Journal of Food Composition and Analysis 125 (2024) 105765

Contents lists available at ScienceDirect

Journal of Food Composition and Analysis

journal homepage: www.elsevier.com/locate/jfca

Sensitive method for total selenium determination in garlic and nuts by

UV–photochemical vapor generation coupled to atomic absorption
spectrometry using mild conditions
Jaime Palatnik a, b, Agustín Londonio a, b, Ezequiel Morzán a, Rodolfo Wuilloud c, d,
Patricia Smichowski a, c, *
a
Comisión Nacional de Energía Atómica, Gerencia Química, Av. Gral Paz 1499, 1650KNA–San Martín, Buenos Aires, Argentina
b
Instituto de Investigación e Ingeniería Ambiental (3iA), Universidad de San Martín, Martín de Irigoyen 3100, 1650–San Martín, Buenos Aires, Argentina
c
Consejo Nacional de Investigaciones Científicas y Técnicas, Av. Godoy Cruz 2290, C1425FQB, Buenos Aires, Argentina
d
Laboratory of Analytical Chemistry for Research and Development (QUIANID), Facultad de Ciencias Exactas y Naturales, Universidad Nacional de Cuyo, Instituto
Interdisciplinario de Ciencias Básicas (ICB), UNCUYO–CONICET, Padre J. Contreras 1300, 5500, Mendoza, Argentina

A R T I C L E I N F O A B S T R A C T

Keywords: Photochemical vapor generation (PVG) coupled on–line to atomic absorption spectrometry using a flow injection
Photochemical vapor generation mode was investigated for the determination of total Se in garlic and nut samples. In the presence of low mo­
Selenium lecular weight organic acid solutions, Se(IV) is converted by UV irradiation to volatile species which are
Photochemical reagents
transported to the heated quartz tube atomizer. Formic and acetic acids were tested as photochemical reagents
Chemometry
Atomic absorption spectrometry
and a mixture of 1.5 % (v/v) formic and 5 % (v/v) acetic acid was the best alternative. Central composite design
Garlic and nut was employed for parameters optimization. The effect of potential interfering elements such as hydride–forming
elements, transition metals and nitric acid was investigated to ascertain possible effects of the matrix on Se(IV)
signal. Under optimized conditions a detection limit (LOD) of 33 ng g–1 (3σ) was reached. Samples of garlic and
nuts purchased at a local market were analyzed and Se concentrations spanned the range 0.36–248 µg g–1. These
results were compared with ICP–MS measurements utilizing an elliptical joint confidence region (EJCR) with
bilinear least squares analysis, and good results were obtained.

1. Introduction (SeMet) is the major species in food. It is found in cereals, grains, and
soybeans. On the other hand, methylselenocysteine (MeSeCys) is mainly
Selenium is an essential trace element that is naturally present in detected in Se-enriched garlic, onions, and wild leeks.
different foods and added to others, and also Se is present in some di­ Chemical vapor generation (CVG) is a well–known, efficient and
etary supplements. It is a constituent of more than two dozen seleno­ consolidated sample introduction technique for the determination of a
proteins that play critical roles in reproduction, thyroid hormone variety of elements at trace and ultra–trace levels. Without any doubts,
metabolism, DNA synthesis, and protection from oxidative damage and hydride generation (HG) based on the chemical reduction of the analyte
infection (Sunde, 2012). This metalloid is a very important and inter­ with a reductant such as tetrahydroborate under acidic conditions is the
esting trace element in terms of human biology and health being food most popular and well established CVG technique for trace elements of
the principal route of Se to enter the body. A special characteristic of Se IV− VI main groups (e.g., As, Bi, Ge, Pb, Sb, Se, Sn, and Te), Hg, and Cd.
is that it can be essential as well as toxic for humans having a narrow The HG of a variety of transition metals was also reported (Marrero and
range between deficiency and toxicity (Dumont et al., 2006). This dual Smichowski, 2002; Pohl et al., 2014; Sturgeon et al., 1996).
behavior and other characteristics have led to Se becoming a permanent Nevertheless, in spite of the important advantages of HG for sample
topic of study. In this context, it is of prime importance to develop introduction, it is worth mentioning that solutions of NaBH4 are quite
methodologies for its determination at the levels that Se can be found in unstable, and that the large amounts of hydrogen generated during HG
food samples. As regard Se species in foodstuffs, selenomethionine process produces instability of the atomization source, especially when

* Corresponding author at: Comisión Nacional de Energía Atómica, Gerencia Química, Av. Gral Paz 1499, 1650KNA–San Martín, Buenos Aires, Argentina.
E-mail addresses: smichows@cnea.gov.ar, patricia.smichowski@gmail.com (P. Smichowski).

https://doi.org/10.1016/j.jfca.2023.105765
Received 26 June 2023; Received in revised form 11 October 2023; Accepted 13 October 2023
Available online 16 October 2023
0889-1575/© 2023 Elsevier Inc. All rights reserved.
J. Palatnik et al. Journal of Food Composition and Analysis 125 (2024) 105765

plasma sources are employed for detection. Interference associated to the formation of volatile Se species such as carbonyl, dimethylselenide,
the presence of transition metals in the sample that by interaction with and diethylselenide depending on the acid utilized. When formic acid is
the NaBH4 provoke its catalytic decomposition is another drawback to used as photochemical agent the volatile products of PVG are mainly
mention. SeH2 and SeCO (carbonyl) in less proportion. In addition, when nitric
In the last years, greener sample preparation and introduction acid is present in the reaction media the only volatile product identified
methods avoiding the use of corrosive and hazardous acids/reagents as was SeCO (Guo et al., 2003c).
well as high pressure and temperature have gained popularity. From the The purpose of this study was to develop an optimized, sensitive,
pioneering studies of Sturgeon and his group (Guo et al., 2003a,b,c), cost–effective and green method using optimized mild reaction condi­
photochemical vapor generation (PVG), utilizing free radicals generated tions to generate volatile species of Se for its determination in different
by a photo–redox reactions in the presence of low molecular weight real food samples. Multivariate techniques were performed to determine
organic acids has appeared as an interesting and powerful alternative to the critical conditions of the experimental parameters.
traditional chemical HG methods for sample introduction (Sturgeon,
2017; Yin et al., 2011). 2. Materials and methods
For analytical purposes, this technique opened novel routes for
introduction of chemical elements into atomic spectrometric sources. It 2.1. Instrumentation
was applied to generate volatile species of inorganic Se at room tem­
perature using ultraviolet radiation (Guo et al., 2003c). PVG has All measurements were made using an atomic absorption spec­
extended in scope to a broader range of elements including nonmetals trometer (AAS) PerkinElmer AAnalyst 200 provided with a quartz cell
such as alkyl halide generation (Leonori and Sturgeon, 2019). Briefly, in covered with an electrically heated mantle. A Se electrode less discharge
PVG the analyte is converted to a volatile species by the addition of a low lamp (PerkinElmer, Norwalk, CT, USA), was used as radiation source,
molecular weight organic acid (photochemical agent) to the sample and operated with a current of 280 mA, at a resonance wavelength of 196.03
subsequent exposure to the action of UV radiation. The generated vol­ nm, and a slit width of 2.3 nm.
atile species are conducted on–line to the detection instrument in a flow A schematic diagram of the instrumental set–up for the
of inert gas. In general, low–pressure Hg discharge lamps are used for UV–PVG–AAS system employed in continuous on–line mode is shown in
irradiation of the media to induce PVG. Different acids have been tested Fig. 1. In detail, a HPLC quaternary gradient pump Jasco model
as source of reducing radicals including, formic, acetic, malonic, and PU–2089 (Jasco, Easton, MD, USA) and a syringe injection six–port
propionic (Guo et al., 2003c; Nováková et al., 2017). valve (Model 7725i, Rheodyne) with a 200 µL loop were used to inject
In subsequent years, other authors reported the successful applica­ and propel samples/reagents/blanks into the stream of the reaction
tion of PVG to the determination of different elements/species in a di­ medium at a flow rate of 5.0 mL min–1. The photoreactor (PR) consists of
versity of matrices using different instrumental approaches. In the last 3.4 m long polytetrafluoroethylene (PTFE) tube (i.d. 1.0 mm/o.d. 1.5
15 years, research on the generation of targeted volatile forms of As (Gao mm, JQ Industries, Argentina) wrapped around a germicide low-
et al., 2015), Bi (Vyhnanovský et al., 2021), Hg (Covaci et al., 2017; Gao pressure Hg vapor UV lamp (254 nm, 15 W, dimension 450 mm, Phil­
et al., 2010), Mo (Šoukal et al., 2018), Se (Campanella et al., 2016; da lips). The PVG reactor was covered by wrapping it with aluminum foil to
Luz Potes et al., 2019), Te (He et al., 2018) was developed to improve prevent operators from UV radiation and also served to increase the
the detection power. In the case of Se, the presence of low molecular efficiency by reflecting UV radiation. Argon with a purity of 99.999 %
weight organic acids (formic, acetic, propionic, and malonic) leads to from Air Liquid (Buenos Aires, Argentina) was used as carrier gas. It was
the conversion of inorganic Se(IV) through UV irradiation, resulting in introduced to the system through a “T” connection inserted between the

Fig. 1. Schematic diagram of the instrumental set-up for the UV-PVG-AAS system employed in continuous on-line mode.

2
J. Palatnik et al. Journal of Food Composition and Analysis 125 (2024) 105765

outlet of the PR and the borosilicate gas liquid–separator (GLS), (part dryer (Virtis, New York, USA) Model 6 Lyophiliser 12 L. For nuts, firstly
number M055G003, PS Analytical, UK), internal volume of 10 mL. they were ground using a mortar and pestle. Subsequently were dried at
Argon flow rate was controlled on–line by a flow–meter (maximum flow: 60ºC until constant weight. After this step, a 6775 Freezer Mill cryogenic
250 mL min− 1). To facilitate Se atomization, the outlet of the GLS was grinder (New Jersey, USA) was employed.
mixed with a flow of H2 (auxiliary gas) introduced in the transport line The pre-reduction step was performed by adding to the dried sample
through a “T” connection between the GLS and the quartz cell. Hydrogen or calibration standard, 300 µL of concentrated HCl; then 2.5 mL of
gas was generated by a Canberra Packard model 9100 Generator DDW were incorporated into the flask and heated in a water bath
(Schwadorf, Austria). The volatile Se compounds generated were (85–90 ◦ C) for 30 min.
transported from the GLS to the atomic absorption apparatus through a Samples under study must be necessarily put in solution before
PTFE tube. PTFE tubes of various inner diameters were used as con­ presenting them to the analytical technique selected for analysis. Three
necting material in the system. For pre–reduction of Se(VI) the samples procedures were used and compared:
were heated on a hot plate (LabTech EH20B). An EthosUP (Milestone,
Bergamo, Italy) microwave apparatus equipped with 15 Teflon–PFA • Procedure 1– For formic acid solubilization, aliquots of 50 mg of
vessels was used to digest food samples. powdered food samples were weighed directly into polypropylene
An ultrasonic bath (Testlab, Buenos Aires, Argentina) was used to tubes and 5 mL of HCOOH were added. The mixture was placed in a
extract the Se compounds from the samples (frequency: 40 kHz; power: thermostated (60 ◦ C) ultrasound bath for two cycles of 30 min each.
160 W). Then, solutions were centrifuged at 4200 rpm for five minutes. An
An inductively coupled plasma single quadrupole mass spectrometer aliquot of 1 mL of the supernatant was transferred into 15 mL
PerkinElmer model NexION 300X was used for Se determination in food volumetric flasks, for the pre-reduction of Se(VI) to Se(IV) and finally
samples for comparison purposes. The instrumental parameters adopted brought to the mark. All experiments were carried out in triplicate
are summarized elsewhere (Londonio et al., 2019). • Procedure 2: Microwave–assisted acid digestion was also imple­
mented in the preparation of food samples. To this end, aliquot of
2.2. Reagents and chemicals 500 mg of each sample were weighted into the digestion vessel and a
mixture of 8 mL HNO3 (65 % w/w) and 1 mL H2O2 (30 % w/w) were
Deionized distilled water (DDW) produced by an Arium® Water added. The MW digestion program used was as follows: temperature
Purification System (Sartorius, Goettingen, Germany) was used for the ramped to 220 ◦ C for 25 min, maintained at 220 ◦ C for 20 min and
preparation of all solutions. All reagents were of analytical grade. finally cooled to room temperature. After cooling, the vessels were
Concentrated nitric acid (65 % w/w), hydrochloric acid (37 % w/w) and opened and the digested solutions were transferred to polypropylene
hydrogen peroxide 30 % (w/w) (Merck, Darmstadt, Germany) of vials and evaporated to almost dryness on a digestion block (90 ◦ C)
analytical grade were used for sample treatment and/or preparation of to eliminate nitric acid. Dried residues were subjected to the pre-
the standards. Nitric acid was additionally purified in a sub–boiling reduction step and redissolved in water to a final volume of 15 mL.
distillation system, Distillacid BSB–939–IR (Berghof, Eningen, All experiments were carried out in triplicate.
Germany). • Procedure 3– Microwave–assisted acid digestion for Se determina­
A stock solution of sodium selenate (Na2SeO4 in HNO3 from Merck) tion by ICP–MS: Samples were prepared as described in Procedure 2,
was used. Solutions of formic (HCOOH) and acetic acid (CH3COOH) but after cooling the digested solutions were transferred to volu­
were employed after appropriate dilution in DDW of the respective metric flasks and DDW was added to reach a final volume of 15 mL.
concentrated reagents (HCOOH, ≥88.0 % from Anedra, and CH3COOH
99.9 % from Biopack). Nitric acid, HCl and H2O2 were used in different 2.5. Analytical procedure
steps of the analytical procedure, as extraction agents, for pre–reduction
of Se(VI) and for microwave–assisted digestion of food samples. The PVG of volatile species of Se was carried out using a flow in­
Standard solutions of As(III), Co(II), Cu(II), Fe(III), Mn(II), Ni(II), Sb jection mode system. Solutions of Se standards or digested samples were
(III), Zn (II) of 1000 mg L–1 (Merck) were employed in the study of in­ injected into a 200 µL sample loop through a six–port valve. Then, the
terferences. Diluted working solutions were prepared daily by serial solution loaded in the loop was injected into the carrier solution flowing
dilutions of the 1000 mg L–1 standard solutions. All glassware and at 5.0 mL min–1 and transported to the PR where they were UV–irradi­
plastic bottles used were cleaned by rinsing with DDW, soaking with a ated. The irradiated solution was mixed with Ar (120 mL min–1) using a
10 % (v/v) HNO3 solution for 24 h and then rinsing several times with T–shaped connector and transported to the GLS to separate the volatile
DDW. All samples and standards were stored in polyethylene bottles Se species from the liquid phase. After the GLS, the vapor was mixed
(50 mL) or Falcon® tubes. with H2 (5 mL min–1) and transported through the transfer line to the
quartz cell. The generated volatile Se compounds were atomized in the
2.3. Samples analyzed AAS instrument. Signal peak area was used for quantification. The whole
procedure took about 1.5 min. A diagram of the experimental system
A total of 16 samples (garlic and nuts) were analyzed: employed is depicted in Fig. 1. More details are given in the Instrumental
section.
(i) 4 samples of garlic (Allium sativum L.) purchased in a local market Table 1 summarizes all optimized experimental conditions employed
in Buenos Aires, Argentina (identified as G1–G4) in this study.
(ii) 4 samples of garlic enriched with Se (identified as GE1–GE4).
Selenium enrichment experiments of garlic samples are described 2.6. Optimization of experimental chemical and physical variables
elsewhere (Pérez et al., 2018).
(iii) 4 samples of nuts (Juglans regia) purchased from a local market in The effect of different parameters on the efficient generation of
the city of Buenos Aires, Argentina (identified as N1–N4). volatile Se species was assessed for achieving the best performance of
(iv) 4 samples of Brazilian nuts (Bertholletia excelsa) purchased in a the system. For all the experiments of the optimization step, a solution
market in Belém, Pará, Brazil (identified as BN1–BN4) containing 100 μg L–1 of Se(IV) was used. All tests were carried out by
triplicate.
2.4. Sample preparation PVG efficiency is highly dependent on the type and concentrations of
the organic acid selected. In a first approach, variables affecting the PVG
Garlic samples were lyophilized and homogenized using a freeze of Se were examined separately. The critical parameters influencing the

3
J. Palatnik et al. Journal of Food Composition and Analysis 125 (2024) 105765

Table 1 species of Se to the atomization cell. Therefore, two factors and five
Instrumental details and operating conditions of the PVG-AAS system for Se (IV) levels full factorial design (FF) were selected requiring a total of 25
determination. experiments.
Instrument Perkin Elmer AAnalyst 200

Cell temperature 940 ºC

2.7. Calibration
EDL lamp current 280 mA
Measurement mode Peak area Two calibration methodologies were followed. Under optimized
Slit width 2.3 nm conditions, an external calibration curve was established within the
Wavelength 196.03 nm
concentration range from LOQ to 100 μg L− 1 of Se (IV) for the deter­
Photochemical vapor generation
UV lamp Power: 15 w mination of Se in garlic samples. For the complexity of the matrix, a
Wavelength: 254 nm three–point method of standard addition was employed for the deter­
PTFE reaction coil Length: 3.4 m mination of Se in nuts to obviate any potential matrix interferences.
i.d. 1 mm
o.d .1.5 mm
CH3COOH concentration 5 % (v/v)
2.8. Limits of detection and reliability criteria
HCOOH concentration 1.5 % (v/v)
Carrier gas (Ar) flow rate 120 mL min-1 Instrumental limits of detection (LOD) and quantification (LOQ)
Auxiliary gas (H2) flow rate 5 mL min-1 were calculated as the concentration associated with three (LOD) and
Sample flow rate 5 mL min-1
ten times (LOQ) the standard deviation of 10 independent measure­
ments of the procedural blank (3σ criterion), respectively. Blanks were
PVG of Se investigated were: (i) concentration of photochemical re­ subjected to the same procedure that the samples.
agent, (ii) Ar (carrier gas) and H2 (auxiliary gas) flow rates, (iii) irradi­ The linearity was checked by an external calibration curve that was
ation time, and (iv) temperature of the inlet solution. made in the range 0–150 μg L–1, which included blanks and standards at
concentrations of 5, 10, 20, 40, 60, 80, 100, and 150 μg L–1 of Se(IV).
2.6.1. Screening studies for the parameters optimization The results obtained from the optimized method were compared
Preliminary tests were performed to find favorable reaction condi­ with the ICP–MS measurement, utilizing an elliptical joint confidence
tions to generate volatile Se species using low concentrations of organic region (EJCR) with bilinear least squares analysis (González et al.,
acids. Formic and acetic acids were investigated. Both acids were tested 1999).
in the range: acetic, 0–10 % and formic, 0–30 %; alone and mixed in The accuracy of the proposed method and matrix effect (interference
different proportions. of the food sample in the analytical response) were verified using an
In this investigation, a HPLC pump was used to propel solutions to analytical recovery test. Independent aliquots from selected previously
ensure a constant flow of the organic acids assessed by mixing three analyzed selected powdered food samples were subsampled and fortified
lines, containing water and the organic acid/acids. In addition, using a with Se. Two concentration levels of 10 and 20 μg L-1, corresponding to
HPLC pump for flow–injection made possible to accurately change the 0.3 and 0.6 μg g-1 in the solid, respectively, were used for the spikes.
composition of the mobile phase during the optimization process. Then, samples were then subject to MW-assisted digestion and analyzed
Sample flow rate is inversely equivalent to sample residence/irra­ using the optimized PVG-AAS method.
diation time for a fixed reactor volume. However, the sample flow rate
was initially adjusted by varying it from 1 to 5 mL min− 1, with the 2.9. Interference study
objective to obtain an appropriate peak for analytical purposes in terms
of height, shape, and width. Using the optimized working conditions, the effect of potentially
For the selected sample flow rate value, the residence time of Se coexisting ions on the determination of Se (IV) by UV–PVG–AAS was
within the irradiation process was adjusted by modifying the length of investigated. To this end, solutions containing 100 μg L− 1 of Se and the
the PTFE tube wrapped around the UV lamp, ranging from 1 to 6 m foreign ion under study, at two different levels, were subjected to the
(corresponding to irradiation times varying from 9.4 to 56.5 s). analytical procedure. Variations over ± 5 % in the analytical signal of Se
Concentrations of gases were studied in the range: Ar, in the presence of potential interfering ions were taken as interference.
20–220 mL min–1, and H2, 0–25 mL min–1. Higher flow rates were not The effect of hydride forming elements and transition metals such as
tested to avoid dilution effects that would cause a decrease in the As(III), Co(II), Cu(II). Fe(III), Mn(II), Ni(II), Sb(III) and Zn(II) was
analytical signal of Se. The temperature of the inlet solution was also studied. Nitric and HCl acids were also investigated to assess their effect
tested in the range from 25 ºC to 55;ºC. on Se signal in the conditions of the experiment.
A multivariate optimization technique was selected in this study as
described in the following pages. 2.10. Photochemical vapor generation efficiency

2.6.2. Multivariate optimization The efficiency of the UV–PVG system was estimated by the ratio of
Two experimental designs namely, Design 1 and Design 2 were the sensitivity between the developed PVG method and conventional
tested: HG–AAS using sodium tetrahydroborate as reductant and the recom­
mended conditions indicated by the manufacturer (Table S1).
• Experimental Design 1 consisted in the optimization of the volatile Se
species generation, and its transport (from the GLS to the atomiza­ 3. Results and discussion
tion cell). The four variables included in this design were (i) acetic
acid concentration, (ii) formic acid concentration, (iii) reactor length, 3.1. Sample treatment and evaluation of sample digestion procedures
and (iv) temperature of the inlet solution. A four factor–central
composite design (CCD) was employed for modeling the response An adequate sample preparation is a key step when PVG–AAS is used
surface. for Se quantification. When Procedure 1 was applied to garlic samples,
• Experimental Design 2 consisted in the optimization of the flow rates an analytical signal corresponding to a recovery of ~ 20 % was obtained;
of the two gases: (i) Ar, employed to transport volatile species of Se to respect to the signal obtained using MW-digestion and ICP-MS quanti­
the GLS, and (ii) H2, used for reduction and transport of volatile fication. However, when the same samples (Procedure 1) were measured
by ICP–MS, the recoveries were ~90–110 %. Similar results were

4
J. Palatnik et al. Journal of Food Composition and Analysis 125 (2024) 105765

reported by Lopes et al. when formic acid was used as extractant (Lopes increased with the increased concentrations of both acids. The presence
et al., 2016). According to Thiry et al. (Thiry et al., 2012) Se in food of acetic acid resulted in a less pronounced dependence of the signal on
samples is present as selenocysteine and selenomethionine (SeMet). The acid concentration.
lower recovery obtained for PVG–AAS compared to ICP–MS measure­ It was observed that the mixture of both acids increased the effi­
ments could be attributed to the mild conditions used in PVG; it could ciency facilitating the photochemical reduction of Se. In this context, a
happen that the amino acid molecules remain practically intact and that multivariate optimization was carried out as described in the next
therefore the pre–reduction step was not as efficient as expected. section.
However, it is reported that Se(VI) may be present which is unreactive to The Ar carrier gas flow rate used to propel the sample through the
PVG and that the direct PVG does occur for selenocysteine and seleno­ transfer line to the GLS determines the liquid–gas separation efficiency
methionine (Suzuki et al., 2012). On the other hand, when Procedure 2 of the volatile Se species. This gas also significantly influences the at­
(MW digestion) was tested, recoveries better than 90% were obtained. In omization of the analyte in the heated quartz cell. The addition of a
this case, after MW digestion, an evaporation step was also introduced to stream of H2 is necessary to achieve an effective atomization; it is due to
eliminate the HNO3 used for digestion since the presence of NO–3 pro­ the generation of free radicals necessary to atomize SeH2. A rapid
duces severe interferences in the production of volatile species of Se removal is advantageous to avoid decomposition of the generated spe­
(described in detail in 3.4, Interference Study section). cies. Both gases produced a significant impact on Se signal.
The Teflon transfer line between the GLS and the quartz tube was
3.2. Screening studies for the parameters optimization kept as short as possible (about 30 cm). However, no significant differ­
ences were observed at greater lengths (up to 75 cm). This fact is in
Screening experiments showed that formic and acetic acids have a concordance with that reported by Guo et al., who reported that once
significant impact on Se response; the intensity of Se(IV) signal formed the volatile species of Se they are very stable and does not suffer

Fig. 2. Empirical response surface fitted for Design 1 with quadratic polynomic: a) 3D visualization of fitted surface for acetic acid and formic acid; b) 3D visu­
alization of fitted temperature and reactor length; c) area predicted by the model vs actual measured value.

5
J. Palatnik et al. Journal of Food Composition and Analysis 125 (2024) 105765

any losses during the transport process (Guo et al.; 2003). 3.4. PVG Interference study

3.3. Multivariate optimization It is well known that the determination of Se by HG technique is

seriously affected by the presence of transition metals, mainly Co, Cu,
The response surface was fitted to the Experimental data of Designs 1 and Ni. Additionally, potential interferences arise from hydride–forming
and 2, leading to the determination of optimized parameters for Se(IV) elements, which can lead to gas–phase interference in the heated quartz
using PVG-AAS. tube. The results of the interference study are presented in Table 2.
Respect to the hydride forming elements tested, the presence of As
3.3.1. Experimental design 1 and Sb did not produce interference in the determination of Se by
The response surface fitted for Experimental Design 1 is shown in PVG–AAS as it occurs when HG is used.
Fig. 2a and b and the equation obtained is included in Supplementary Copper is a typical transition metal that can produce depression on
Information. The best fitting model was quadratic polynomial. Se signal at concentrations higher than 1.0 mg L–1. According to Cam­
The evaluation of model fitness was evaluated by ANOVA test panella et al. (2016) the depressing effect could be ascribed to the for­
described by Bezerra et al. (2008), and the model fitness obtained can be mation of metallic Cu(0), producing a heterophasic system in the
considered satisfactory. Furthermore, to show its fitness actual reaction coil. Rybínová et al. (2016) also reported that Cu(II) was the
measured value vs value predicted by the model was plotted and shown most interfering transition metal affecting seriously the determination of
in Fig. 2c. The figure shows that the points are in a straight line of slope Se by PVG. In our study, Cu depressed severely Se signal as follows: –37
~ 1 and intercept ~ 0. % and –57 % when 1.0 mg L–1 and 10 mg L–1 Cu were added to the test
The model shows that temperature has no significant contribution to
Se (IV) signal, at least in the range studied. These results agree with Table 2
those reported by Guo et al. (2003b). The optimal parameters were: Effect of potential interfering ions on Se instrument response. Results are
acetic acid concentration: 5 % (v/v); formic acid concentration: 1.5 % expressed as mean concentration ± standard deviation (n = 3).
(v/v); reaction coil length: 3.4 m (equivalent to an irradiation time of Interferent Interferent concentration Relative response [%]
32 s) and temperature of the inlet solution: 20–25 ºC. [mg L-1]

As(III) 0.1 105 ± 6

3.3.2. Experimental design 2 1 104 ± 6
The response surface fitted for Experimental Design 2 obtained is Co (II) 1 102 ± 5
shown in Fig. 3a and the equation obtained is included in Supplemen­ 10 98 ± 5
Cu (II) 1 57 ± 3
tary Information. The best fitting model was cubic polynomial.
10 37 ± 2
The evaluation of model fitness evaluated by ANOVA test and the Fe (III) 1 101 ± 6
model fitness obtained can be considered satisfactory. Furthermore, to 10 86 ± 5
show its fitness, actual measured value vs values predicted by the model Mn (II) 1 91 ± 5
were plotted and shown in Fig. 3b, The points of the curve are in a 10 78 ± 5
Ni (II) 1 95 ± 6
straight line with a slope ~ 1 and intercept ~ 0. 10 92 ± 5
The optimal parameters found for the system under study were: Ar: Sb(III) 0.1 107 ± 5
120 mL min–1; H2: 5 mL min–1. 1 107 ± 6
Factors and levels used in the factorial design are depicted in Zn (II) 1 100 ± 5
10 96 ± 6
Table S2.
* The studied elements were tested at a concentration of Se of 100 μg L-1

Fig. 3. Empirical response surface fitted for Design 2 with cubic polynomic: a) 3D visualization of fitted surface for H2 and Ar; b) area predicted by the model vs
actual measured value.

6
J. Palatnik et al. Journal of Food Composition and Analysis 125 (2024) 105765

solutions, respectively. No significant interference from Co, Fe, Mn, Ni, Table 3
Co and Zn was detected when 1.0 mg L–1 were present in the Se con­ Recovery test for five selected food samples*. Data are expressed as means
taining sample solution. On the other hand, when concentrations of concentration standard deviation ± standard deviation (n = 3).
10 mg L–1 were tested, Se signal was reduced. It is important to note that Garlic sample Added Se (µg g-1) Found (µg g-1) Recovery (%)
concentrations of Fe, Mn and Ni of 10 mg L–1 are not levels expected in S1 - 1.7 ± 0.1 -
the samples analyzed in this study. 0.3 1.9 ± 0.2 94 ± 6
Positive interferences were also reported in the literature. In this 0.6 2.4 ± 0.2 107 ± 5
direction, other authors added transition metals at trace levels as sen­ S2 - 1.6 ± 0.1 -
0.3 1.8 ± 0.2 96 ± 5
sitizers/catalysts for PVG. They include metals such as Ni(II), Cu(II), and
0.6 2.0 ± 0.2 91 ± 5
Cd(II) to increase the production of volatile species of Pb, Co, As, Cl, Se S3 - 1.8 ± 0.2 -
and Te when generated in solutions of UV–irradiated acetic and formic 0.3 2.0 ± 0.2 96 ± 6
acids (Leonori and Sturgeon, 2019). This behavior was used by Xu et al. 0.6 2.5 ± 0.2 102 ± 6
(2018) to enhance the in situ production of Se(VI) by adding Cd(II) to an S4 - 0.8 ± 0.1 -
0.3 1.0 ± 0.1 95 ± 6
acetic acid reaction medium forming a nano–CdSe photocatalyst. In our 0.6 1.4 ± 0.2 105 ± 5
study, no significant enhancing effects were observed for the elements S5 - 0.3 ± 0.0 -
tested. 0.3 0.6 ± 0.0 93 ± 6
There is another group of concomitants that deserves to be studied. 0.6 1.0 ± 0.0 105 ± 5
Nitric and hydrochloric acids are common concomitant since they are *
Se(IV) solution were added to the powdered garlic samples to perform the
frequently used for sample digestion. According to our screening ex­ recovery test.
periments, the presence of both acids produced an increase in the gen­
eration efficiency of Se volatile compounds. Leonori & Sturgeon 3.6. Photochemical vapor generation efficiency
(Leonori and Sturgeon, 2019) postulated mechanisms involving redox
termination of radical chain reactions, especially from oxidants such as The efficiency of the PVG is the fraction of the analyte Se (IV) that
dissolved O2, NO–3 and NO–2; being nitrite interference more severe. In can be removed from solution and determined in the whole process. The
addition, they reported that Se signal intensity increases in the same way sensitivity values of the optimized PVG method and conventional
by the presence of NO-3 or NO2 - anion deriving from either sodium ni­ HG–AAS were 0.0132 L µg − 1 and 0.0292 L µg− 1, respectively. These
trate or sodium nitrite. According to Mollo and Knochen (Mollo and values were used to calculate the efficiency. In both analytical tech­
Knochen, 2020) nitrate, reacts with the reductant radicals decreasing its niques, the efficiency depends on a series of parameters namely, vapor
availability and yielding oxidizing species. When this occurs, the me­ generation; separation of the volatile analyte in GLS from condensed
dium is no longer reductant and the formation of the volatile species is phase; and transportation from GLS to the AAS instrument. For calcu­
quenched. Several authors investigated the adverse effects caused by lation, we assumed that conventional CVG–AAS has efficiency of 100 %
NO–3 as well as the different methods used for the abatement of nitrate efficiency. The efficiency achieved in this study was 46%, which is
interference (Lopes et al., 2016; Mollo and Knochen, 2020). comparable to the value reported by Guo et al. of 50 ± 10 % using acetic
Fig. S1a shows the dependence of Se absorbance when HNO3 was and malonic acids, and significantly higher than the reported efficiency
tested indicating that the nitric acid has a marked influence on the Se of 10–15 % when formic acid was used alone (Guo et al., 2003c).
signal. Our results are in good agreement with those reported by Nevertheless, this value was lower than the 79 % efficiency reported by
Rybínová et al. (2016). In Fig S1b, similar results were obtained when da Luz Potes et al., 2019, who employed PVG coupled to graphite
HCl was assessed. Consequently, in our study, in the MW–assisted furnace atomic absorption spectrometry (GF–AAS) (da Luz Potes et al.,
digestion of real samples, the amount of oxidizing acids was minimized 2019). The disparity in efficiency may be attributed to the fact that the
and digested samples were evaporated to almost dryness and quartz PR exhibits much higher UV–C transmission compared to the
re-solubilized in diluted HCl. PTFE tubing employed in our study. Nevertheless, PTFE tubing has the
advantage of its flexibility and durability.

3.5. Analytical performance

3.7. Analysis of real samples
Using optimized analytical conditions, the obtained values of LOD
and LOQ were 1.1 µg L–1 and 3.6 µg L–1, which correspond to 33 ng g-1 Data concerning Se determination in food samples is scarce. In our
and 109 ng g-1 in the solid, respectively. For comparative purposes, our opinion, it can be attributed to the complexity of methods used for Se
results were compared in Table S3 (supplementary information) with detection and quantification in food at trace and ultra–trace levels.
those reported in the literature. In general terms, our results are of the Recently, da Silva et al. reported an illustrative study on the accumu­
same order of magnitude that those reported previously, except those lation and localization of Se and Ba in nuts using spectroanalytical
reported by Rybínová et al. (2016). techniques (da Silva Junior et al., 2022). In another study, the accu­
Linearity was attained from levels close to the LOQ up to 100 μg L–1. mulation of Se in plant foods and Se intake of residents in a moderately
The repeatability expressed as the relative standard deviation (% RSD, Se-enriched area was described (Li et al., 2023).
n = 10) was 2 % at a Se(IV) level of 100 μg L− 1. Although PVG has been used for some analytes determination as
The coefficient of determination (R2) of the calibration curve was described before, matrix effects and sample preparation remain a chal­
better than 0.997 until Se(IV) concentration exceeded 100 μg L− 1. From lenge for this technique when applied to the analysis of real samples. For
this value R2 began to decline. This effect could be attributed to the fact this reason, it can be observed in the literature that published studies
that at higher Se (IV) concentrations the efficiency of generation is generally analyze liquid samples and only a minority part of the studies
lower. analyze solid samples.
The comparison of results from the PVG–AAS optimized method with For comparative purposes, samples under study were also deter­
the ICP–MS measurement is shown in Fig. S2a. The optimized method is mined by ICP–MS after MW–sample digestion (Table 4). Determination
as accurate as the ICP–MS method, as depicted in Fig. S2b, due to the of Se by ICP–MS is in general a difficult task because it is hampered by
ideal point (coordinates: a=0, b=1) is inside the EJCR. spectral interferences. It is affected by the presence of polyatomic ions,
Recoveries in the garlic samples analyzed varied between 93 % and such as 40Ar+ 12 35 +
2 or C Cl2 , and the effects of these ions may be overcome
107 %. Results of the recovery test are depicted in Table 3. by using different strategies, such as (i) the selection of another isotope,

7
J. Palatnik et al. Journal of Food Composition and Analysis 125 (2024) 105765

Table 4 Declaration of Competing Interest

Comparison of data determined by PVG-AAS and ICP-MS in different samples.
Concentrations of Se are expressed in µg g-1 *. The authors declare that they have no known competing financial
Sample Se concentration Se concentration Error (%) interests or personal relationships that could have appeared to influence
(PVG-AAS) (ICP-MS) the work reported in this paper.
G 1 (market) 5.1 ± 0.52 4.50 ± 0.45 13
G 2 (market) 0.36 ± 0.03 0.33 ± 0.03 9 Data availability
G 3 (market) < 0.11 < 0.08 -
G 4 (market) < 0.11 < 0.08 -
No data was used for the research described in the article.
GE 1 (enriched) 211 ± 19 205 ± 20 3
GE 2 (enriched) 112 ± 10 123 ± 12 -9
GE 3 (enriched) 248 ± 25 245 ± 25 1 Acknowledgments
GE 4 (enriched) 221 ± 22 230 ± 23 -4
N 1 (Argentina) 0.73 ± 0.07 0.71 ± 0.07 3 The authors are thankful to Comisión Nacional de Energía Atómica
N 2 (Argentina) 0.56 ± 0.06 0.58 ± 0.05 -3
N 3 (Argentina) < 0.11 < 0.08 -
for the continuous support of our research. Financial support from
N 4 (Argentina) < 0.11 < 0.08 - CONICET through project PIP 00078 is acknowledged. PS thanks M. G.
BN 1 (Brazil) 100 ± 10 103 ± 10 -3 R. Vale for providing the nut samples and for the useful discussion. L.
BN 2 (Brazil) 78.8 ± 8.0 72.8 ± 7.2 8 Valenzuela and F. González (CNEA) are also acknowledged.
BN 3 (Brazil) 30.4 ± 3.0 30.5 ± 2.9 0
BN 4 (Brazil) 40.7 ± 4.1 43.4 ± 4.3 -6
Appendix A. Supporting information
*
Three set of replicates, each set consisting of three measurements.
Supplementary data associated with this article can be found in the
(ii) the use of mathematical correction, (iii) the use of cool plasma or (iv) online version at doi:10.1016/j.jfca.2023.105765.
the use of reactions and/or collisions by the introduction of a variety of
gases. Strategy (i) was selected in this study, and the 82Se isotope, which References
is free from the Ar-Ar interference, was chosen. However, its low natural
abundance of 8.73% yields a poor LOD. To solve this problem, a 3 % v/v Bezerra, M.A., Santelli, R.E., Oliveira, E.P., Villar, L.S., Escaleira, L.A., 2008. Response
surface methodology (RSM) as a tool for optimization in analytical chemistry.
methanol (final concentration) was added to the sample and standards Talanta 76 (5), 965–977. https://doi.org/10.1016/j.talanta.2008.05.019.
to enhance the signal (Sullivan et al., 2013). Concentrations of Se Campanella, B., Menciassi, A., Onor, M., Ferrari, C., Bramanti, E., D’Ulivo, A., 2016.
(expressed in µg g-1) in garlic and nuts samples are shown in Table 4. Studies on photochemical vapor generation of selenium with germicidal low power
ultraviolet mercury lamp. Spectrochim. Acta Part B: At. Spectrosc. 126, 11–16.
Results were compared with those obtained by ICP-MS and percentage https://doi.org/10.1016/j.sab.2016.10.007.
errors were calculated. Covaci, E., Senila, M., Ponta, M., Darvasi, E., Petreus, D., Frentiu, M., Frentiu, T., 2017.
Methylmercury determination in seafood by photochemical vapor generation
capacitively coupled plasma microtorch optical emission spectrometry. Talanta 170,
4. Conclusions
464–472. https://doi.org/10.1016/j.talanta.2017.04.036.
da Luz Potes, M., Nakadi, F.V., Frois, C.F.G., Vale, M.G.R., da Silva, M.M., 2019.
On the basis of the outcome of this study, it can be concluded that Investigation of the conditions for selenium determination by photochemical vapor
generation coupled to graphite furnace atomic absorption spectrometry. Microchem.
PVG of Se(IV) using FI–AAS and a low power UV Hg lamp can be ach­
J. 147, 324–332. https://doi.org/10.1016/j.microc.2019.03.053.
ieved using mild conditions with a mixture of 1.5 % (v/v) formic and 5 % da Silva Junior, E.C., Duran, N.M., de Lima Lessa, J.H., Ribeiro, P.G., de Oliveira Wadt, L.
(v/v) acetic acid as photochemical reagents. No additional chemicals H., da Silva, K.E., de Oliveira Junior, R.C., 2022. Unraveling the accumulation and
were essentially needed to generate volatile Se compounds and the localization of selenium and barium in Brazil nuts using spectroanalytical
techniques. J. Food Compos. Anal. 106, 104329.
interface is simple and can be adopted as alternative for rapid and Dumont, E., Vanhaecke, F., Cornelis, R., 2006. Selenium speciation from food source to
friendly determination of other elements. In comparison with conven­ metabolites: a critical review. Anal. Bioanal. Chem. 385, 1304–1323. https://doi.
tional HG methods, tetrahydroborate is avoided and therefore the gen­ org/10.1007/s00216-006-0529-8.
Gao, Y., Yang, W., Zheng, C., Hou, X., Wu, L., 2010. On-line preconcentration and in situ
eration of hazardous substances is reduced or eliminated. In addition, photochemical vapor generation in coiled reactor for speciation analysis of mercury
concentrated HCl is not necessary for volatile species generation. and methylmercury by atomic fluorescence spectrometry. J. Anal. At. Spectrom. 26
Optimization of the different variables involved in the PVG process (1), 126–132. https://doi.org/10.1039/C0JA00137F.
Gao, Y., Sturgeon, R.E., Mester, Z., Hou, X., Yanga, L., 2015. Multivariate optimization of
was performed using chemometric tools that allowed mapping the entire photochemical vapor generation for direct determination of arsenic in seawater by
experimental domain studied. inductively coupled plasma mass spectrometry. Anal. Chim. Acta 901, 34–40.
Data obtained for the analysis of real solid samples show that https://doi.org/10.1016/j.aca.2015.10.020.
González, A.G., Herrador, M.A., Asuero, A. n G., 1999. Intra-laboratory testing of method
UV–PVG could become an alternative route of volatile compounds
accuracy from recovery assays. Talanta 48 (3), 729–736. https://doi.org/10.1016/
generation due to its relatively low cost, simplicity, high sensitivity and S0039-9140(98)00271-9.
green profile. In addition, the necessary instrumentation is easy to find Guo, X., Sturgeon, R.E., Mester, Z., Gardner, G.J., 2003a. Photochemical alkylation of
inorganic selenium in the presence of low molecular weight organic acids. Environ.
in any analytical laboratory.
Sci. Technol. 37 (24), 5645–5650 https://doi.org/0.1021/es034418j.
Our future PVG studies will be focused to study the use of photo­ Guo, X., Sturgeon, R.E., Mester, Z., Gardner, G.J., 2003b. UV light-mediated alkylation of
catalysts, including nanomaterials, to improve the conversion efficiency inorganic selenium. Appl. Organomet. Chem. 17, 575–579. https://doi.org/
of elements for achieving higher sensitivities. 10.1002/aoc.473.
Guo, X., Sturgeon, R.E., Mester, Z., Gardner, G.J., 2003c. UV vapor generation for
determination of selenium by heated quartz tube atomic absorption spectrometry.
CRediT authorship contribution statement Anal. Chem. 75, 2092–2099. https://doi.org/10.1021/ac020695h.
He, H., Peng, X., Yu, Y., Shi, Z., Xu, M., Ni, S., Gao, Y., 2018. Photochemical vapor
generation of tellurium: synergistic effect from ferric ion and nano-TiO2. Anal.
Jaime Palatnik: Conceptualization, Investigation, Methodology, Chem. 90 (9), 5737–5743. https://doi.org/10.1021/acs.analchem.8b00022.
Visualization, Validation, Writing – review & editing. Agustin Londo­ Leonori, D., Sturgeon, R.E., 2019. A unified approach to mechanistic aspects of
nio: Conceptualization, Methodology, Writing – review & editing. photochemical vapor generation. J. Anal. At. Spectrom. 34 (4), 636–654. https://
doi.org/10.1039/C8JA00354H.
Ezequiel Morzan: Software, Formal analysis. Rodolfo Wuilloud: Li, X., Luo, Y., Zeng, C., Zhong, Q., Xiao, Z., Mao, X., Cao, F., 2023. Selenium
Validation, Resources. Patricia Smichowski: Conceptualization, accumulation in plant foods and selenium intake of residents in a moderately
Writing – Original Draft, Writing – review & editing, Supervision. selenium-enriched area of Mingyueshan, Yichun, China. J. Food Compos. Anal. 116,
105089 https://doi.org/https://doi.org/10.3389%2Ffnut.2022.962312.

8
J. Palatnik et al. Journal of Food Composition and Analysis 125 (2024) 105765

Londonio, A., Morzán, E., Smichowski, P., 2019. Determination of toxic and potentially Sturgeon, R., Liu, J., Boyko, V., Luong, V., 1996. Determination of copper in
toxic elements in rice and rice-based products by inductively coupled plasma-mass environmental matrices following vapor generation. Anal. Chem. 68 (11),
spectrometry. Food Chem. 284, 149–154. https://doi.org/10.1016/j. 1883–1887. https://doi.org/10.1021/ac951259g.
foodchem.2019.01.104. Sturgeon, R.E., 2017. Photochemical vapor generation: a radical approach to analyte
Lopes, G.S., Silva, F.L., Grinberg, P., Sturgeon, R.E., 2016. An evaluation of the use of introduction for atomic spectrometry. J. Anal. At. Spectrom. 32, 2319–2340. https://
formic acid for extraction of trace elements from Brazil nut and babassu coconut and doi.org/10.1039/C7JA00285H.
its suitability for multi-element determination by ICP-MS. J. Braz. Chem. Soc. 27, Sullivan, D., Zywicki, R., Yancey, M., 2013. Method for the determination of total
1229–1235. https://doi.org/10.5935/0103-5053.20160018. selenium in a wide variety of foods using inductively coupled plasma/mass
Marrero, J., Smichowski, P., 2002. Evaluation of vapor generation for the determination spectrometry. J. AOAC Int. 96 (4), 786–794. https://doi.org/10.5740/jaoacint.12-
of nickel by inductively coupled plasma-atomic emission spectrometry. Anal. 389.
Bioanal. Chem. 374, 196–202. https://doi.org/10.1007/s00216-002-1372-1. Sunde, R. , 2012. Selenium. In A. C. Ross, B. Caballero, R. J. Cousins & K. L. Tucker
Mollo, A., Knochen, M., 2020. Towards the abatement of nitrate interference on selenium (Eds.), Modern nutrition in health and disease. Philadelphia, PA: Lippincott Williams
determination by photochemical vapor generation. Spectrochim. Acta Part B: At. & Wilkins: Jones & Bartlett Learning.
Spectrosc. 169, 105875 https://doi.org/10.1016/j.sab.2020.105875. Suzuki, T., Sturgeon, R.E., Zheng, C., Hioki, A., Nakazato, T., Tao, H., 2012. Influence of
Nováková, E., Linhart, O., Červený, V., Rychlovský, P., Hraníček, J., 2017. Flow injection speciation on the response from selenium to UV-photochemical vapor generation.
determination of Se in dietary supplements using TiO2 mediated UV-photochemical Anal. Sci. 28 (8), 807–811. https://doi.org/10.2116/analsci.28.807.
volatile species generation. Spectrochim. Acta Part B: At. Spectrosc. 134, 98–104. Thiry, C., Ruttens, A., De Temmerman, L., Schneider, Y.-J., Pussemier, L., 2012. Current
https://doi.org/10.1016/j.sab.2017.06.007. knowledge in species-related bioavailability of selenium in food. Food Chem. 130
Pérez, M.B., Maniero, M.Á., Londonio, A., Smichowski, P., Wuilloud, R.G., 2018. Effects (4), 767–784. https://doi.org/10.1016/j.foodchem.2011.07.102.
of common cooking heat treatments on selenium content and speciation in garlic. Vyhnanovský, J., Yildiz, D., Štádlerová, B., Musil, S., 2021. Efficient photochemical
J. Food Compos. Anal. 70, 54–62. https://doi.org/10.1016/j.jfca.2018.04.004. vapor generation of bismuth using a coiled Teflon reactor: effect of metal sensitizers
Pohl, P., Jamroz, P., Welna, M., Szymczycha-Madeja, A., Greda, K., 2014. Chemical- and analytical performance with flame-in-gas-shield atomizer and atomic
vapor generation of transition metals through the reaction with tetrahydroborate in fluorescence spectrometry. Microchem. J. 164, 105997 https://doi.org/10.1016/j.
recent achievements in analytical atomic spectrometry. Trends Anal. Chem. 59, microc.2021.105997.
144–155. https://doi.org/10.1016/j.trac.2014.04.010. Xu, F., Zou, Z., He, J., Li, M., Xu, K., Hou, X., 2018. In situ formation of nano-CdSe as a
Rybínová, M., Červený, V., Hraníček, J., Rychlovský, P., 2016. UV-photochemical vapor photocatalyst: cadmium ion-enhanced photochemical vapour generation directly
generation with quartz furnace atomic absorption spectrometry for simple and from Se (VI). Chem. Commun. 54 (38), 4874–4877. https://doi.org/10.1039/
sensitive determination of selenium in dietary supplements. Michrochem. J. 124, C8CC01513A.
584–594. https://doi.org/10.1016/j.microc.2015.10.004. Yin, Y., Liu, J.-F., Jiang, G., 2011. Photo-induced chemical-vapor generation for sample
Šoukal, J., Sturgeon, R.E., Musil, S., 2018. Efficient photochemical vapor generation of introduction in atomic spectrometry. Trends Anal. Chem. 30 (10), 1672–1684.
molybdenum for ICPMS detection. Anal. Chem. 90 (19), 11688–11695. https://doi. https://doi.org/10.1016/j.trac.2011.04.021.
org/10.1021/acs.analchem.8b03354.