Talanta 161 (2016) 547–553 Contents lists available at ScienceDirect Talanta http://d 0039-91 n Corr E-m journal homepage: www.elsevier.com/locate/talanta Effect of different precursors on generation of reference spectra for structural molecular background correction by solid sampling high- resolution continuum source graphite furnace atomic absorption spectrometry: Determination of antimony in cosmetics Ariane Isis Barros, Diego Victor de Babos, Edilene Cristina Ferreira n, José Anchieta Gomes Neto Univ. Estadual Paulista - Unesp, Departamento de Química Analítica, P.O. Box 355, 14800-060, Araraquara - SP, Brazil a r t i c l e i n f o Article history: Received 22 July 2016 Received in revised form 2 September 2016 Accepted 4 September 2016 Available online 5 September 2016 Keywords: SiO molecular spectrum Background correction Facial cosmetics Antimony x.doi.org/10.1016/j.talanta.2016.09.017 40/& 2016 Elsevier B.V. All rights reserved. esponding author. ail address: edilene@iq.unesp.br (E.C. Ferreira a b s t r a c t Different precursors were evaluated for the generation of reference spectra and correction of the back- ground caused by SiO molecules in the determination of Sb in facial cosmetics by high-resolution con- tinuum source graphite furnace atomic absorption spectrometry employing direct solid sample analysis. Zeolite and mica were the most effective precursors for background correction during Sb determination using the 217.581 nm and 231.147 nm lines. Full 23 factorial design and central composite design were used to optimize the atomizer temperature program. The optimum pyrolysis and atomization tem- peratures were 1500 and 2100 °C, respectively. A Pd(NO3)2/Mg(NO3)2 mixture was employed as the chemical modifier, and calibration was performed at 217.581 nm with aqueous standards containing Sb in the range 0.5–2.25 ng, resulting in a correlation coefficient of 0.9995 and a slope of 0.1548 s ng�1. The sample mass was in the range 0.15–0.25 mg. The accuracy of the method was determined by analysis of Montana Soil (II) certified reference material, together with addition/recovery tests. The Sb concentration found was in agreement with the certified value, at a 95% confidence level (paired t-test). Recoveries of Sb added to the samples were in the range 82–108%. The limit of quantification was 0.9 mg kg�1 and the relative standard deviation (n¼3) ranged from 0.5% to 7.1%. From thirteen analyzed samples, Sb was not detected in ten samples (blush, eye shadow and compact powder); three samples (two blush and one eye shadow) presented Sb concentration in the 9.1�14.5 mg kg�1 range. & 2016 Elsevier B.V. All rights reserved. 1. Introduction Despite the fact that hazardous metals are not acceptable in cosmetics, antimony-containing compounds may be present as pigments to give red, yellow, or blue colors to facial makeup products [1]. Antimony and its compounds can cause a number of human health effects, including respiratory and gastrointestinal diseases and contact dermatitis [1,2]. German and Canadian reg- ulatory agencies have established a concentration of 10 mg kg�1 as the maximum acceptable level of Sb in cosmetics [3,4]. Hence, appropriate methods for the determination of toxic metals in makeup products are needed in order to ensure product quality and the safety of cosmetics for users worldwide. The major components commonly present in the composition of facial makeup (including alumina, silica, titanium dioxide, mica, ). and inorganic pigments) are refractory, which makes analysis difficult. The presence of these refractory components necessitates extreme conditions for the dissolution of makeup to give a clear solution, and conventional procedures are based on dry ashing or wet digestion [5–7]. Disadvantages of these techniques are that they are laborious, time consuming, and susceptible to con- tamination and analyte losses. They also involve high energy consumption, the use of large quantities of hazardous chemical reagents, and the generation of substantial amounts of waste. As an alternative, direct solid sample (DSS) analysis is an en- vironmentally friendly method that avoids the lengthy pretreat- ments associated with conventional analysis of makeup samples [8,9]. Among spectrometric techniques for elemental analysis, high- resolution continuum source graphite furnace atomic absorption spectrometry (HR-CS GF AAS) hyphenated with SS is a versatile technique that offers several advantages. Solid samples can be measured directly without any previous treatment, sample throughput can be greatly increased, waste generation is low, and www.sciencedirect.com/science/journal/00399140 www.elsevier.com/locate/talanta http://dx.doi.org/10.1016/j.talanta.2016.09.017 http://dx.doi.org/10.1016/j.talanta.2016.09.017 http://dx.doi.org/10.1016/j.talanta.2016.09.017 http://crossmark.crossref.org/dialog/?doi=10.1016/j.talanta.2016.09.017&domain=pdf http://crossmark.crossref.org/dialog/?doi=10.1016/j.talanta.2016.09.017&domain=pdf http://crossmark.crossref.org/dialog/?doi=10.1016/j.talanta.2016.09.017&domain=pdf mailto:edilene@iq.unesp.br http://dx.doi.org/10.1016/j.talanta.2016.09.017 A.I. Barros et al. / Talanta 161 (2016) 547–553548 the method is relatively inexpensive. The technique enables elimination of the fine structured background caused by diatomic molecular species, using a least-squares background correction (LSBC) algorithm [10,11]. Nevertheless, the efficacy of LSBC de- pends on correct selection of the precursor used to generate a reference spectrum representing the structured molecular back- ground of the sample matrix [12,13]. In GFAAS, blanks, standards, and samples are analyzed using the same atomizer heating pro- gram, but the molecular background reference spectrum, which is typically generated from pure substances or mixtures, may not be entirely similar to the spectra of real samples. Concomitants pre- sent in the sample matrix can affect the kinetics of formation of diatomic molecules, hence altering the background absorption profile [12]. Therefore, the identification and use of suitable pre- cursors in HR-CS SS-GF AAS is not as straightforward as one might think. The determination of Sb by HR-CS SS-GF AAS has been de- scribed for dust [14], airborne particulate matter [15], and sedi- ment [16] matrices. All these complex samples contain Si as a major concomitant, but structured SiO backgrounds have only been reported for determination of Sb using the lines at 212.739 and 231.147 nm, with Ru [15] and Ir [16] as permanent modifiers, respectively. The interferences caused by SiO and PO in the de- termination of Sb in sediment were eliminated when the LSBC employed the combined molecular background reference spectra for pure SiO2 and NH4H2PO4 [16]. Preliminary tests in our la- boratories showed that the use of SiO2 alone was inadequate for pre-recording a suitable structured SiO background to be used as the reference spectrum in determination of Sb in makeup by HR- CS SS-GF AAS. Considering the above, this paper describes the evaluation of different precursors used to generate SiO background reference spectra, in order to develop a simple and accurate method for Sb determination in facial makeup samples by HR-CS SS-GF AAS. 2. Material and methods 2.1. Instrumentation An Analytik Jena contrAA 700 high-resolution continuum source atomic absorption spectrometer equipped with a xenon short-arc lamp (XBO 301, 300 W, GLE, Berlin, Germany) as a con- tinuum radiation source was used throughout this work. The spectrometer was fitted with a compact high-resolution mono- chromator, comprising a prism and an Echelle grating with a spectral bandwidth of less than 2 pm per pixel in the far ultra- violet range, together with a charge-coupled device (CCD) array detector. Atomization was performed using pyrolytic carbon- coated graphite tubes (without a dosing hole) and graphite plat- forms. The samples were weighed directly onto the platforms using a micro-balance (WZ2PW, Sartorius Göttingen, Germany) with precision of 0.001 mg, and were introduced into the atomi- zation compartment using an Analytik Jena SSA 600 automated solid sampling accessory. High purity argon (99.996%, White Martins, São Paulo, Brazil) was used as the purge and protective gas. 2.2. Reagents, analytical solutions, and samples The high purity water (resistivity 18.2 MΩ cm) used to prepare all the solutions was obtained from a Millipore Rios 5 s reverse osmosis system coupled to a Milli-Q Academic deionizer (Milli- pore, Bedford, MA, USA). Standard solutions containing 250 and 600 μg L�1 Sb were prepared daily by appropriate dilution of the 1000 mg L�1 Sb stock solution (Titrisols, Merck) in 0.1% (v/v) Suprapur HNO3 (Merck, Darmstadt, Germany). Different aliquots of 250 and 600 μg L�1 standard solutions were injected onto the platforms for calibra- tions using the Sb lines at 217.581 nm (0.5, 0.75, 1.0, 1.25, 1.5, and 2.0 ng Sb) and 231.147 nm (1.2, 2.4, 3.6 and 4.8 ng Sb), respectively. A modifier solution containing 1000 mg L�1 Pd(NO3)2 plus 500 mg L�1 Mg(NO3)2 was prepared by the appropriate dilution of 10 g L�1 Pd(NO3)2 and Mg(NO3)2 stock solutions (Merck, Darm- stadt, Germany)in 0.05% (m/v) Triton X-100 (Mallinckrodt Baker, Paris, KY, USA). Different precursors were evaluated for generation of the SiO(g) structured background: SiO2 (99%, 0.5–10 mm, Sigma-Aldrich, USA); TiO2 (99.8%, Sigma-Aldrich, USA); Mica Covasil 4.05 (K2O �3Al2O3 �6SiO2, Sensient Cosmetic Technologies, France); and CBV100 Zeolite Y (Na form, SiO2/Al2O3¼5, Zeolyst International, USA). Eye shadow, blush, and compact powder samples of various brands and in diverse colors were purchased locally in the city of Araraquara (São Paulo, Brazil). Makeup reference materials are unavailable, so the Montana Soil (II) certified reference material (CRM NIST 2711a) (Gaithersburg, MD, USA) was selected for the evaluation of accuracy, since the major components of its matrix are similar to those present in the composition of makeup. All plastic bottles and glassware materials were cleaned by soaking in 10% (v/v) HNO3 for at least 24 h. Before use, these materials were rinsed abundantly with deionized water. 2.3. Effect of precursors on background correction Preliminary tests showed that the Sb line at 217.581 nm was within the range of strong molecular absorption bands with sig- nificant fine structure, attributed to SiO, which needed to be cor- rected in order to avoid interferences. The feasibility of generating SiO molecular reference spectra suitable for background correction in the determination of Sb in cosmetics was evaluated using the following precursors: SiO2; mixture of SiO2 and TiO2 oxides with mass ratios of 1:1, 1:2, 1:3, 1:4, and 1:5; mica; and zeolite. These studies were carried out using a preliminary heating program under the conditions described elsewhere [16,17]. The optimum precursor obtained was then employed in further studies. 2.4. Optimization of the atomizer heating program An eye shadow sample was selected to optimize the main parameters of the atomizer heating program for Sb determinations by HR-CS SS-GF AAS in the presence of the 5.0 μg Pd (NO3)2þ2.5 μg Mg(NO3)2 as the chemical modifier. This optimi- zation was performed using a full (23) factorial design, considering the following factors: pyrolysis temperature, atomization tem- perature, and sample size. Considering these factors at high and low levels, coded as þ(1600 °C pyrolysis temperature; 2200 °C atomization temperature; 0.45�0.55 mg sample mass) and – (1400 °C pyrolysis temperature; 2000 °C atomization temperature; 0.15�0.25 mg sample mass), respectively, eight experiments were designed and performed in triplicate, as detailed in Table S1. The response monitored in these experiments was the nor- malized integrated absorbance at 217.581 nm, corrected for the background. The experiments were performed in random order to avoid systematic errors, and the experimental data were processed using Statgraphics Centurion XVI (v.16.1.15) software. It was found that the pyrolysis and atomization temperatures significantly influenced the normalized integrated absorbance, so a new experimental design was carried out to provide a better evaluation of these factors. A central composite design (CCD) was used, considering the following coded levels: high (þ) (1600 °C pyrolysis temperature; 2200 °C atomization temperature), low (-) A.I. Barros et al. / Talanta 161 (2016) 547–553 549 (1400 °C pyrolysis temperature; 2000 °C atomization tempera- ture), central (0) (1500 °C pyrolysis temperature; 2100 °C atomi- zation temperature),+ 2 (1640 °C pyrolysis temperature; 2240 °C atomization temperature) and − 2 (1360 °C pyrolysis tempera- ture; 1960 °C atomization temperature). The corresponding values of the levels of the factors and their combination are shown in Table S2. The experiments were performed in random order to avoid systematic errors, and the response monitored was the normalized integrated absorbance at 217.581 nm, corrected for the back- ground. The Statgraphics Centurion XVI (v.16.1.15) software was used for data processing. 2.5. Analytical procedure The possibility of using aqueous standards calibration for direct solid sample analysis was evaluated by means of multiple effects matrices, comparing the characteristic masses (m0) and the slopes of the analytical curves (in the 0–4.8 ng mass range) constructed for the standards in (i) 0.1% (v/v) HNO3 and (ii) the soil CRM. Accuracy was assessed by analyzing the soil CRM, and com- paring the found and certified values using the unpaired t-test. The Sb line at 231.147 nm (34% relative sensitivity) was used because the certified Sb value of the CRM (24.3871.66 mg kg�1) was higher than the upper limit of the linear response established for the line at 217.581 nm. It is worth highlighting that the structured background observed for the line at 231.147 nm was efficiently corrected by LSBC using the reference spectrum from zeolite. Accuracy was also evaluated by means of addition/recovery tests. Eye shadow, blush, compact powder, zeolite, and mica samples (0.2 mg) were spiked with 5.0 and 7.0 mL aliquots of 250 mg L�1 stock standard solution in order to obtain final con- centrations of 6.2 and 8.7 mg kg�1, respectively. The spiked sam- ples were transferred to the solid sample platform for analysis. The precision of the method was evaluated in terms of the relative standard deviation (RSD) obtained for three successive measurements of each sample. The limits of detection (LOD) and quantification (LOQ) were determined according to the IUPAC recommendations [18]: 3� SDblank/b (LOD), and 10� SDblank/b (LOQ), where SD is the standard deviation for ten blank measurements (using an empty platform) and b is the angular coefficient of the calibration curve. After optimizing the calibration conditions, the method was applied in determination of Sb in commercial makeup samples. Aliquots of aqueous standards were manually injected on to the SSA 600 platform using micropipettes. Sample masses in the range 0.15–0.25 mg were manually transferred to the graphite platforms, weighed, and introduced into the atomization compartment using the automated solid sampling accessory. Atomic absorption mea- surements were carried out at 217.581 and 231.147 nm for the samples and the CRM, respectively. The optimized graphite tube heating program is provided in Table 1. For the purposes of Table 1 Optimized heating program for determination of Sb in eye shadow, blush, and compact powder samples. Step Temperature (°C) Ramp (°C s�1) Hold time (s) Argon flow rate (L min�1) Drying 1 110 10 10 2.0 Drying 2 130 5 10 2.0 Pyrolysis 1500 50 30 2.0 Auto-zeroa 1500 0 5 0 Atomization 2100 2100 5 0 Cleaning 2650 500 5 2.0 a Step to ensure that the atomization starts without the presence of argon. comparison, the integrated absorbance values obtained with DSS were normalized to 1.0 mg of sample. All measurements were carried out in triplicate (n¼3), and the integrated absorbance was equivalent to 3 pixels. 3. Results and discussion 3.1. Evaluation of different precursors for background correction Preliminary experiments revealed a pronounced effect of spectral interference from the structured background on the de- termination of Sb in facial cosmetics and the CRM. These inter- ferences were attributed to strong SiO absorption bands that overlapped with the Sb line sat 217.581 and 231.147 nm. These interferences could be corrected automatically by the LSBC algo- rithm available in the HR-CS GFAAS software, using a spectrum representing the sample background as a reference. To this end, different precursors were evaluated for the generation of a spec- trumwith a profile close to the structural molecular background of the sample spectra. The correction procedure adopted here began with the acquisition of a spectrum for the following precursors: 0.2 mg zeolite, 0.2 mg mica, 0.2 mg SiO2, 0.1 mg SiO2þ0.1 mg TiO2, 0.1 mg SiO2þ0.2 mg TiO2, 0.1 mg SiO2þ0.3 mg TiO2, 0.1 mg SiO2þ0.4 mg TiO2, and 0.1 mg SiO2þ0.5 mg TiO2. The background of the makeup samples was attributed essen- tially to SiO, so the first experiments were focused on generation of a background from SiO2(s) alone. Fig. 1 shows time-resolved absorbance spectrum for Sb at 217.581 nm for an eye shadow sample, without correction (Fig. 1a) and after LSBC (Fig. 1c). It can be seen that the profiles of the SiO background generated from SiO2(s) and the sample were different, suggesting a substantial influence of the sample matrix on generation of SiO(g). An attempt was therefore made to determine the influence of matrix effects on the background structure. A facial cosmetic such as eye shadow contains around 5% TiO2 and 56% mica (KAl2Si3AlO10(OH,F)2) [19]. In order to evaluate the influence of sample composition on the SiO background structure, reference spectra were obtained from the SiO2:TiO2 mixtures at different ratios (1:1, 1:2, 1:3, 1:4, and 1:5 w/w) and from the mica and zeolite samples. Zeolite was used here because, like mica, it is an aluminum silicate. Fig. S1 shows the temporal profiles of the SiO backgrounds generated from SiO2 (s), the eye shadow sample, and the SiO2/TiO2 mixtures. It can be seen that the maximum transient absorbance of SiO(g) generated from SiO2 appeared faster than obtained using the sample. As the TiO2 in the mixture increased, the corresponding transient signals became longer. The times for the SiO(g) peak maxima were closely similar for the sample and SiO2:TiO2 at a ratio of 1:3 (w/w). A delay in the time for formation of SiO(g) was observed when zeolite and mica were used as precursors (Fig. S2). It is likely that the constituents TiO2 and Al2O3 acted as chemical modifiers, de- laying the formation of SiO(g). According to the literature, the thermal decomposition of Al2O3(s) begins before the decomposi- tion of SiO2(s), since energies of 501.6 and 799.6 kJ mol�1 are re- quired to dissociate the Al-O and Si-O bonds, respectively. The thermal decomposition products of Al2O3 (such as Al2O2(g), Al2O(g), and Al(g)) are oxidized to Al2O3(ad), Al2O2(ad), and Al2O(ad) due to the presence of the Mg(NO3)2 modifier [20,21], and may react with SiO(g) to produce Si(g) (Eq. 1). The Si(g) can then be oxidized to SiO(g) by oxygen. Hence, the time required for formation of SiO(g) from SiO2/Al2O3 is longer than observed in the presence of SiO2(s) alone. The TiO2 present in the mixed precursor may act similarly to Al2O3, increasing formation of the target SiO(g) molecule. ( )+ → + ( )( ) ( ) ( ) ( ) ( )SiO Al O Al O Al O Al O Si, , 1g ad ad ad ad g2 2 2 2 2 2 3 Fig. 1. Time-resolved absorbance spectrum, in the vicinity of the 217.581 nm Sb line, for (a) eye shadow sample (9.170.9 mg kg�1) without correction, (b) SiO generated from SiO2, and (c) eye shadow sample (9.170.1 mg kg�1) after correction with LSBC. A.I. Barros et al. / Talanta 161 (2016) 547–553550 Fig. S3 shows the SiO spectra generated from the following precursors: eye shadow sample, SiO2, SiO2/TiO2, mica, and zeolite. Similar SiO(g) band profiles were obtained in the spectra for the sample, SiO2/TiO2, mica, and zeolite. Subsequently, LSBC was ap- plied to the measurements of Sb in one eye shadow sample, using the reference spectra generated from SiO2, SiO2/TiO2, mica, and zeolite (Fig. 2). Use of the zeolite spectrum resulted in the best atomic absorption spectrum for Sb in the eye shadow sample, so zeolite was selected as the precursor in subsequent studies. 3.2. Optimization of pyrolysis and atomization temperatures and sample size The pyrolysis temperature, atomization temperature, and sample size are the factors that have the greatest influence in absorbance measurements by HR-CS SS-GF AAS, and were firstly evaluated at two levels using a full 23 factorial design. For each experimental condition, the integrated absorbance was measured, normalized, and corrected by LSBC using the zeolite reference spectrum. The experimental conditions and the corresponding responses (normalized integrated absorbance) were submitted to statistical analysis to estimate the significance of factors that in- fluenced the absorbance. The results are shown in the form of a Pareto chart (Fig. 3). Factors showing t42.12 (indicated by the bars crossing the vertical line) had a significant effect on the re- sponse (at a 95% confidence level). The pyrolysis temperature (tA) and the combined pyrolysis and atomization temperatures (tAB) had significant effects on the response evaluated. The negative value for tA indicated that the response could be increased by decreasing the pyrolysis temperature. The positive effect for in- teraction between the pyrolysis and atomization temperatures (tAB) indicated that these factors were dependent on each other Fig. 2. Atomic absorption spectra of eye shadow sample with LSBC based on background spectrum generation using: (a) SiO2, (b) SiO2:TiO2 (1:3, w/w), (c) mica, and (d) zeolite. tA tB tC tAB tBC tAC tABC +0.56 +1.09 -0.93 2.63 -0.56 -0.51 Estimated effect (absolute value) -3.39 Fig. 3. Pareto chart generated from the experimental conditions and responses of the 23 factorial design. tA: Student's t-value for pyrolysis temperature; tB: Stu- dent's t-value for atomization temperature; tC: Student's t-value for sample weight; tAB, tBC, tAC, and tABC: Student's t-values for combined effects. A.I. Barros et al. / Talanta 161 (2016) 547–553 551 and could be optimized together. Considering these findings, a new experiment was designed in order to find the optimum values for the pyrolysis and atomization temperatures. New experiments using central composite design (CCD) were run using a fixed sample mass of around 0.15�0.25 mg. Fig. S4 shows a contour chart constructed using the levels of the factors investigated and the corresponding normalized integrated absorbances (corrected by LSBC using the zeolite reference spectrum). The brown circle shows the region where the pyrolysis and atomization tempera- tures produced the best response, with the maximum indicated by the symbol “þ”. Hence, the pyrolysis and atomization tempera- tures selected for the heating program of the graphite furnace atomizer for Sb determination in facial cosmetics were 1500 and 2100 °C, respectively. The optimized heating program is provided in Table 1. 3.3. Analytical method: features and application Standard reference materials for facial cosmetics are not com- mercially available, while aqueous standards are readily accessible and enable simple calibration. The possibility of using calibration with aqueous standards for direct solid sample analysis of makeup samples was evaluated by comparing the slopes of analytical curves in the concentration range 0–4.8 ng Sb, constructed using standards in 0.1% (v/v) HNO3 (aqueous medium) or the soil CRM (solid medium). In both cases, the modifier was 5.0 μg Pd (NO3)2þ2.5 μg Mg(NO3)2 in 0.05% (m/v) Triton X-100. The slopes of the curves obtained were 0.144070.0020 and 0.150370.0044 s ng�1 using the aqueous and solid media, respectively, which were not significantly different at a 95% confidence level. The char- acteristic mass (m0) values calculated for the aqueous and solid media were 30.970.19 and 29.973.8 pg, respectively. These findings suggest that matrix effects were minimal, confirming the efficacy of the optimized heating program for use indirect solid sample analysis with calibration using aqueous standards. These results were close to m0 found in the literature (28 pg) for Sb at line 231.147 [16]. Table 2 Analytical figures of merit for Sb determination in facial cosmetic by HR-CS SS-GF AAS. Parameter Sb (217.581 nm) LOD (mg kg�1) 0.3 LOQ (mg kg�1) 0.9 Linear working range (ng) 0.5–2.5 Characteristic mass (pg) 27.270.6 r 0.9995 Table 3 Results (mean7SD) for Sb (mg kg�1) determined (n¼3) in spiked eye shadow sample and Montana Soil II CRM (NIST 2711a) without and with LSBC using dif- ferent precursors. Date between parentheses indicate recoveries in %. Precursor Sb7SD (Recovery) Eye shadown CRM NIST 2711ªnn Without LSBC 13.270.7 (151) 38.671.3 (162) tcalculated¼11.0 tcalculated¼19.7 SiO2 10.970.6 (124) 21.471.3 (90) tcalculated¼6.21 tcalculated¼3.19 SiO2:TiO2 1:3 (m/m) 9.470.4 (108) 20.470.3 (86) tcalculated¼2.99 tcalculated¼19.6 Mica 9.770.6 (111) 24.271.8 (102) tcalculated¼2.86 tcalculated¼0.38 Zeolite 9.870.5 (112) 23.270.8 (97) tcalculated¼3.74 tcalculated¼1.30 tcritical (α¼0.05; df¼2)¼4.30. n Eye shadow sample spiked with 8.7 mg kg�1 Sb. nn Certified value: 23.871.4 mg kg�1 Sb. Table 4 Recoveries (%) of Sb spiked in different samples. Values in parentheses indicate the RSD values. Sample Spike (mg kg�1) Recovery (%) Compact powder 1 - coral 6.2 82.572.7 (3.3) 8.7 90.174.0 (4.5) Blush 1 - pink 6.2 91.474.5 (4.9) 8.7 86.775.1 (5.9) Eye shadow 1 - purple 6.2 107.874.8 (4.5) 8.7 91.873.8 (4.2) Mica 6.2 94.170.5 (0.5) 8.7 97.676.4 (6.6) Zeolite 6.2 89.872.4 (2.7) 8.7 91.676.5 (7.1) Table 5 Results (mean7SD) for determination of Sb (n¼3) in blush and eye shadow samples by the proposed method. Sample Sb (mg kg�1) Blush 2 - orange 12.770.2 Blush 3 - pink 14.571.2 Eye shadow 2 - purple 9.170.9 A.I. Barros et al. / Talanta 161 (2016) 547–553552 The main figures of merit of the proposed method for Sb de- termination using the 217.581 nm absorption line were a dynamic working range of 0.50�2.25 ng Sb, correlation coeffi- cientZ0.9995, calibration sensitivity of 0.1518 s ng�1, detection limit of 0.3 mg kg�1, and quantification limit of 0.9 mg kg�1. The figures of merit are provided in Table 2. The effectiveness of background correction by LSBC using the zeolite reference spec- trum was evaluated by analyzing a soil CRM and an eye shadow sample, both spiked with 8.7 mg kg�1 Sb. For comparison, the CRM and the sample were also analyzed using LSBC with spectra for the SiO2, SiO2/TiO2, and mica precursors. The critical and cal- culated t-values obtained from unpaired t-test (Table 3) showed that more accurate results were obtained when mica and zeolite were employed as precursors for CRM, and SiO2/TiO2, mica, and zeolite were employed for spiked sample. Although accurate result was obtained when SiO2 precursor was employed for CRM, there was an overcorrection of the spectrum, as shown in Fig. 2a. These findings reinforced the selection of mica or zeolite as the precursor for generating the SiO reference spectrum in the background correction by LSBC. Accuracy studies were also carried out using recovery tests for compact powder, blush, eye shadow, mica, and zeolite samples spiked with 6.2 and 8.7 mg kg�1 Sb. Recoveries of Sb added to the samples varied within the range 82�107% (Table 4). Relative standard deviations were r7.1%, which are acceptable when di- rect solid sample analysis is employed [22]. The method was then applied in analysis of thirteen commer- cial samples of eye shadow, blush, and compact powder. From thirteen analyzed samples, Sb was not detected in ten samples (blush, eye shadow and compact powder); three samples (two blush and one eye shadow) presented Sb concentration in the 9.1�14.5 mg kg�1 range (Table 5). Only two samples presented Sb contents higher than the maximum acceptable level in cosmetics (10 mg kg�1) established by German and Canadian regulatory agencies [3,4]. 4. Conclusion The LSBC algorithm is a valuable tool for background correction in the HR-CS SS-GF AAS technique, enabling appropriate pre- cursors to be used for the generation of reference background spectra. An important finding was that the same chemical species could generate different background signals, as a function of time, because concomitant substances present in the samples acted as chemical modifiers and altered the formation time. Accurate de- termination of Sb in makeup samples could be achieved using precursors containing aluminum or titanium oxides that acted as chemical modifiers, retarding the SiO absorbance signals observed in the samples. Acknowledgements The authors would like to thank Fundação de Amparo à Pes- quisa do Estado de São Paulo (FAPESP) for financial support of this work (Grant #2014/12595-1). The authors are also grateful to Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) for fellowships to D.V.B. (33004030072P8), and to Con- selho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) for fellowships to A.I.B. (Grant #147166/2012-8) and M.A.B. (Grant #140934/2013-8) and a research fellowship to J.A.G.N (Grant #303255/2013-7). Appendix A. Supporting information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.talanta.2016.09. 017. References [1] B. Bocca, A. Pino, A. Alimonti, G. Forte, Toxic metals contained in cosmetics: a status report, Regul. Toxicol. Pharmacol. 68 (2014) 447–467, http://dx.doi.org/ http://dx.doi.org/10.1016/j.talanta.2016.09.017 http://dx.doi.org/10.1016/j.talanta.2016.09.017 http://dx.doi.org/10.1016/j.yrtph.2014.02.003 http://dx.doi.org/10.1016/j.yrtph.2014.02.003 http://dx.doi.org/10.1016/j.yrtph.2014.02.003 A.I. Barros et al. / Talanta 161 (2016) 547–553 553 10.1016/j.yrtph.2014.02.003. [2] M. De Boeck, M. Kirsch-Volders, D. Lison, Cobalt and antimony: Genotoxicity and carcinogenicity, Mutat. Res. Fundam. Mol. Mech. Mutagenesis 533 (2003) 135–152, http://dx.doi.org/10.1016/j.mrfmmm.2003.07.012. [3] Bundesinstitut Für Risikobewertung. Kosmetische mittel: BfR empfiehlt schwermetallgehalte über reinheitsanforderungen der ausgangsstoffe zu re- geln, Stellungnahme Nr. 025/2006, 2006. 〈http://www.bfr.bund.de/cm/343/〉, (accessed 05.03.13). [4] HC-SC, Health Canada-Santé Canada. Guidance on Heavy Metal impurities in Cosmetics. 〈http://www.hc-sc.gc.ca/cps-spc/pubs/indust/ heavy_metals-me taux_lourds/index-eng.php〉, (accessed 05.03.13). [5] EPA 3052, Microwave assisted acid digestion of siliceous and organically bade matrices, 1, 1996, pp. 1–20. 〈https://www.epa.gov/sites/production/files/2015- 12/documents/3052.pdf〉, (accessed 04.03.13). [6] M.G. Volpe, M. Nazzaro, R. Coppola, F. Rapuano, R.P. Aquino, Determination and assessments of selected heavy metals in eye shadow cosmetics from China, Italy, and USA, Microchem. J. 101 (2012) 65–69, http://dx.doi.org/ 10.1016/j.microc.2011.10.008. [7] N.M. Hepp, W.R. Mindak, J. Cheng, Determination of total lead in lipstick: Development and validation of a microwave-assisted digestion, inductively coupled plasma-mass spectrometric method, J. Cosmet. Sci. 60 (2009) 405–414, http://dx.doi.org/10.1111/j.1468-2494.2010.00577_2.x. [8] M. Resano, M. Aramendía, M. a Belarra, High-resolution continuum source graphite furnace atomic absorption spectrometry for direct analysis of solid samples and complex materials: a tutorial review, J. Anal. At. Spectrom. 29 (2014) 2229–2250, http://dx.doi.org/10.1039/C4JA00176A. [9] M.A. Belarra, M. Resano, F. Vanhaecke, L. Moens, Direct solid sampling with electrothermal vaporization/atomization: What for and how? Trends Anal. Chem. 21 (2002) 828–839, http://dx.doi.org/10.1016/S0165-9936(02)01206-2. [10] B. Welz, S. Mores, E. Carasek, M.G.R. Vale, M. Okruss, H. Becker-Ross, High- resolution continuum source atomic and molecular absorption spectrometry – a review, Appl. Spectrosc. Rev. 45 (2010) 327–354, http://dx.doi.org/10.1080/ 05704928.2010.483669. [11] B. Welz, D.L.G. Borges, F.G. Lepri, M.G.R. Vale, U. Heitmann, High-resolution continuum source electrothermal atomic absorption spectrometry – an ana- lytical and diagnostic tool for trace analysis, Spectrochim. Acta - Part B At. Spectrosc. 62 (2007) 873–883, http://dx.doi.org/10.1016/j.sab.2007.03.009. [12] A.R. Borges, E.M. Becker, L.L. François, A. De Jesus, M.G.R. Vale, B. Welz, M. B. Dessuy, J.B. De Andrade, Investigation of spectral interferences in the de- termination of lead in fertilizers and limestone samples using high-resolution continuum source graphite furnace atomic absorption spectrometry, Spectrochim. Acta - Part B At. Spectrosc. 101 (2014) 213–219, http://dx.doi.org/ 10.1016/j.sab.2014.08.040. [13] I.M. Dittert, D.L.G. Borges, B. Welz, A.J. Curtius, H. Becker-Ross, Determination of silver in geological samples using high-resolution continuum source elec- trothermal atomic absorption spectrometry and direct solid sampling, Mi- crochim. Acta 167 (2009) 21–26, http://dx.doi.org/10.1007/ s00604-009-0211-x. [14] A.A. Shaltout, B. Welz, I.N.B. Castilho, Determinations of Sb and Mo in Cairo's dust using high-resolution continuum source graphite furnace atomic ab- sorption spectrometry and direct solid sample analysis, Atmos. Environ. 81 (2013) 18–24, http://dx.doi.org/10.1016/j.atmosenv.2013.08.049. [15] R.G.O. Araujo, F. Vignola, I.N.B. Castilho, B. Welz, M.G.R. Vale, P. Smichowski, S. L.C. Ferreira, H. Becker-Ross, Determination of silver in airborne particulate matter collected on glass fiber filters using high-resolution continuum source graphite furnace atomic absorption spectrometry and direct solid sampling, Microchem. J. 109 (2013) 36–40, http://dx.doi.org/10.1016/j. microc.2012.05.009. [16] R.G.O. Araujo, B. Welz, F. Vignola, H. Becker-Ross, Correction of structured molecular background by means of high-resolution continuum source elec- trothermal atomic absorption spectrometry-Determination of antimony in sediment reference materials using direct solid sampling, Talanta 80 (2009) 846–852, http://dx.doi.org/10.1016/j.talanta.2009.08.004. [17] A. de Jesus, M.B. Dessuy, C.S. Huber, A.V. Zmozinski, Á.T. Duarte, M.G.R. Vale, J. B. Andrade, Determination of antimony in pet containers by direct analysis of solid samples using graphite furnace atomic absorption spectrometry and leaching studies, Microchem. J. 124 (2016) 222–227, http://dx.doi.org/10.1016/ j.microc.2015.08.016. [18] L.A. Currie, Limits for qualitative detection and quantitative determination. Application to radiochemistry, Anal. Chem. 40 (1968) 586–593, http://dx.doi. org/10.1021/ac60259a007. [19] Making Cosmetics, Pressed Powder Eye Shadow “Nude Glitter”, 2014. 〈http:// www.makingcosmetics.com/recipes/46-Press edPowderEyeShadowNudeGlitter.pdf〉 (accessed 04.11.14). [20] D.L. Styrus, D.A. Redfield, Mechanism of graphite furnace atomization of alu- minum by Molecular beam sampling mass Spectrometry, Anal. Chem. 59 (1987) 2891–2897, http://dx.doi.org/10.1021/ac00151a012. [21] B. Welz, M. Sperling, Atomic Absorption Spectrometry, Wiley-VCH, Weinheim, 1998. [22] U. Kurfust, Solid Sample Analysis - Direct and Slurry Sampling Using GF-AAS and ETV-ICP, Springer, Fulda, Germany, 1998. http://dx.doi.org/10.1016/j.yrtph.2014.02.003 http://dx.doi.org/10.1016/j.mrfmmm.2003.07.012 http://dx.doi.org/10.1016/j.mrfmmm.2003.07.012 http://dx.doi.org/10.1016/j.mrfmmm.2003.07.012 http://www.bfr.bund.de/cm/343/ http://www.hc-sc.gc.ca/cps-spc/pubs/indust/%20heavy_metals-metaux_lourds/index-eng.php http://www.hc-sc.gc.ca/cps-spc/pubs/indust/%20heavy_metals-metaux_lourds/index-eng.php http://https://www.epa.gov/sites/production/files/2015-12/documents/3052.pdf http://https://www.epa.gov/sites/production/files/2015-12/documents/3052.pdf http://dx.doi.org/10.1016/j.microc.2011.10.008 http://dx.doi.org/10.1016/j.microc.2011.10.008 http://dx.doi.org/10.1016/j.microc.2011.10.008 http://dx.doi.org/10.1016/j.microc.2011.10.008 http://dx.doi.org/10.1111/j.1468-2494.2010.00577_2.x http://dx.doi.org/10.1111/j.1468-2494.2010.00577_2.x http://dx.doi.org/10.1111/j.1468-2494.2010.00577_2.x http://dx.doi.org/10.1039/C4JA00176A http://dx.doi.org/10.1039/C4JA00176A http://dx.doi.org/10.1039/C4JA00176A http://dx.doi.org/10.1016/S0165-9936(02)01206-2 http://dx.doi.org/10.1016/S0165-9936(02)01206-2 http://dx.doi.org/10.1016/S0165-9936(02)01206-2 http://dx.doi.org/10.1080/05704928.2010.483669 http://dx.doi.org/10.1080/05704928.2010.483669 http://dx.doi.org/10.1080/05704928.2010.483669 http://dx.doi.org/10.1080/05704928.2010.483669 http://dx.doi.org/10.1016/j.sab.2007.03.009 http://dx.doi.org/10.1016/j.sab.2007.03.009 http://dx.doi.org/10.1016/j.sab.2007.03.009 http://dx.doi.org/10.1016/j.sab.2014.08.040 http://dx.doi.org/10.1016/j.sab.2014.08.040 http://dx.doi.org/10.1016/j.sab.2014.08.040 http://dx.doi.org/10.1016/j.sab.2014.08.040 http://dx.doi.org/10.1007/s00604-009-0211-x http://dx.doi.org/10.1007/s00604-009-0211-x http://dx.doi.org/10.1007/s00604-009-0211-x http://dx.doi.org/10.1007/s00604-009-0211-x http://dx.doi.org/10.1016/j.atmosenv.2013.08.049 http://dx.doi.org/10.1016/j.atmosenv.2013.08.049 http://dx.doi.org/10.1016/j.atmosenv.2013.08.049 http://dx.doi.org/10.1016/j.microc.2012.05.009 http://dx.doi.org/10.1016/j.microc.2012.05.009 http://dx.doi.org/10.1016/j.microc.2012.05.009 http://dx.doi.org/10.1016/j.microc.2012.05.009 http://dx.doi.org/10.1016/j.talanta.2009.08.004 http://dx.doi.org/10.1016/j.talanta.2009.08.004 http://dx.doi.org/10.1016/j.talanta.2009.08.004 http://dx.doi.org/10.1016/j.microc.2015.08.016 http://dx.doi.org/10.1016/j.microc.2015.08.016 http://dx.doi.org/10.1016/j.microc.2015.08.016 http://dx.doi.org/10.1016/j.microc.2015.08.016 http://dx.doi.org/10.1021/ac60259a007 http://dx.doi.org/10.1021/ac60259a007 http://dx.doi.org/10.1021/ac60259a007 http://dx.doi.org/10.1021/ac60259a007 http://www.makingcosmetics.com/recipes/46-PressedPowderEyeShadowNudeGlitter.pdf http://www.makingcosmetics.com/recipes/46-PressedPowderEyeShadowNudeGlitter.pdf http://www.makingcosmetics.com/recipes/46-PressedPowderEyeShadowNudeGlitter.pdf http://dx.doi.org/10.1021/ac00151a012 http://dx.doi.org/10.1021/ac00151a012 http://dx.doi.org/10.1021/ac00151a012 http://refhub.elsevier.com/S0039-9140(16)30686-5/sbref17 http://refhub.elsevier.com/S0039-9140(16)30686-5/sbref17 http://refhub.elsevier.com/S0039-9140(16)30686-5/sbref18 http://refhub.elsevier.com/S0039-9140(16)30686-5/sbref18 Effect of different precursors on generation of reference spectra for structural molecular background correction by solid... Introduction Material and methods Instrumentation Reagents, analytical solutions, and samples Effect of precursors on background correction Optimization of the atomizer heating program Analytical procedure Results and discussion Evaluation of different precursors for background correction Optimization of pyrolysis and atomization temperatures and sample size Analytical method: features and application Conclusion Acknowledgements Supporting information References