method involves gas chromatography with high-resolution mass spectrometry or tandem mass spectrometry A possibility of using various reagents for the purification of extracts by chemical reactions and fractionation is studied The physicochemical properties of PBDEs and polychlorinated biphenyls (PCBs) have significant differences, and to determine the full range of PBDEs, it is necessary to use other methods of sample preparation than in the case of PCBs The conditions selected for the purification of extracts in a column filled with potassium silicate, Florisil, and silica impregnated with sulfuric acid and for their fractionation using activated neutral alumina ensure the PBDE recoveries of at least 75% Purification of the extracts can be carried out without the use of chlorinated organic solvents Applied aspects of instrumental analysis and measurement quality assurance are also described Keywords: polybrominated diphenyl ethers, sample preparation, fractionation, food and feed, biological samples, organic pollutants DOI: 10.1134/S1061934819040130 Polybrominated biphenyl ethers are products of target industrial synthesis and are used to reduce the flammability of polymeric materials The production of PBDEs began in the 1970s in Germany There are three main industrial products: penta-, octa-, and decabromodiphenyl ether (DeBDE) The last product is mainly used in the electronics industry, accounting for approximately 82% of world production; the other two compounds are congener mixtures used in the plastics industry and in the furniture industry [1, 2] There are 209 of PBDE congeners in total; they contain from one to ten bromine atoms Because of cumbersome names of the systematic nomenclature, the PBDE names use arithmetic numbers that coincide with the IUPAC numbers for polychlorinated biphenyls with substituents in the same positions of aromatic rings [3] For example, BDE-99 corresponds to 2,2',3',4',5-PeBDE; to denote isomer groups in bromination degrees, conventional prefixes derived from the roots of Greek and Latin numerals (mono-, di-, tri-, tetra-, etc.) and abbreviations (MoBDE, DiBDE, TrBDE, TBDE, etc.) are used Active studies of environmental pollution and biological samples with PBDEs began approximately 20 years ago [4] The results of these studies were the rea- son for banning or restricting the use of PBDEs in the United States and the European Union In 2009, technical mixtures of penta- and octabromodiphenyl ethers were included in the expanded list of the Stockholm Convention on Persistent Organic Pollutants; DeBDE is a candidate for inclusion in this list Conventional approaches to the determination of these substances have not been formed By their structure PBDEs are similar to PCBs, and it can be assumed that the methods for their isolation are similar In practice, this is true only for medium-brominated compounds, which are the most common substances for determination [5] Highly brominated congeners, including DeBDE, are determined less often because of difficulties in chromatographic separation of these compounds Mono- and dibromodiphenyl ethers are determined even less often, and the authors of [6–8], recognizing the possibility of their determination, point out the problem of low recoveries or not give them at all for MoBDE This article is devoted to the development of a method of the purification of fat-containing extracts, which enables the determination of PBDEs with any degree of bromination and monitoring the level of contamination of feed and food Other applied aspects 574 A NEW METHOD FOR PURIFYING FAT-CONTAINING EXTRACTS of analysis and their relationship with the sample preparation procedure are also described PECULIARITIES OF DETERMINATION OF POLYBROMINATED DIPHENYL ETHERS Like most methods of organic trace analysis, the determination of PBDEs in various samples consists of three main stages: extraction, purification (separation of the target compounds from other extracted components of the matrix), and instrumental analysis Extraction from samples of animal lipophilic organic pollutants including PBDEs actually comes down to the extraction of fat, which is a relatively simple task It is more challenging to extract PBDEs from samples of plant origin, to which the analytes pass from the atmosphere or from the soil In both cases, extraction techniques that have proven effective for PCBs or polychlorinated dibenzo-p-dioxins and dibenzofurans (PCDD/PCDF) can be used; there are no prerequisites to assume that the extraction method, useful for PCBs and even more so for PCDD/F, would be unsuitable for PBDEs In this regard, we not consider the stage of extraction in this paper The size of a sample and the corresponding procedure for the purification of extracts depend on the sensitivity requirements of the method of quantitative determination and the available measurement equipment The levels of PBDEs in samples can be conventionally described as significantly lower than the levels of PCBs but higher than those of PCDD/PCDF The primary method for determining PBDEs is gas chromatography–high-resolution mass spectrometry (GC–HR-MS), a method used to quantify PCDD/PCDF when the highest sensitivity requirements are imposed Because of the high sensitivity and selectivity of GC–HR-MS, small samples with minimal purification can be analyzed, for example, by passing the extract through a Pasteur pipette filled with silica impregnated with sulfuric acid and/or filtering the sample through activated silica or Florisil Such an approach not only saves solvents and adsorbents but also helps to decrease contamination of the blank sample because PBDEs are present in almost all solvents and adsorbents In our case, this technique is not applicable, since it is necessary to be able to work with samples containing a sufficiently large amount of fat Because of the structural similarity of PBDEs and PCBs, it is usually proposed to use techniques that have previously been tested for PCBs for purification of extracts The main method can be considered the destruction and adsorption of labile matrix components on a multilayer column consisting of layers of silica impregnated with sulfuric acid and potassium hydroxide or silicate, separated by anhydrous sodium sulfate, and fractionation on alumina, when the analytes are eluted with a mixture containing several perJOURNAL OF ANALYTICAL CHEMISTRY Vol 74 575 cents of dichloromethane (DCM) in hexane A similar purification algorithm is implemented in the most well-known automatic sample preparation system from FMS (Waltham, United States) However, when analyzing fish meal samples using the FMS TotalPrep system according to the procedure proposed by the manufacturer, we obtained recovery rates of 44– 77% for medium-brominated BDEs; MoBDE was absent in the extract, and the recovery of DiBDE did not exceed 15% The contamination level of the blank BDE-47 sample was approximately 20 pg when using specialized FMS disposable columns for PBDEs Other users of FMS systems also encountered the problem of losing low-brominated congeners [9, 10] A radical increase in the volume of solvents and the use of pure dichloromethane instead of its mixture with hexane does not allow the recovery to reach even 10% for MoBDE [11] The procedure recommended by the Ministry of the Environment of the Canadian Province of Ontario, using the FMS automatic sample preparation system, ensures the determination of PBDEs containing at least three bromine atoms [12] The developers of Method 1614, which is an official method for determining PBDEs of the US Environmental Protection Agency, also probably faced the problem of extracting MoBDE and DiBDE, as indicated by the absence of criteria for assessing recovery rates for these compounds [13] However, they did not use the automatic sample preparation system but proposed a procedure similar to Method 1668 for determining PCBs [14] A separate problem in the determination of PBDEs is their chromatographic separation: in addition to the absence of columns capable of separating all existing isomers [15, 16], these substances have low volatility with insufficient thermal stability Polybrominated diphenyl ethers containing up to 5–6 bromine atoms can be determined by GC–MS using DB-5ms column (5% of 1,4-bis(dimethylsiloxy)phenylenomethylpolysiloxane) or HT-8 columns (8% of phenylpolycarbonate siloxane), 25–30 m in length with a stationary phase layer thickness of 0.22–0.25 μm, typical for determining PCDD/PCDF or PCBs In the case of heavier congeners, a sharp decrease in sensitivity is observed up to the complete disappearance of chromatographic peaks There is information [15] on the determination of BDE-209 using long columns with different stationary phases; however, the amount of substance injected into the chromatograph must be taken into account In our experience, the problem of chromatography of BDE-209 resembles the situation with DDT, when some constant amount of substance is subject to thermal decomposition; that is, the higher amount of the substance introduced, the smaller the relative loss In order to determine subnanogram quantities of highly brominated PBDEs reliably, it is advisable to use a special J&W DB-5ht chromatographic column 10–15 m in length with a thinner layer (0.1 μm) of inert stationary phase (95% of methylsi- No 2019 576 SHELEPCHIKOV et al loxane, 4% of phenylsiloxane, and 1% vinylsiloxane) Using such a column, PBDEs with any degrees of bromination can be detected, but the quality of the separation of the isomers could be rather low, and a substantial distortion of the peaks in the initial part of the chromatogram could occur However, the insufficient purification of the extracts usually does not affect the highly brominated compounds To obtain reliable quantitative results, at least two different chromatographic columns should be used EXPERIMENTAL The published data and our experience show that the loss of low-brominated diphenyl ethers occurs during fractionation on activated alumina and they may be absent in simplified methods, for example, when removing the bulk of fat by freezing and further purification in a multilayer column [7] or using gel chromatography [17] Unfortunately, these and other options for purification of extracts without fractionation cannot be considered as universal methods for the routine determination of traces of PBDEs in fatcontaining matrices There are data [18] on low, but not zero recoveries of mono- and dibrominated diphenyl esters, obtained using alumina cartridges for the purification of soil extracts or adding it to the cartridge for accelerated solvent extraction; however, no information about the brand of cartridges or alumina is given There is also no data on cartridges used in the procedure for determining PBDEs of any degree of bromination in milk [19] The inability to elute MoBDE from activated basic alumina with dichloromethane and toluene quantitatively suggested that debromination or another chemical transformation of the substances takes place; in other words, the basic principle of quantitative analysis method, i.e., the absence of chemical reactions (except target derivatization) between the substances to be determined and the reagents used, is violated We decided to create a new procedure for determining the full range of PBDEs rather than to adapt the available procedures, for which it was necessary to study the possibility of using different adsorbents and techniques Test samples The effectiveness of the developed procedure was tested using three types of samples: pork fat, fish oil, and sunflower oil They are representatives of the three main fat-containing matrices: animal, fish, and vegetable There are no values of maximum permissible concentrations or other standards for the concentration of PBDEs in feed and food products; the European Union directive on monitoring these compounds in food products specifies a limit of determination of no less than 10 pg/g of wet weight [20] Considering that the same instruments are used to determine PBDEs and PCDD/PCDF, and the weighed portions of the latter contain no more than 3–5 g of fat in their routine determination, we can assume that 0.5 g of fat should be sufficient to estimate the concentration of PBDEs Equipment At the preliminary stage of research, a Thermo TSQ8000 Evo triple quadrupole was used in the MS/MS mode with a Trace 1310GC gas chromatograph equipped with a Thermo TR-5MS column 30 m in length, 0.25 mm in diameter, and the thickness of the stationary phase layer of 0.25 μm A sample of 1.5 μL in volume was injected in the splitless mode at the injector temperature of 290°C; purging of the injector was 1.5 after the injection The temperature program of the chromatographic separation was as follows: the initial temperature of the thermostat was 140°C; holding at this temperature for min; heating to 220°C at a rate of 10 deg/min; then heating to 245°C at a rate of deg/min and to 290°C at a rate of 10 deg/min; holding at this temperature until the end of the elution Under these conditions, ethers from MoBDE to HxBDE and sometimes HpBDE can be determined The use of MS/MS techniques in some cases gives mass-chromatograms that are more convenient for interpretation, especially, in the case of DiBDE, the exact masses and retention times of which are close to PeCB; however, the error in determining the recoveries of isotope-labeled reference compounds is higher, which is described in more detail below The remaining PBDEs were detected, and a confirmatory determination of low-brominated congeners was performed using a Waters AutoSpec Premier high-performance chromatography–mass spectrometer with J&W DB-5ht column (length 10 m, internal diameter 0.25 mm, and thickness of the stationary phase layer 0.1 μm), SGE BPX-5 column (length 25 m, internal diameter 0.22 mm, and thickness of the stationary phase layer 0.25 μm), and SGE HT-8 column (length 25 m, internal diameter 0.25 mm, and thickness of the stationary phase layer 0.25 μm), connected to the mass spectrometer via a 2.5-m capillary with an internal diameter of 0.15 μm The temperature conditions are given in Table Two characteristic isotope ions were detected for each isotope-labeled and native PBDE, the isotopic ratio was checked for correctness, and the average value of the ionic current of the isotopic cluster was calculated, which was then used for quantitative calculations and for determining the recovery Extraction was carried out using a Dionex ASE 200 and a Thermo ASE 350 accelerated solvent extractors with 33- and 100-mL cells For chemical purification and fractionation of samples, glass columns with a length of 200 mm and an inner diameter of 14 mm and columns with a length of 150 mm and an inner diameter of 10 mm were used, having a narrowing on one side and a 14/23 ground JOURNAL OF ANALYTICAL CHEMISTRY Vol 74 No 2019 A NEW METHOD FOR PURIFYING FAT-CONTAINING EXTRACTS 577 Table Conditions of chromatographic separation Column Parameter J&W DB-5ht SGE BPX-5 SGE HT-8 Initial temperature, holding time 170°С, 1.5 160°С, 135°С, Heating, rate To 240°С, 20 deg/min to 270°С, 15 deg/min to 295°С, 10 deg/min To 220°С, deg/min to 295°С, deg/min To 170°С, 15 deg/min to 270°С, deg/min to 295°С, deg/min Holding time at final temperature 13.5 20 5.2 Injector temperature 290°С 285°С 275°С Sample injection mode μL, splitless, 1.5 μL, splitless, 1.9 μL, splitless, 0.8 mL/min, constant flow 0.8 mL/min, constant flow Carrier gas (helium) flow rate 1.3 mL/min, constant flow glass connector on the other side to connect to the tank Solvents and materials The following adsorbents were used: basic alumina with Brockmann I activity (Sigma-Aldrich, 199443); neutral alumina with Brockmann I activity (Sigma-Aldrich, 199974); neutral alumina, type WN-6, with Super I activity (Sigma-Aldrich, A1522-500); Florisil (0.150– 0.250 mm, Merck, 1.12994.1001); Florisil PR (Merck, 20280); silica gel 60 (0.063–0.100 mm, Merck, 1.07734.9025), and high-purity silica (70–230 mesh, Merck, 7754) The method of preparation of the listed materials differed in different experiments and is discussed below; activated adsorbents were cooled to 80°C, and transferred to an airtight container, where they were stored until use Sodium sulfate (Acros Organics, 196640050) was calcined for 16 h at 550°C, cooled to 80°C, and transferred to an airtight container, where it was stored until use Potassium silicate was synthesized by adding silica to an equimolar solution of potassium hydroxide in methanol under constant stirring; the reaction mixture was left for day in a desiccator Then, excess methanol was decanted, and the product was dried and kept for 16 h at 250°C Silica impregnated with sulfuric acid was prepared by mixing the activated silica with conc H2SO4 to form a homogeneous mass During the fractionation of samples, the column with the adsorbent was conditioned with 15–20 mL of hexane before injecting the sample A mixture of isotope-labeled PBDE standard reference compounds (MBDE-MXG and PBDE-ISS-G) and a mixture of native congeners (BRF-PAR) were purchased from Wellington Laboratories Solvents from various suppliers were tested for the absence of interfering components The problem of contamination in the blank sample is discussed below Recoveries and quality assessment criteria for purification When using isotope-labeled internal stanJOURNAL OF ANALYTICAL CHEMISTRY Vol 74 dards, especially in the version of the isotopic dilution method, the recoveries are often not essential, and the error in their determination can be very high For example, in the determination of PCDD/PCDF, the recoveries range from 16 to 279% according to the EPA procedure [21]; a range of 60–120% is considered acceptable in the European Union for the quantitative determination of PCDD/PCDF [22] For PBDEs containing from three to nine bromine atoms, according to the EPA Method 1614, the recovery of PBDEs should be in the range from 25 to 150%, and for DeBDE, they can vary from 20 to 200% [13] Currently, several mixtures of isotope-labeled internal PBDE standards are available, containing at least one isomer of each bromination degree To estimate recovery rates, it is proposed to use mixtures of no more than three PBDEs containing four, six, and nine bromine atoms (in our case, congeners 79, 138, and 206), which, because of the significant difference in masses of characteristic ions and thermal decomposition of PBDEs inevitably increases the error of determination of the recovery In our work with MS/MS detection, the sensitivity coefficient of BDE-3 relative to BDE79 differed by 1.5–2 times on different days This effect is less pronounced for a magnetic sector instrument, but it is still necessary to repeat the injection of the calibration mixture regularly The distortion of the results of determination of the recovery may also occur because of the overlapping of the signals of matrix components on those of the detected substances, causing a local loss of sensitivity of the mass spectrometer This effect manifests itself as “subsidence” of the recoveries or their sharp increase in overlapping with the peak of the compound being determined In the isotopic dilution method, this does not lead to a distortion of the quantitative results (assuming no imposition on the recorded ions occurs), but with the general implementation of the internal standard method, the results of the quantitative analysis may be distorted several times No 2019 578 SHELEPCHIKOV et al Another source of error in determining the recoveries is caused by thermal destruction or other losses during chromatography This problem is most pronounced for DeBDE; in some cases, the residual components of the matrix lower this effect, because of which the calculated values of the recoveries can systematically exceed 100% The effect of this factor can be estimated, and correction factors can be introduced by comparing the change in the magnitude of the analytical signal in a series of isotope-labeled standards used to calculate extraction rates for pure mixtures and samples under study In addition to the loss of PBDEs during purification, the possibility of light-induced decomposition of high-brominated congeners is mentioned [11] It is also evident that MoBDE has rather high volatility, and special attention should be paid to the preconcentration of samples and their storage Because of the high uncertainty in estimating recovery rates in pilot experiments, we only present a semiquantitative estimate below Any routine analysis procedure is almost always a compromise between purification quality, cost, and recovery rate; the better these parameters are balanced, the more effective the procedure can be considered High recovery rates alone are not an essential requirement for routine analysis A more critical criterion is their stability when working with different matrices If one does not consider the extreme case when the residual amount of matrix components in the final extract is such that it is impossible to obtain mass chromatograms, the quality of purification is a somewhat subjective parameter In addition to the purely visual characteristic, i.e., the absence of staining or turbidity during the preconcentration of the purified extract to ~10 μL, we used the following criteria for assessing the quality of purification: —no distortion of chromatographic peaks compared with pure standards; —the absence of sharp degradation of the chromatographic column (constancy of retention times); —no over-peak or “humps” in the total ion current mass chromatograms RESULTS AND DISCUSSION Chemical destruction of impurities We estimated the stability of the existing isotope-labeled PBDEs by passing them through silica impregnated with sulfuric acid or potassium silicate at room temperature and at 85°C in a Dionex ASE 200 accelerated solvent extraction unit The mixtures were used that were made of silica activated at a temperature from 130 to 180°C, with the concentration of sulfuric acid from 30 to 44% The results did not show significant losses in using potassium silicate In the case of silica impregnated with sulfuric acid, PBDEs with a bromination degree of three or higher can be considered stable For low-brominated congeners, ambiguous results were obtained, indicating destruction at least in freshly prepared, highly active mixtures The current trend in analytical practice is the use of units for accelerated solvent extraction (ASE) not only for the extraction of various samples but also for the purification of extracts or for combining both stages [5, 23] The increased temperature during ASE increases the rate of chemical destruction of the matrix; however, ASE itself and the method of extraction practically exclude the possibility of effective adsorption purification A controversial point is also the efficiency of extraction with aliphatic solvents In our case, in analyzing fish meal with different variants of filling the extraction cell with silica impregnated with sulfuric acid, potassium silicate, and Florisil, the recoveries varied from 25 to ~100% with low purification quality This result could be predicted, since, based on the experience of determining PCBs, it can be argued that the vast majority of biological samples require at least a two-stage purification combining chemical destruction of the matrix and fractionation Fractionation For the purification of PCB and PCDD/PCDF by fractionation, it is often recommended to use a basic form of aluminum oxide We conducted an experiment with a column containing g of adsorbent (activated at 600°C for 16 h), with successive elution with 20 mL of hexane, 20 mL of a mixture of hexane–DCM (19 : 1, vol), and 50 mL of a mixture of hexane–DCM (2 : 3, vol) Under these conditions, PCBs enter the second fraction, except for the coplanar congeners, which, together with PCDD/PCDF, are eluted into the third fraction Mono- and dibrominated diphenyl ethers were lost; the remaining PBDEs were partitioned between the last two fractions, which shows significant differences in the physicochemical properties of PCBs and PBDEs Along with the basic form of aluminum oxide, the EPA methods provide for the possibility of using the acid form for determining PCB, PBDE, and PCDD/PCDF, but the developers of the methods indicate that it has less activity and offers smaller purification efficiency We did not find any examples of applied use of this adsorbent; however, we checked the possibility of its use In the experiment with the acidic form of aluminum oxide (4 g, activated at 130°C for 16 h) with successive elution with 20 mL of hexane, 20 mL of a mixture of hexane–DCM (19 : 1, vol), 20 mL of a mixture of hexane–DCM (3 : 1, vol), and 20 mL of a mixture of hexane–DCM (2 : 3, vol), PBDEs were partitioned between the second and third fractions without obvious loss, that is, this adsorbent can be considered as an option for additional purification of samples Significant differences in the properties of PBDEs and PCBs were also observed when using Florisil PR JOURNAL OF ANALYTICAL CHEMISTRY Vol 74 No 2019 A NEW METHOD FOR PURIFYING FAT-CONTAINING EXTRACTS This adsorbent is used in the determination of PCBs and various pesticides in cases where the use of aluminum oxide is impossible or not effective enough [24, 25] On a column with g of Florisil PR (activated at 180°C for 16 h) with the same elution sequence, PCDEs were distributed among all fractions, while PCBs was quantitatively eluted with hexane or a mixture with a small amount of DCM in hexane Although there is no reason to believe that a loss of analyte substances occurs, the use of Florisil PR seems to be unreasonable We tested conventional Florisil (activated at 180°C for 16 h and at 675°C for 24 h) In both cases, the columns were eluted successively with 30 mL of hexane, 25 mL of a mixture of hexane–DCM (3 : 1, vol), and 40 mL of DCM In the first system, the reference substances were partitioned between the first two factions; in the second system, the main part was in the hexane fraction, and only trace amounts were present in the DCM fraction Thus, this adsorbent is not suitable for the adsorption of PBDEs from solutions but can be used to remove other components of the matrix The last tested adsorbent was neutral alumina, the use of which is not recommended by the EPA methods for PCBs or PBDEs, but it is effective in determining polycyclic aromatic hydrocarbons In the first experiments (4 g, activated at 400°C for 16 h), there was no leakage of PBDEs during washing with hexane, and all reference substances were quantitatively eluted with 20 mL of a hexane–DCM mixture (4 : 1, vol); however, we later observed a loss of mono- and dibrominated diphenyl ethers When the activation temperature was decreased to 200°C, there was no loss, but the quality of purification deteriorated significantly Apparently, the loss of mono- and dibrominated diphenyl ethers is associated with the problem of desorption rather than with chemical transformations This hypothesis was confirmed by elution with methanol, when MoBDE and DiBDE were desorbed quantitatively However, this elution method has no practical significance, since methanol dissolves alumina, which precipitates from solution upon preconcentration Successive elution of the column with a mixture of hexane–methanol (19 : 1, vol) and mixtures of hexane–DCM–methanol in volume ratios of 17 : : 2, : : 1, and 10 : : (hereinafter, each fraction of 20 mL) did not achieve quantitative elution of MoBDE When eluted with pure toluene, zero recoveries were obtained for MoBDE and DiBDE; isopropanol eluted less than 50% Quantitative desorption of all PBDEs was achieved using a hexane–diethyl ether mixture (4 : 1, vol) A mixture containing two times less ether eluted at least 70% of low-brominated PBDEs, and the rest analytes were eluted quantitatively Two-stage purification Fractionation is an essential tool for fine purification, but nonselective adsorpJOURNAL OF ANALYTICAL CHEMISTRY Vol 74 579 tion barely enables the quantitative separation of trace components from the main components of the matrix It was noted earlier that a combination of fractionation and chemical purification is a more effective approach The use of silica impregnated with sulfuric acid is a conventional method of removing macro amounts of fat and many other extractable components of the matrix in the determination of PCBs and other substances that withstand such effects, but the tarring or saponification of organic substances under the effect of sulfuric acid leads to the “sticking” of the column, because of which the rate of passage of the solvent decreases In addition, the reaction products have uncontrollable adsorption properties, which complicates the work, leads to the loss of analyte substances, and increased solvent consumption We used a different approach, namely, the binding of the gross amount of fatty acids with potassium silicate and Florisil (magnesium silicate) Despite the similar nature of these substances, they are likely to bind different components of the fat matrix in different ways The highest efficiency was shown by a column containing a layer of Florisil between two layers of potassium silicate, separated by a layer of anhydrous sodium sulfate, and a layer of silica impregnated with sulfuric acid in the lower part of the column to remove residual components of the matrix Samples in mL of hexane were applied to a dry column; the substances to be determined were eluted with 50 mL of hexane The eluate was fractionated in a neutral alumina column The solution can be applied on a column with aluminum oxide without preconcentration or by evaporating it up to 2–3 mL in a rotary evaporator The purification procedure and the amount of adsorbents and solvents are shown in Fig The results of the determination of PBDEs in a sample of fish oil are given in Table Each purified extract was analyzed twice using the long and short chromatographic columns The recovery rates of all PBDE congeners in fish oil samples were not less than 83%; only in the blank experiment, lower values were obtained for BDE-3 and BDE-197 The results are characterized by excellent reproducibility in determining both the concentrations of native PBDEs and the recovery rates (Table 2), which demonstrates the reliability of the proposed method High recovery rates make it possible to decrease the consumption of solvents, which is 110 mL; this is more than four times smaller than when using the FMS Total-Prep automatic sample preparation unit Contamination of the blank sample is comparable to the values obtained with the use of the FMS unit, which is acceptable for moderately contaminated samples but yields distorted results in the case of low concentrations of PBDEs Problem of the blank experiment In the case of PBDEs and PCBs, the problem of the blank experiment cannot be solved entirely, unlike PCDD/PCDF, when, at least for congeners that make a significant contribution to the total equivalent toxicity [26], a No 2019 580 SHELEPCHIKOV et al ~0.5 g of fat in ~5 mL of hexane Flow direction Column to remove fat (internal diameter 14–15 mm) Na2SO4 ~ g K2SiO3 ~ g Na2SO4 ~ 1.5 g Florisil ~ 4.5 g K2SiO3 ~ g Na2SO4 ~ 1.5 g H2SO4 /SiO2(30%) ~ g Elution with 50 mL of hexane Without preconcentration or down to 1–2 mL Fractionation on neutral alumina (column with internal diameter mm) g, activated at 400°C for 16 h, conditioning with 15 mL of hexane Washing with 15 mL of hexane (waste) Elution with 20 mL of a diethyl ether– hexane mixture (1 : 4, vol) Preconcentration to ~10 µL or GLC−MS analysis Low- and medium-brominated congeners, SGE HT-8 or SGE-5 column 25–30 m (or equivalent) High- and mediumbrominated congeners, J&W DB-5ht column 10–15 m Fig Purification of fat-containing extracts in the determination of PBDEs with the degrees of bromination from to 10 vanishingly small level of the blank experiment can be achieved Sources of PBDEs and PCBs in the blank sample are of a universal nature; therefore, solutions to the problem are also similar We should consider all adsorbents, solvents, glassware, synthetic polymer materials, and even air in the laboratory potential sources of contamination The contribution of each source varies, depending on the qualification of the adsorbent or solvent, and may vary from batch to batch The contribution to the contamination of solvents of “pesticide grade” qualification or intended for the determination of PCDD/PCDF and PCBs is usually very low, but it is reasonable to minimize their consumption, and not JOURNAL OF ANALYTICAL CHEMISTRY Vol 74 No 2019 A NEW METHOD FOR PURIFYING FAT-CONTAINING EXTRACTS 581 Table Concentrations and recoveries (Rex) of polybrominated diphenyl ethers in the analysis of fish oil samples Sample Sample Sample RSD, % Blank sample Analyte c, pg/g BDE-3 BDE-15