Total Reflection X-Ray Fluorescence Spectroscopy (TXRF) Method Validation: Determination of Heavy Metals in Dietary Supplements

Faculty of Nursing and Nutriology, Chihuahua Autonomous University, Chihuahua 31125, Mexico Faculty of Chemical Sciences, Chihuahua Autonomous University, Chihuahua 31125, Mexico Environment and Energy Department, Advances Materials Research Center (CIMAV) S.C, Miguel de Cervantes 120, Chihuahua 31109, Mexico Laboratory of Environmental Analytical Chemistry, University of the Balearic Islands, Cra. Valldemossa km 7’5 Palma de Mallorca, Palma, Spain


Introduction
Dietary supplements are products that contain a "dietary ingredient" which includes vitamins, minerals, and amino acids, as well as other substances that can be used to supplement the diet [1]. ey can be prepared from decoctions from fresh or dried crude herbs from algae or plants, but are usually made into different presentations, including powders, tablets, capsules, energy bars, and liquids [2]. e consumers believe that since these products contain ingredients that come from plants, they are safe and without any adverse effect [3]. However, many ingredients are derived from natural products, which can be toxic due to the presence of heavy metals.
Dietary supplements could contain heavy metals due two fundamental reasons. (a) eir distribution is not uniform, such that some soils may contain higher concentrations derived from natural components or geological sources as well as from human activities or anthropogenic causes. Taken up by plants, heavy metals may enter the food chain in significant amounts. Hence, people could be at risk of adverse health effects from consuming dietary supplements made with plants grown in soils containing elevated metal concentrations [4]. (b) Contamination may occur during the production processes (manufacturing, handling, storage, processing, or distribution). Heavy metals such as Pb, As, Hg, and Cr are most commonly the subject of attention in manufacturing dietary supplements [4,5]. e botanical ingredients in dietary supplements may contain elevated concentrations of As, contributing to the overall exposure to this element [5]. Arsenic exposure affects all organ systems, including the cardiovascular, dermatologic, nervous, renal, gastrointestinal, and respiratory systems. Also, it is now well accepted that exposure to high levels is associated with an increased risk of various types of cancers, most notably, skin, urinary bladder, and lung [6].
Lead is poisonous to humans and can affect people of any age or health status. Lead is especially harmful to vulnerable populations, including infants, young children, pregnant women, and their fetuses, and others with chronic health conditions [7]. On the other hand, Pb poisoning mainly affects the nervous system (headache, decreased attention span, irritability, loss of memory, and dullness are the early symptoms) [8].
Chromium is a mineral nutrient that is essential for the proper functioning of many systems in the human body. Some studies have observed that Cr (III) could improve insulin action, but the relation between Cr and insulin is not well known.
is element also could participate in lipid metabolism and could have some effect on body composition [9]. e major problem with some dietary supplements is that their labels do not show the concentration of either the oxidation estate of Cr. e oxidation state should consider when evaluating the toxicity of Cr compounds. Due to this, metal may form complexes with peptides, proteins, and DNA, resulting in DNA-protein cross-links, DNA strand breaks, and alterations in cellular signaling pathways, which may contribute to the toxicity and carcinogenicity of chromium compounds. Meanwhile, the products of metabolic reduction of Cr (VI) (free radicals and Cr IV and V), and the newly generated Cr (III), are thought to be in part responsible for the carcinogenic effects seen in human and animal studies [10].
Abdulla et al. presented reports of mercury contamination of dietary supplements [11]. Mercury interferes with many enzymatic reactions, causes problems in migration and division of cells, and is responsible for cell damage, or even death [12].
Heavy metals, even at trace levels, are very harmful to human beings. us, their analytical determination is a very critical and essential topic, which has attracted considerable attention [7], particularly due to the actual amount of Pb, As, Cr, and Hg contained in these dietary supplements, which could exceed the acceptable standard. e standards defined by different international organizations presented discrepancies in what concerns the listing of elements and compliance limits for each element.
e World Health Organization (WHO) guidelines proposed Pb permissible levels (10 μg·g −1 ) in medicinal raw plant materials and dietary supplements [13], whereas, for plant-based food supplements, the European Commission set limits of 3 μg·g −1 and 0.1 μg·g −1 for Pb and Hg, respectively [14]. e United States Pharmacopeia (USP) recommended permissible limits (in μg·g −1 ) for Pb (0.5), As (0.15), and Hg (1.5) in dietary supplements [15,16]. Besides, the growing interest, popularity, and the lack of quantitative and qualitative research on the composition of supplements before being place into the market are important topics to attend [17]. e most common techniques used for the elemental analysis in dietary supplements include inductively coupled plasma optical emission spectrometry (ICP-OES) [18], inductively coupled plasma mass spectrometry (ICP-MS) [3,19,20], wavelength dispersive-x-ray fluorescence spectrometry (WD-XRF) [16], scanning electron microscopyenergy dispersive x-ray spectrometry (SEM-EDX), and ion chromatography (IC) [21]. However, to our knowledge, there is not any publication about the validation of an analytical method for heavy metals determination in dietary supplements by Total Reflection X-ray Fluorescence Spectroscopy (TXRF).
TXRF is a well-established analytical technique for multielement determination in various sample types, especially liquids and powdered microsamples. e x-ray emitted in the TXRF is characteristic of each individual element and its intensities are proportional to their concentrations in the sample [22]. e incoming radiation is incident on the sample at less than the critical angle and is totally reflected; this is one difference with the XRF technique. Advantages include the element and its concentration being unaffected by matrix effects, its low sensitivity (ppb), requiring small amounts of sample (μg or μL), the use of small quantities of reagents, and not utilizing argon or nitrogen gases. e low background levels result in improved LODs and even 2 pg may be detected for a variety of elements with a counting time of 1,000 s [23][24][25]. Actually, the development and commercialization of TXRF instrumentation, which offers simple operations with a low-cost compact design, have promoted its application in many different fields [24].
In view of these premises, the aim of this work was the validation and application of a simple and accurate methodology to determine the content of Pb, As, Cr, and Hg in dietary supplements by TXRF. Also, there were two research purposes, the first being to contribute to the risk assessment analysis regarding the consumption of dietary supplements and the second being to propose the TXRF technique as a powerful tool for quality standardization and control of both raw material and finished product.
Individual single-element stock solutions containing 1000 mg·L −1 Pb, As, Cr, Hg, and Ga were used as calibrations standards. Gallium was employed as an internal standard (IS) to improve the precision of quantitative analysis. For each element, six different known concentrations were prepared into 10 mL volumetric flasks and filled to the mark with 0.31 mol·L −1 of HNO 3 y 10 μg·L −1 of Ga (IS) solution. For Hg analysis, was utilized a dilute solution which contains 0.01 mol·L −1 of HNO 3 , 0.01 mol·L −1 of EDTA, and 10 μg·L −1 of Ga. All glassware was kept overnight in HNO 3 (10 % v/v) and rinsed several times with Millipore-Q water before its use to avoid cross-contamination.

Dietary Supplement Samples' Pretreatment.
Different mixtures have been reported for the digestion procedure of dietary supplements. e most common compounds were the concentrated HNO 3 [19], mixtures of HNO 3 + HCl, and concentrated HNO 3 + H 2 O 2 [3,20]. e present work utilized a digestion procedure using concentrated HNO 3 + HCl + H 2 O 2 for the determination of Pb, As, Cr, and Hg in six dietary supplements [2,24].
Four dietary supplements, fat burner (A-01), laxative (A-02), tejocote root (A-03), and Omega 6, 12 oil (A-04), were randomly collected from local Mexican suppliers. e selection criterion was based on high consumption and popularity in the treatment of chronic diseases (diabetes, high blood pressure, cancer, and obesity). e composition of all dietary supplements includes organic (carbohydrates, proteins, lipids, and vitamins) and inorganic compounds (Zn, Cu, Fe, Ca, among others). For confidentiality reasons, the studied products shall not be identified. e digestion procedure consisted of weighing approximately 0.020-0.050 g of each sample in polypropylene tubes. en, 5 mL of H 2 O and 1 mL nitric acid concentrated solutions were added into glass testing tubes. Reagents and sample mixtures were allowed to stand for 15 minutes. Subsequently, 100 μL of H 2 O 2 , 485 μL of concentrated HNO 3 , and 100 μL of concentrated HCl were incorporated into tubes, which were covered with their respective stopper. A vortex was used to shake the samples in the testing tubes, which were then placed into a bathwater at 80°C for eight hours. After acid digestion, samples were filtered. Finally, samples were prepared into 25 mL volumetric flasks and filled to the mark with 0.31 mol·L −1 of HNO 3 and 10 μg·L −1 of Ga (IS) solution. For Hg analysis, a dilute solution was utilized which contains 0.01 mol·L −1 of HNO 3 , 0.01 mol·L −1 of EDTA, and 10 μg·L −1 of Ga. Blank samples were made to identify any possible source of contamination. Analyses were carried out in triplicate, and analytical blanks were prepared by following the same procedure used for the samples.

Analysis of Certified Reference Material.
e USP mentioned some names of plants to develop standard mixtures that can be used in validation studies [15].
However, dietary supplements are usually made into different presentations, including powders, tablets, capsules, energy bars, liquids, and oils, among others [2], which implies a wide variety of matrices.
is is because it is difficult to find certified reference materials, with the validation purpose. However, a BCR-610 (groundwater, CRM) was analyzed, adjusting the HNO 3 concentration to 0.30 mol L −1 and added Ga as IS, without any prior digestion procedure. Samples were examined in triplicate by the TXRF_1 method.

Intercomparison Studies between TXRF and ICP-OES.
All chemicals utilized were of analytical reagent grade. Hydrogen peroxide (H 2 O 2 , 30% w/w) and nitric acid (65% w/w) were obtained from Scharlau (Barcelona, Spain) as well as standard solutions of As, Cr, and Pb (1000 mg·L −1 in 2% of HNO 3 ). Two samples (A-05 and A-06) were acquired from local Spain suppliers and employed for the intercomparison studies between TXRF and ICP-OES methods. e purpose was an estimate of the accuracy in the measurements of both methods. Student's t-test was utilized to find significant differences in the metal determinations.
According to its label, the A-05 sample is a dietary supplement utilized for the organism detoxification, while the A-06 (Fat Burner) is used to increase acutely fat metabolism and for the loss of weight. Both samples were weighed (0.5 g approximately) into a microwave Teflon vessel, to which 1 mL of H 2 O 2 , 1 mL of HCl, and 3 mL of HNO 3 were added. After digestion, they diluted with 25 mL of water and analyzed by ICP-OES (Optimal 5300 DV PerkinElmer Inc.). In the TXRF method, A-05 and A-06 samples were pretreated by following the procedure indicated in Section 2.2.

Instrumentation and Experimental Procedure.
e TXRF measurements were performed with a commercial S2 PICOFOX TXRF spectrometer (Bruker AXS Microanalysis GmbH, Berlin, Germany), equipped with either a molybdenum (Mo) x-ray tube operated at 600 μA and 50 kV. e automatic sample changer (holder) allows up to 25 sample carriers to be loaded and analyzed. e software used in the instrument control, data collection, and data analysis was Spectra PICOFOX ® 7. e spectral lines utilized: L -α 10.5 KeV, K −α 5.4 KeV, L −α 9.2 KeV, K −α 10.5 KeV y, and L −α 10 KeV, for Pb, Cr, Ga, As, and Hg, respectively. Sample carriers and glassware were subjected to a strict cleaning protocol. e experimental procedure applied was that reported by Beltrán and Cols. [24].

Design of the Experiments.
A response surface methodology involves several experimental techniques for the evaluation of the relationship between a group of controlled experimental factors and measured responses based on one or more criteria. Besides, it can be used to estimate the effect of individual parameters, the interaction of variables, and the optimum conditions for responses [17]. A central composite design (CCD) was used to investigate the impact of the selected parameters on the TXRF measurements. e HNO 3 concentration (mol·L −1 ), EDTA concentration (mol·L −1 ), and sample-reading time (s) were selected as the independent variables, while analytical signals (Pb, As, Cr, and Hg) were considered as the response (dependent variables). e independent variables vary between the lowest level of-l, center level 0, and the highest level of +l. en, their studied ranges were as follows; HNO 3 concentration (0.01-1 mol·L −1 ), EDTA concentration (0 -0.1 mol L −1 ), and sample-reading time (100 -500 s). e range of independent variables was based on preliminary experiments and determined literature [24]. e experimental design points consist of the 2 3 factorial points, six axial points, and six central points. Center points were used to determine the reproducibility of the data and the experimental error [17]. Results were examined by the software Minitab ® 17. Twenty experiments were performed to establish the optimal conditions utilizing the minimum reagents concentration and lower reading time.

Figures of Merit. Figures of merit considered in this
work were (the linearity and range, limit of detection, limit of quantification, accuracy, and precision) according to the recommendations of IUPAC [26].
Calibration curves were prepared for each of the metals and running a range of concentration between 2 and 100 μg·L −1 (Pb, As, and Cr), and for Hg analysis in a range 5-100 μg·L −1 . It plotted on the y-axis (analyte peak area/Ga peak area) against the corresponding ones (x-axis). e concentration of the analyzed samples was estimated by the assessment of the calibration curve data corresponding to each element: m (the slope of the calibration curve) and b (interception with the y-axis).
e LODs and LOQs were calculated as 3 s/m and 10 s/m, respectively, where s is the standard deviation of the analytical signal of ten blanks [26]. In addition, regression equations, determination coefficient (R 2 ), analysis frequency (h −1 ) were estimated. e precision was considered in terms of the intra-day repeatability and intra-day reproducibility by the use of 20 μg·L −1 concentration level for all analytes and evaluated as the percentage of relative standard deviation (% RSD).

Analysis of Real and Spiked Samples.
e accuracy of the TXRF_DS_1 and TXRF_DS_2 methods were evaluated by the recovery percentage of heavy metals by known amount added to selected dietary supplements. Samples (A-01, A-02, A-03, and A-04) were spiked with a standard solution of 20 μg·L −1 Pb, As, Cr and Hg during the digested process. ree replicates of each spiked samples were analyzed by TXRF under optimum conditions. Table 1 shows the experimental design with 20 experiments with their analytical responses. e experiments were run randomly. e ANOVA table showed a low fit for the tested models (Lineal, lineal + interactions, lineal + squares, and full quadratic).

Optimization of the TXRF Variables.
us, such as alternative, the optimal values of the variables that affected the proposed TXRF methods were obtained utilizing the desirability function. e function can be maximized by using a random starting point and then proceeding along the steepest slope up to a maximum. Starting from several points in the design space increases the chance of achieving the "best" maximum among all possible maxima of the function [27].
is mathematical model transforms each response into an individual desirability (d i ) value, coded from 0 (undesired response) to 1 (desired response). In this analysis, (d i � 1) corresponds to a desired response (maximum analytical signal for all elements), while (d i � 0) represents a minimum analytical signal. Finally, the individual desirability of each analyte was combined into a single response, which means the overall desirability (CD) through the geometric mean [16]. e best experimental conditions were chosen according to the higher composite desirability (CD). e analysis of data obtained in the experiment design included the four dependent variables and their response. In the first analysis, the CD value was low (0.73). Probably, due to the Hg response had a different behavior than the other elements, that is the "EDTA concentration" variable was not significant for Pb, As and Cr responses, but it was for Hg. erefore, a second analysis was carried out, excluding Hg data. In this case, the best local maximum was found to be at the sample-reading time of 500 s, 0.31 mol L −1 HNO 3 , 0 mol L −1 EDTA, while the CD value was 0.9802. However, looking for a compromise between the sensitivity of the technique and the lamp lifetime, the variable "samplereading time" was finally set to 401 s, without the desirability value (0.9170) being affected (Figure 1). e desirability of 0.9170 indicates that the estimated function may express the desired conditions and the experimental model. Under these conditions, the method was called TXRF_DS_1. e individual analysis for Hg analytical response showed that the best local maximum was found to be at the sample-reading time of 431 s, HNO 3 concentration of 0.01 mol·L −1 , EDTA concentration of 0.01 mol·L −1 , and CD value � 1. In the same way, the sample-reading time variable was established in 401 s and the CD value was not affected (Figure 2). e Hg analysis method was called TXRF_DS_2. Merit. Under the optimal conditions, the analytical performance of the proposed procedure was evaluating. Four calibration curves were built. Table 2 shows the linear calibration ranges and their respective determination coefficient (R 2 ), LODs, LOQs, repeatability, and reproducibility for all analyses using TXRF_DS_1 and TXRF_DS_2 methods. e calibration plots are depicted in Figure 3. e limit of detection (LOD) is the lowest concentration that can be detected, but not necessarily quantified. Detection limits for Pb, As, Cr, and Hg were lower than LODs obtained with an ICP-OES method for the quantitative analysis of toxic elements in some dietary supplements and diet products [18]. However, the excellent analytical sensitivity that ICP-MS allows is indisputable. Table 3 shows other methods that had better LODs than those achieved in proposed TXRF methods [3,19]. Nevertheless, these methods utilized a larger amount of samples and reagents for digestion procedures in comparison with TXRF_DS_1 and TXRF_DS_2 methods. Besides, our methods use a final dilution of samples between 10 and 25 mL (4-fold lower than other proposed methods). In addition, a sample volume of 10 μL was required, which contributes to the minimization of residues. e TXRF neither requires an argon gas nor hydride generation system. Moreover, all  LODs were lower than levels suggested by international organizations (WHO and US Pharmacopeia) [13,15]. e LOQs were calculated and produced a peak with ten times the signal-noise ratio for Pb, As, Cr, and Hg, respectively, and were in the range of 1.35-2.50 μg·L −1 .

Figures of
e precision under intraday repeatability and intraday reproducibility was assessed as the relative standard deviation (% RSD). e intraday repeatability was calculated based on five consecutive measurements of a standard solution (20 μg·L −1 ) using the same measurement procedure and the same operating conditions. e intraday reproducibility can be obtained with stated precision by five consecutive measurements of (20 μg·L −1 ) standard solution in a different analysis day. For all elements retained in the analysis, % RSD values are lower than 20%, as recommended by the US Pharmacopeia [28]. Table 2 shows figures of merit for the proposed methods. e calibration curves achieved statistically satisfactory results (R 2 > 0.999). e analysis frequency was nine samples per hour.

Analysis of Certified Reference Material.
To complete the validation of the TXRF_DS_1 method, a BCR-610 certified reference material was analyzed for Pb and As. e results   Journal of Chemistry were obtained by three replicates (n � 3) and expressed as the means and their standard deviation. e t-test for the comparison of means revealed that there were no significant differences at a 95% confidence level between the values obtained (7.85 ± 0.04 μg·L −1 and 10.79 ± 0.40 μg·L −1 for Pb and As, respectively) and the certified values (7.78 ± 0.13 μg·L −1 for Pb and 10.40 ± 0.25 μg·L −1 for As).

Intercomparison Studies between TXRF and ICP-OES.
Two dietary supplements were analyzed in the intercomparison study between TXRF and ICP-OES methods ( Table 4). Student's t-tests were performed to compare the experimental data obtained between the TXRF-DS_1 method and ICP-OES method. Sample pretreatment was carried out according to Section 2.2.2. First, in the A-05 sample, the Pb concentrations were in agreement in both methods; in the case of Cr and As determinations, it was not possible to establish a comparison, because some values were below ICP-OES LOQ. e results found for Pb and Cr determinations in the A-06 sample showed that there were no significant differences at a confidence level of 95% for n � 3. However, As was not detected by ICP-OES. e results demonstrated that the accuracy of the TXRF_DS_1 method is acceptable.

Analysis of Real and Spiked Samples.
e developed procedures were applied for the determination of Pb, As, Cr, and Hg in real and spiked dietary supplements.

Journal of Chemistry
Sample pretreatment was carried out according to Section 2.2. Concentrations of the target analytes in samples were obtained by using calibration curves and are shown in Table 5. e obtained recoveries of Pb, As, and Cr in A-01 and A-02 were ranging from 91% to 105%. e Hg concentrations in these samples were less than the proposed LOD. ese samples were analyzed by the TXRF_DS_1 method. e analysis recoveries of Hg were probed once more in the two samples A-03 and A-04 by the method TXRF_DS_2. e high recovery values indicated the absence of analyte loss during the sample preparation step and that sensitivity was not influenced by the dietary supplements matrix.
ese samples were analyzed with the TXRF_DS_2 method, exclusively for Hg. e results show that methods TXRF_DS_1 and TXRF_DS_2 are suitable options to quantify without any problem to heavy metals at trace levels, even below the maximum permissible limits imposed by international standards [13][14][15][16]. e TXRF technique is presented as a useful tool for monitoring heavy metals in food supplements, through which the quality of these products can be guaranteed before going to market and consumers ingest them.

Conclusions
e scientific and regulatory challenges in terms of the quality, safety, and efficacy of dietary supplements are common to all countries as the marketplace for them becomes increasingly global. In this study, two methods for quantitative analysis for the determination of toxic metals (Pb, As, Cr, and Hg) were developed, validated, and applied for the analysis of dietary supplements samples. It utilized a central composite design and desirability function for the method validation. Results obtained as linearity, accuracy, precision, LODs, and LOQs were within satisfactory borders, because we recommend TXRF_DS_1 and TXRF_DS_2 methods for heavy metal determinations in dietary supplements at trace level. No significant differences were found in the inter-comparison studies between TXRF and ICP-OES methods. TXRF methods improve the economy of laboratories in the research centers, universities, and industry. As they are environmentally friendly since the consumption of samples and reagents is minimum, they do not use toxic reagents, and it is not necessary to use carrier gases. ese characteristics allow their use in toxicological, environmental, biological, and food studies. Finally, this research proposes two methods for the standardization and control of the quality of both the raw material and the dietary supplement finished product before human consumption.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
All authors declare that they have no conflicts of interest.