**Spectroscopic Ellipsometry Detection of Prostate Cancer Bio-Marker PCA3 Using Specific Non-Labeled Aptamer: Comparison with Electrochemical Detection †**

**Sarra Takita 1,\* , Alexei Nabok <sup>1</sup> , David Smith <sup>2</sup> and Anna Lishchuk <sup>3</sup>**


**Abstract:** The most common prostate cancer (PCa) diagnostics, which are based on detection of prostate-specific antigens (PSA) in blood, have specificity limitations often resulting in both falsepositive and false-negative results; therefore, improvement in PCa diagnostics using more specific PCa biomarkers is of high importance. Studies have shown that the long noncoding RNA Prostate Cancer Antigen 3 (lncPCA3) that is over-expressed in the urine of prostate cancer patients is an ideal biomarker for non-invasive early diagnostics of PCa. Geno-sensors based on aptamer bioreceptors (aptasensors) offer cost- and time-effective, and precise diagnostic tools for detecting PCa biomarkers. In this study, we report on further developments of RNA-based aptasensors exploiting optical (spectroscopic ellipsometry) measurements in comparison with electrochemical (CV and IS) measurements published earlier. These sensors were made by immobilization of thiolated CG-3 RNA aptamers on the surface of gold. Instead of a redox-labelled aptamer used previously in electrochemical measurements, a non-labelled aptamer was used here in a combination with total internal reflection ellipsometry (TIRE) measurements. The results obtained by these two methods were compared. The method of TIRE is potentially highly sensitive and comparable in that respect with electrochemical methods capable of detection of PCA3 in sub-pM levels of concentration. The required selectivity is provided by the high affinity of PCA3-to-aptamer binding with KD in the 10−<sup>9</sup> M range. The spectroscopic ellipsometry measurements provided additional information on the processes of PCA3 to aptamer binding.

**Keywords:** aptamer; mRNA; PCA3; biosensor; TIRE

#### **1. Introduction**

Prostate cancer (PCa) is considered as one of the most common types of cancer worldwide, and is the second leading cause of mortality among men after lung cancer [1,2]. There are clinical challenges for PCa early-stage diagnosis related to the asymptomatic nature of the disease and the similarity of its symptoms to benign prostatitis [3]. Early diagnosis of PCa can reduce mortality rates and increase the opportunity for effective medical interventions, therefore, the development of reliable diagnostics of PCa is of high importance [4,5]. Current diagnostics of PCa is based on the detection of total serum prostate-specific antigens (PSA) in blood followed by (if PCa suspected) digital rectal examination and imaging studies [6,7]. However, the lack of specificity of PSA markers often leads to both false-positive and false-negative results of the PSA test [8]. Hence, identifying

**Citation:** Takita, S.; Nabok, A.; Smith, D.; Lishchuk, A. Spectroscopic Ellipsometry Detection of Prostate Cancer Bio-Marker PCA3 Using Specific Non-Labeled Aptamer: Comparison with Electrochemical Detection. *Chem. Proc.* **2021**, *5*, 65. https://doi.org/10.3390/ CSAC2021-10453

Academic Editor: Maria Emília de Sousa

Published: 30 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

alternative specific prostate cancer biomarkers and developing methods for their detection in the early stage of the disease is of high importance nowadays [9,10]. Several PCa biomarkers have been identified as being over-expressed in prostate tumours [11]. The differential display code 3 (DD3) gene, also known as prostate cancer antigen 3 (PCA3), the long non-coding RNA (lncRNA) discovered in 1999 [12] has been widely accepted as one of the specific biomarkers for malignant PCa [13–15]. PCA3 levels can predict the prostatic biopsies' outcome, especially in combination with other PCa biomarkers, such as PSA, and can reduce the likelihood of false-positive results. The Prognesa® test based on simultaneous detection of PCA3 and PSA using quantitative nucleic acid amplification with high sensitivity and specificity is approved in USA [16]. However, such a test is time-consuming and expensive. The development of PCA3 biosensors for express, accurate and cost-effective diagnostics of PCa is a subject of high importance. Recent developments in biosensing technologies related to the use of aptamers, synthetic bioreceptors having specifically designed sequences of RNA or DNA oligonucleotides to provide the antibodylike function towards a wide range of analytes, leads to substantial progress in cancer diagnostics (including prostate cancer) [17,18]. The RNA-based aptamer (CG3 aptamer), with a high affinity towards 277-bases section of PCA3 transcript, has been developed recently [19]. The successful application of the CG3 aptamer functionalized with ferrocene at C5 terminal and immobilized on the surface of gold screen-printed electrodes via thiol group at C3 terminal for electrochemical in-vitro detection of PCA3 in low concentrations down to sub-pM range was reported for the first time in [20].

One of the most attractive optical biosensing technology developed in the last decade was the method of total internal reflection ellipsometry (TIRE) which is a combination of spectroscopic ellipsometry (SE) and surface plasmon resonance (SPR) [21]. The method of TIRE has a high sensitivity (10 times higher than conventional SRP) and thus is particularly attractive for detection of small molecules, such as mycotoxins, in concentrations down to ppt level [22,23]; it is also suitable for the study of adsorption kinetics and subsequent evaluation of the affinity of bioreceptors (antibodies and aptamers). This work is mostly focused on TIRE detection of PCA3 in direct assay with unlabelled CG3 aptamers immobilized on the surface of gold. The results are compared to our data of electrochemical detection of PCA3 using redox-labelled CG3 aptamer reported earlier in [20]. Our observations are a step towards the long-term aim of developing a novel, accurate, simple, and cost-effective diagnostic tool for the early detection of prostate cancer.

#### **2. Materials and Methods**

#### *2.1. Chemicals*

HEPES binding buffer (HBB) pH 7.2–7.6, sodium phosphate (Na2HPO4), potassium phosphate (KH2PO4), potassium chloride (KCl), magnesium chloride (MgCl2), dithiothreitol (DTT), and sodium chloride (NaCl), were procured from Sigma-Aldrich (UK). All reagents were of analytical grade. The biological target, the 277 nt target analyte fragment of lncRNA PCA3, was purchased from Eurofins Genomics (Germany). The label-free CG-3 RNA-based aptamer (5 -AGUUUUUGCGUGUGCCCUUUUUGUCCCC-3 SH) for optical measurements was acquired from Sangon-Biotech, China. The same aptamer, but labelled with ferrocene at 5 , was used previously [20] for electrochemical detection of PCA3. Before immobilization, the stock solution of aptamer (100 μM) was diluted at desired concentration with PBB supplemented with 2 mM of DTT, then diluted aptamer solution was activated by thermocycling in PCR unit (Prime TC3600) by heating to 90 ◦C for 5 min and cooling down to 4 ◦C for 5 min. Immobilization of aptamers was carried out in 100 M HEPES buffer (pH 7.4) with the addition of 2.5 mM DTT and 3 mM MgCl2. Optical measurements were performed in the detection buffer, e.g., 100 mM PBS (pH 7.4), prepared by dissolving 10 mM Na2HPO4, 1.56 mM KH2PO4, 2.5 mM KCl, and 135 mM NaCl in water. Milli-Q water was used for all preparations.

#### *2.2. TIRE-Optical Bio-Transducer*

The TIRE experimental set-up schematically shown on inset in Figure 1 is based on J.A. Woollam M2000 spectroscopic ellipsometer with the addition of a 68◦ glass prism (providing the light coupling at total internal reflection conditions) which was optically connected via index matching fluid with the gold coated glass slide. The PTFE cell of 0.2 mL in volume was sealed against the gold coated glass slide; the inlet and outlet tubes allow the injection of the required liquid reagents in the cell. The principles of total internal reflection ellipsometry (TIRE) measurements and data acquisition were described in detail previously in [21–24].

**Figure 1.** Results of TIRE spectra measurements: (**a**) TIRE spectra for Au layer with aptamers immobilized on the surface (solid line) and the same sample after binding 1 nM of PCA3 to aptamer (dotted line); (**b**) A series of TIRE D-spectra demonstrating the "blue" spectral shift caused by binding PCA3 of different concentrations: aptamer before exposure (1) and after exposure to PCA3 0.09 nM (2), 0.5 nM (3), 1 nM (4), 10nM (5), and 100 nM (6).

Standard microscopic glass slides were cleaned in hot piranha solution (3:1 mixture of H2SO4 and H2O2) for 10 min. followed by rinsing with deionized Milli-Q water and drying under a stream of nitrogen gas. Gold layers of about 20 to 25 nm in thickness were evaporated on glass slides using Edwards E306A metal evaporator unit; an intermediate layer of Cr (3 to 5 nm) was used to improve the adhesion of gold to glass. For TIRE measurements, gold coated glass slides functionalized with label-free aptamer were used.

The PCA3 solutions were prepared by diluting the original stock solution (100 μM) in PBS buffer to obtain the required concentrations of 0.09, 0.5, 1, 10, 100 nM. TIRE measurements were performed in a sequential adsorption manner (starting with the injection of lowest concentration of PCA3) and rinsing the cell after each adsorption step; the initial TIRE measurements of pure buffer were used as a reference. The TIRE setup allows two types of ellipsometric measurements: (i) single spectroscopic spectral scans performed in PBS after completing each stage of molecular adsorption, and (ii) dynamic measurements, e.g., recording of several spectroscopic scans during the binding of analytes (PCA3) to receptors (CG3 aptamer) which provides the information on the reaction's kinetics.

#### **3. Results and Discussion**

#### *3.1. Results of TIRE Single Spectroscopic Measurements*

Typical TIRE spectra of Ψ and Δ of Au/Cr layer on the glass slide functionalized with aptamers are shown in Figure 1. The maximum and minimum in the spectrum of the amplitude-related parameter Ψ correspond, respectively, to the conditions of total internal reflection (TIR) and surface plasmon resonance (SPR), while a sharp drop in a spectrum of phase related parameter Δ is a new quantity non-existent in traditional SPR. The position of such a phase drop is highly sensitive to changes in the optical density of a molecular layer adsorbed the surface of gold; the increase in the molecular layer thickness causes a "red" spectral shift, while thickness decrease causes a "blue" spectral shift of Δ spectrum.

As one ca see in Figure 1a, binding PCA3 from its 0.5 nM solution in PBS to its specific aptamer results in a blue shift of both Ψ and Δ spectra (dotted lines) as compared to the spectra of unperturbed aptamer layer (solid lines).

As shown in Figure 1b, the increase in concentration of PCA3 causes a progressive increase in the blue (negative) spectral shift until the saturation of binding sites, e.g., aptamers) occurs at concentrations larger than 1nM. The saturation of the TIRE sensor response at the level of about −9.5 nm shown as inset in Figure 1b corresponds to a decrease in the film thickness of around 2 nm. The negative control tests were carried out by adsorbing molecules having scrambled the sequence of PCA3, resulting in the "red" (positive) spectral shift of about 20 nm associated with the thickness increase of about 4 nm.

These results can be understood using the model shown in Figure 2 which schematically illustrates the process of PCA3 to aptamer binding during which the aptamer engulfs the target resulting in the thickness decrease. Contrarily, non-specific binding of the scrambled PCA3 to the aptamer results in the thickness increase. Interestingly, that prolonged exposure of aptamer to large concentrations of PCA3 (10 nM and 100 nM) also causes the red spectral shift which is due to the non-specific adsorption (or piling up) of PCA3 molecules. Another explanation of the saturation of the sensor response at relatively low concentrations of PCA3 is in high density of the immobilized aptamers which have no room for coiling around large target molecules of PCA3; optimization of the aptamer concentration is therefore required.

It is worth mentioning that the binding of 90 pM (the lowest concentration used) of PCA3 causes a substantial spectral shift of about 5 nm. Considering the high accuracy of ellipsometry measurements with the noise level of Ψ and Δ in the second decimal digits the actual limit of detection (LOD) could be much lower. In this work, however, the evaluation of LOD was not a priority.

Additional spectroscopic ellipsometry measurements were carried out on dried samples, in order to evaluate the thickness of the aptamer layer covalently bound on the surface of gold. The thicknesses of Au (17 nm to 20 nm) and Cr (5 nm to 7 nm) layers were evaluated first from the measurements on bare metal layers. The aptamer layer thickness of 2.5 nm was found. The thermocycling of gold coated slides with immobilized aptamer in

PCR unit (e.g., heating up to 95 ◦C and cooling down to 5 ◦C during 10 min.) has resulted in the aptamer layer thickness increase up to 4.5 nm. This shows that aptamer molecules tend to coil in a dry state, while thermocycling in a buffer solution containing Mg2+ ions stabilizes the aptamer structure in the original stretched form suitable for sensing. Such a procedure is recommended for "refreshing" the samples of aptamers immobilized on gold prior to conducting sensing tests.

**Figure 2.** Schematic diagram of specific and non-specific binding of molecules to unlabeled aptamers immobilized on the surface of gold.

#### *3.2. TIRE Study of the Binding Kinetics*

TIRE spectral measurements were carried out (at certain time intervals) during binding PCA3 to aptamers immobilized on the surface of gold. The resulted massive data files can be processed by plotting the time dependencies of either Ψ or Δ at fixed wavelength typically selected on the left side of the resonance (see red dotted line in Figure 1a). A typical example of TIRE binding kinetics of PCA3 (0.5 nM) to aptamers is given in Figure 3a as the time dependence of Ψ at 700 nm. These data were fitted to the rising exponential function with the parameters of the equation given as an inset. The parameter of interest was the time constant (τ). Such measurements were carried out at different concentrations of PCA3 (*C*) and the characteristic time constants (τ) were evaluated at each concentration. According to the theory of molecular adsorption [20], the rates of adsorption and desorption (*ka* and *kd*) can be found, respectively, as the gradient and intercept of the following linear equation: <sup>1</sup> *<sup>τ</sup>* = *kaC* + *kd*, then the association and affinity constants (*KA* and *KD*) can be found as *KA* = *ka*/*kd* , *KD* = 1/*KD*.

Linear dependence of 1/(τ) vs. *C* given in Figure 3b in both logarithmic and linear coordinates yields the values of *KA* = 3.87 × <sup>10</sup><sup>10</sup> <sup>M</sup>−<sup>1</sup> and *KD* = 2.58 × <sup>10</sup>−<sup>9</sup> M which are very similar to those obtained earlier by the electrochemical method of CV [20]. Also, the TIRE experiments revealed anomalous kinetics at high concentrations of PCA3 when soon after reaching the saturation the response started to rise again. This is most likely associated with the non-specific adsorption of PCA3.

**Figure 3.** Evaluation of the PCA3 to aptamer binding affinity from dynamic TIRE measurements: (**a**) example of Ψ time dependence upon binding PCA3 (0.5 nM) to aptamer immobilized on the surface of Au; The values of Ψ at 700 nm were presented; (**b**) evaluation of KA and KD from the 1/τ(C) dependence given in logarithmic and linear scales.

#### **4. Conclusions: Comparison of the Electrochemical and Optical Detection Strategies**

The use of the TIRE method for detection of the lncRNA transcript PCA3 in direct assay with non-labelled aptamer immobilized on the surface proved to be promising. In contrast to electrochemical detection based on labelled aptamers, the optical detection using TIRE showed the increase in the molecular layer thickness caused by the non-specific binding of PCA3 which cannot be detected with the electrochemical method. In terms of sensitivity, the method of TIRE is potentially capable of detection of PCA3 in low concentrations (much lower than 0.09 nM used in this work) which could be comparable with the values of LOD reported earlier for electrochemical CV (0.35 pM–0.78 pM) and EIS (0.26 pM) methods [20]. More detailed optical study in a wider concentration range of PCA3 and optimization of aptamer concentration is currently underway. The high affinity of unlabeled aptamer towards PCA3 was confirmed by TIRE kinetics study which gave similar values KA and KD to those obtained previously from CV measurements for an aptamer labelled with ferrocene [20].

From the point of view of sensing, electrochemical methods are more attractive because of the low cost and simplicity of use, however the optical method of TIRE provides important complementary information on the thickness of molecular layers which allows for better understanding of the processes of aptamer-target interaction.

**Author Contributions:** Conceptualization A.N.; methodology, A.N. and A.L.; validation, S.T.; formal analysis, A.N.; investigation, S.T.; resources, S.T. and A.N.; writing—original draft preparation, A.N. and S.T.; writing—review and editing, A.N. and D.S.; supervision, A.N., A.L. and D.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data are not publicly available; The data files are stored on corresponding instruments and on personal computers.

**Acknowledgments:** We would like to acknowledge Material and Engineering Research Institute at Sheffield Hallam University for providing full access to its resources and materials.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Proceeding Paper* **Titanium Based Materials for High-Temperature Gas Sensor in Harsh Environment Application †**

**Roussin Lontio Fomekong 1,\* and Bilge Saruhan <sup>2</sup>**


**Abstract:** Gas selective sensors that are capable of sensing at hot-gas environment are increasingly gaining importance for the monitoring of combustion and thermal processes releasing NO, NO2 and H2 containing hot gases at temperatures exceeding 600 ◦C. Despite some drawbacks, TiO2 is capable of operating as a gas sensor above 500 ◦C. In this context, Ni-doped TiO2, Co-doped TiO2 and Rh-doped BaTiO3 have been prepared by oxalate coprecipitation route and fully characterized. Co-doping of TiO2 promotes p-type behavior exhibiting good sensing properties to NO2 while Ni-doping displays the maintenance of n-type behavior and better H2-sensing properties at 600 ◦C. Rh-doped BaTiO3 shows excellent NO sensing properties at 900 ◦C.

**Keywords:** Ni-doped TiO2; Co-doped TiO2; Rh-doped TiO2; coprecipitation; high-temperature gas sensor

**Citation:** Fomekong, R.L.; Saruhan, B. Titanium Based Materials for High-Temperature Gas Sensor in Harsh Environment Application. *Chem. Proc.* **2021**, *5*, 66. https:// doi.org/10.3390/CSAC2021-10480

Academic Editor: Huangxian Ju

Published: 30 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

#### **1. Introduction**

High temperature gas sensors are mainly designed to solve gas detection and monitoring problems with a high operating temperature environment, such as gas turbine, nuclear power plants and automobile internal combustion engine emission [1]. As far as target gases are concerned NOx (NO2, NO) and H2 are among the most important. NOx is a severely toxic gas with a pungent odor arising mostly from the related human activities involving high temperature such as combustion of coal and oil at electric power plants, combustion of chemical plant and also in emissions from automotive and aircraft engines. NOx-emission leads to harmful effects on the environment and health. There is an urgent need to develop sensor control systems for exhaust emission gases to directly monitor NOx at temperatures in the range of 400–900 ◦C [2]. As far as hydrogen is concerned, it is the best candidate to replace the hydrocarbon-based fuels used in many combustion engines such as those in automobiles and aircraft, which are responsible for much of today's air pollution [3]. Hydrogen seems to be a green, renewable energy carrier that can help solve the problems of non-sustainable energy use (fossil fuels). However, the efficient application of hydrogen requires careful consideration of the relevant safety concern. In fact, its physico-chemical properties make hydrogen a highly explosive gas [3,4]. Moreover, as hydrogen is colorless, odorless and tasteless, the ability to detect a hydrogen leak by means of selective sensors is highly desired.

Chemiresistive gas sensors based on semiconductor metal oxides have been drawing more and more attention because of their advantages such as low cost, lightweight, fast response/recovery times and high compatibility with microelectronic processing. Graphene and its derivatives, as well as some organic semiconductor-based materials constitute also an interesting family of chemo-resistive gas sensor due to their large surface area, good electrical, thermal and mechanical properties [5,6].

Cost effective metal oxide-based gas sensors such as SnO2, WO3, ZnO, NiO or CuO operate mostly at temperatures below 400 ◦C [7–10]. There are only few reports in literature focusing on their gas sensing above 400 ◦C. TiO2 is one of them to be capable of operating above 500 ◦C. The additional benefits of TiO2 are non-toxicity, easy fabrication, and the good chemical stability [11]. However, TiO2 is a high resistive n-type semiconductor with relatively poor conductivity for sensing oxidative gases such as NO2. This disadvantage was previously reported to be overcome through addition of low valence dopant atoms which alter the electronic structure [12–15]. Another strategy is to use catalytically doped perovskite-based titanium compounds such as BaTiO3. In this work, we report the synthesis of Co-doped TiO2, Ni doped TiO2 and Rh-doped BaTiO3 by co-precipitation method and demonstrate gas sensing ability toward NO2, NO and H2 above 500 ◦C. Our results yield that Co-doping of TiO2 promotes p-type behavior exhibiting good sensing properties to NO2 while Ni-doping displays the maintenance of n-type behavior and better H2-sensing properties at 600 ◦C. More interestingly, Rh-doped BaTiO3 shows excellent NO sensing properties even at 900 ◦C.

#### **2. Materials and Methods**

#### *2.1. Preparation of the Different Sensing Materials*

Ni-doped TiO2 nanoparticles were prepared using the co-precipitation synthesis route followed by calcination. The starting precursor solutions were first prepared by dissolving nickel acetate in acetic acid while pure ethanol was used separately to dissolve titanium iso-propoxide (TTIP). The adjustment between the previously prepared nickel and titanium solution has been performed so that the final obtained mix would contain 0.0, 0.5, 1.0 and 2.0 mol.% of nickel dopant in TiO2 and were labeled as TN0, TN05, TN1 and TN2, respectively. The solutions were then mixed and stirred for 5 min. Oxalic acid was used as the precipitating agent. It was dissolved in absolute ethanol solution and poured progressively into the previously mixed solutions. In order to achieve a total precipitation, the resulting mixtures were stirred for 1 h at room temperature, followed by the filtration and drying of the obtained precipitate at 80 ◦C. A white powder was obtained for the undoped precursor and a yellowish colored powder for the Ni-doped precursor samples. The prepared precursor powders were then calcined in a muffle furnace under static air for 3 h at 700 ◦C to obtain the nano-particulate powders.

Co-doped TiO2 nanoparticles were synthesized by employing the same processing route. The cobalt acetate and titanium iso-propoxide solutions were adjusted to obtain 0.5 and 1 mol.% of cobalt in TiO2 and were labeled as 0.5Co-doped TiO2 and 1Co-doped TiO2, respectively. The so-obtained precipitate was filtered and dried in oven at 80 ◦C yielding a pink colored powder for the Co-doped samples.

The synthesis of Rh-doped BaTiO3 (designated as BTR1-OX-900) was also prepared by coprecipitation by following the same procedure. The amount of the aqueous Rh-nitrate solution, barium acetate and titanium iso-propoxide were adjusted to yield perovskites with the following composition: BaTi0.98Rh0.02O3. The as-prepared precursor powder was calcined in a ceramic combustion boat holder at 900 ◦C in a muffle furnace (5 ◦C min−1) for one hour under static air. In order to activate this material, it was treated under 2% of hydrogen at 900 ◦C for two hours.

#### *2.2. Materials Characterization*

The XRD diffractograms of all the samples were obtained at room temperature with a D5000 Siemens Kristalloflex θ–2θ powder diffractometer which has a Bragg-Brentano geometry and equipped with Cu-Kα radiation (λ = 1.54178 Å) and a standard scintillation counter detector.

Bruker Senterra Raman spectrometer (from Bruker Optik GmbH, Ettlingen, Germany) was used to record all the Raman spectra at room temperature under 532 nm and 0.2 mW power laser excitation, which was focused on samples through a 50X objective (Olympus MPlan N 50X/0.75).

The particles' morphology was determined by Scanning Electron Microscopic (SEM) analysis and was carried out in a Zeiss Ultra 55 microscope.

#### *2.3. Sensor Preparation*

The as prepared materials (Co-doped TiO2, Ni-doped TiO2 and Rh-doped BaTiO2) powders were deposited as thick films (~20 μm) using a simple drop-coating method on alumina substrates that were previously fitted with interdigitated electrodes. The sensor response for n-type semiconductors is defined by (Rgas/Rair − 1) × 100 and (Rair/Rgas − 1) × 100 for oxidizing and reducing gases, respectively, while for p-type semiconductor, (Rgas/Rair − 1) × 100 and (Rair/Rgas − 1) × 100 for reducing and oxidizing gases respectively. R is the electrical resistance of the sensor material in air (Rair) or in gas (Rgas)

#### **3. Results and Discussion**

The phase identification derived from powder XRD analysis is presented on Figure 1. As can be observed in Figure 1a for Ni-doped TiO2, the results indicate that anatase is present in those from undoped to 1% Ni-doped TiO2, while rutile is the only phase present in the 2% Ni-doped TiO2 powders. The amount of anatase (JCPDS 21-1272) decreases as the amount of nickel increases while the amount of rutile (JCPDS 21-1276) follows the opposite trend. A trace amount of ilmenite, NiTiO3 is observed at 32.5◦ in the TN1 and TN2 samples (JCPDS 33-0960). These results indicate that Ni promotes the transition of anatase to rutile. As far as Co-doped TiO2 is concerned, the results depicted on Figure 1b reveal that, the undoped and the 0.5Co-doped TiO2 samples showed pure TiO2 consisted of its two polymorphs; anatase (majority) and rutile phases (minority). No other phase containing Co was observed. The 1Co-doped TiO2 sample showed only single phase TiO2 but this time the rutile polymorph was the major phase as anatase phase was in trace amount. The X-ray results indicate that the cobalt dopant promotes also the anatase-to-rutile phase conversion of TiO2 but not progressively as nickel. The results of XRD analysis performed on the Rh-doped BaTiO3 (before and after the activation) are presented in Figure 1c. As shown, the major phase is BaTiO3 (according to JCPDS 075-0462) and a trace amount of BaCO3 phase is observed. No evidence of an Rh phase was encountered indicating the substitution of Ti by Rh.

**Figure 1.** (**a**) Powder XRD patterns of Ni-doped TiO2, (**b**) Co-doped TiO2 and (**c**) Rh-doped BaTiO3.

Figure 2 shows the morphology of the synthesized powders investigated by SEM. Concerning Ni-doped TiO2 (Figure 2a), as far as the undoped powder sample is concerned, the microstructural investigation reveals spherical nanoparticles with sizes around 70 nm, which tend to agglomerate. While the TN05 sample shows more agglomerated spherical particles, the TN1 and TN2 samples show less agglomeration with the appearance of small pores and particle size reduction for the TN2 samples. Figure 2b shows the morphology of the Co-doped TiO2 synthetized powders. The SEM investigation revealed the formation of spherical nanoparticles with sizes around 70 nm. As the sample 0.5Co-doped TiO2 shows more agglomerated spherical particles, the sample 1Co-doped TiO2 presented larger and

well-faceted rhombohedral crystallites with less agglomeration. As SEM pictures display in Figure 2b the Rh-doped BaTiO3 has the well-defined and homogeneously distributed spherical nanoparticles (~50 nm).

**Figure 2.** SEM images of (**a**) undoped and Ni-doped TiO2 powders and (**b**) Co-doped TiO2 and Rh-doped BaTiO3.

The Raman spectra of the samples were obtained between the wavenumbers of 175–800 cm−<sup>1</sup> and the results are presented in Figure 3. The results of Ni-doped TiO2 Raman analysis (Figure 3a) show that the main signals came from TiO2. The samples TN0 and TN05 show very strong Raman signals, with peaks at 196(Eg), 396(B1g), 517(A1g) and 638(B1g) cm−<sup>1</sup> from the typical anatase TiO2 phase [16]. A weak peak at 447 cm−<sup>1</sup> (Eg) attributed to rutile is observed in TN05. Samples TN1 and TN2, which contain larger amounts of Ni, present the Raman signals corresponding to both the anatase and rutile (447(Eg), 612(A1g) cm−1) phase [17]. In addition to the anatase and rutile phases, another set of Raman vibrations emerges in the TN1 and TN2 samples. The peaks at 244, 345 and 706 cm−<sup>1</sup> are assigned to a trace amount of ilmenite, NiTiO3. In the Figure 3b showing the results of Co-doped TiO2, the Raman lines observed at 197, 390, 511, 637 cm−<sup>1</sup> can be assigned to Eg, B1g, A1g, or B1g and Eg modes of anatase phase respectively. The spectra show that the peak intensities decrease drastically after doping, due certainly to the decrease of the amount of anatase and the formation of rutile as indicated by XRD. Moreover, the Raman spectra of 1Co-doped TiO2 yielded a smaller shift towards lower wavelengths while new peaks (436 cm−1) appeared indicating the presence of rutile polymorph, as mentioned in literature [17]. Raman spectra of Rh-doped BaTiO3 are displayed in Figure 3c. It shows the peaks at 270, 308, 525 and 725 cm−<sup>1</sup> which are assigned respectively to the A1(TO2), E(TO2), A1(TO3), and A1(LO3) of barium titanate modes of the room temperature P4mm phase.

Based on our previous results on undoped, Al and Cr-doped TiO2, 600 ◦C was chosen as the optimum sensing temperature for Ni and Co doped TiO2 in this work [12,13].

**Figure 3.** (**a**) Raman spectra of undoped and Ni-doped TiO2, (**b**) Co-doped TiO2 and (**c**) Rh-doped BaTiO3.

The responses of the undoped TiO2 and all Ni-doped TiO2 towards 10,000 ppm H2 in dry synthetic air at 600 ◦C are shown in Figure 4a. The sensor responses are 42, 72, 70 and 62% for TN0, TN05, TN1 and TN2, respectively. It can be observed that the sensor response increases greatly as Ni-content increases up to 0.5 mol.% and then decreases slowly with further increase in the Ni-content to 2.0 mol.%. This implies that the sensor reaches its maximum response of 72% with 0.5 mol.% of Ni dopant. It can be assumed that this enhancement of gas sensor response may be due to the formation of a n-n junction between the anatase (Eg = 3.2 eV) and rutile (Eg = 3.0 eV) phases. As revealed by XRD results, the TN05 sample contains almost the same amount of anatase and rutile phases (which is not the case with the other samples in this work), and thus, the highest amount of n-n junctions are expected to be present in this sample. This kind of junction effect has also been reported for other n-n junction systems such as ZnO-SnO2 [18] and SnO2-WO3 [19]. The achievement of a great selectivity towards the target gas is a key parameter and a very important characteristic. Therefore, the responses of the TN05 gas sensor towards a variety of interference gases including NO2, CO, and NO at 600 ◦C in dry synthetic air were explored to evaluate its selectivity. As observed in Figure 4b, the response of this sensor towards 600 ppm of H2 (35%) is at least a factor of two higher than that towards 300 ppm of CO (12%), 300 ppm of NO2 (11%) and 300 ppm of NO (7%). The sensor's H2-response was a factor of two greater compared to that for 300 ppm of all the tested interfering gases, yielding the highest value and indicating a relatively high selectivity potential of the sensor towards H2.

**Figure 4.** (**a**) Response of gas sensors based on undoped TiO2 and Ni-doped TiO2 to 10,000 ppm of hydrogen gas at the optimum operating temperature of 600 ◦C. (**b**) Response of gas sensor based on 0.5% Ni-doped TiO2 to various gases including 600 ppm of H2, 300 ppm of NO2, NO and CO all in dry air at 600 ◦C.

As Figure 5a shows, the sensors yield higher response towards H2 than NO2. The H2 sensor responses are 42, 23 and 33% for undoped, 0.5Co-doped TiO2 and 1Co-doped TiO2, respectively. The undoped sample shows the highest response toward H2. The presence of Co-dopant seems to decrease the H2-sensing performance of TiO2 even though doping creates more oxygen vacancies in TiO2. This behavior can be attributed to the increase of rutile polymorph content on doping with cobalt, as previously reported, rutile is the less active (in term of functional properties) polymorph of TiO2 [20]. On the other hand, the 1Co-doped TiO2 which contains predominantly rutile polymorph showed a higher response toward H2 than 0.5Co-doped TiO2. This discrepancy can be explained by alteration of the conductivity from n-type to p-type. In fact, the dynamic response of the sensors toward H2 given in Figure 5b reveals that undoped and 0.5Co-doped TiO2 exhibit n-type semi-conductivity (i.e., their electrical resistance decreases upon interaction with hydrogen) while the 1Co-doped TiO2 yields p-type conductivity (its electrical resistance increases when reducing gas is introduced). The sensor responses measured toward the oxidizing gas NO2 were 5, 3 and 8% for undoped, 0.5Co-doped TiO2 and 1Co-doped TiO2 respectively. In the case of NO2 sensing, the 1Co-doped TiO2 showed the highest response. This may be mainly due to the electronic alteration of TiO2 from n to p-type semiconductor. Previous literature points out that this alteration can be utilized for the detection of oxidizing gas. Our current results confirm that the dominant factor for the gas sensing property of the Co-doped TiO2 depend on the existing polymorphs as well as the nature of target gas (oxidizing or reducing). In the case of reducing gases, the type of polymorphs has more influence on the gas sensitivity than the type of electronic structure, while an opposite trend can be observed for oxidizing gases.

**Figure 5.** (**a**) Sensor response of undoped and Co-doped TiO2 towards NO2 and H2 at 600 ◦C and (**b**) their dynamic responses towards H2.

Figure 6a shows the sensor responses of the hydrogen treated Rh-doped BaTiO3 towards 200 ppm of nitrogen oxide (NO) at a different operating temperature under dry and humid (10% of RH) synthetic air and the dynamic response at 700 and 900 ◦C, respectively. The sensor responses are 14, 2, 6, 7% in dry air, 6, 12, 16 and 18% in humid

air at 600, 700, 800, and 900 ◦C, respectively. In dry air, the sensor response decreases from 14 to 7% in general with the increasing temperature, while in humid air, the sensor response increases from 6 to 18% as the temperature increases. The maximum sensor response is therefore obtained at 900 ◦C under humidity. This enhancement of sensing properties in the presence of humidity can be explained by the affinity between adsorbed hydroxyl group (generated after thermal decomposition of water) and the NO. In fact, at high temperatures, H2O in water vapor decomposes, and hydroxyl is adsorbed on the sensing layer. As the temperature increases, more decomposition will occur, and more hydroxyl groups will be adsorbed on the surface, enhancing the NO sensor response. To the best of our knowledge, this is the first time that NO detection is reported at such a high temperature in the humid. Therefore, we have investigated intensely further gas sensing characteristics of this material under these extreme conditions (e.g., at 900 ◦C under humid air). Two other main products for fuel combustion are NO2 and CO. Their presence in the exhaust gas stream at a high temperature can cause important hinderance for the NO gas sensing application. Our sensor's selectivity toward NO against CO and NO2 at 900 ◦C under humid air was investigated. Figure 6b shows the different responses of the sensor to 200 ppm of NO, NO2 and CO. The results indicate that at 900 ◦C, the response to 200 ppm of NO (18%) is higher than that of 200 ppm of NO2 (8.7%) and 200 ppm of CO (8.4%). This implies that this sensor is at least twice as much sensitive to NO than NO2 and CO. This good selectivity is ascribed to the catalytic effect of Rhodium-NPs on the oxidation of NO, which will promote and enhance the adsorption and the oxidation of NO preferentially. It is reported in the literature that Rhodium which is currently and often used in TWC, is a suitable catalyst for NO oxidation [21].

#### **4. Conclusions**

This paper reports the successful synthesis of Ni-doped TiO2, Co-doped TiO2 and Rh-doped BaTiO3 nanoparticles by a facile co-precipitation route through the use of oxalic acid. Their NO2, NO and hydrogen sensing properties at high temperatures (≤600 ◦C) were investigated. According to structural characterization the substitution of Ti4+ by dopant (Ni2+, Co3+ and Rh3+) was effective and creates more oxygen vacancies which promotes the anatase-to-rutile transformation in the case of Ni and Co doped TiO2. Enhanced sensing properties with respect to H2 were observed for 0.5% Ni-doped TiO2 in comparison to undoped and 1 and 2% Ni-doped TiO2. The sample 1Co-doped TiO2 which reveals p-type conductive behavior yields an enhanced NO2 response at 600 ◦C under air as carrier gas. With Rh-doped BaTiO3, it was possible to detect NO at 900 ◦C under humid air with a good response (18% for 200 NO ppm) and good selectivity (twice as much sensitive to NO than CO and NO2). Titanium based materials appear as a promising high temperature gas sensor in harsh environment.

**Author Contributions:** R.L.F.: conceptualization, data curation, formal analysis concerning Raman spectroscopy and SEM analysis, methodology and investigation concerning sensor testing, writing—original draft preparation; B.S.: supervision, project administration, funding acquisition, validation, visualization, writing—review and editing. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by DAAD-DLR postdoc fellowship program, grant number 284.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data is contained within the article.

**Acknowledgments:** The authors thank Shujie You of the Luleå University of Technology in Sweden for the Raman spectroscopy measurements.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Proceeding Paper* **Nutritional Composition of the Atlantic Seaweeds** *Ulva rigida***,** *Codium tomentosum***,** *Palmaria palmata* **and** *Porphyra purpurea* **†**

**Javier Echave <sup>1</sup> , Catarina Lourenço-Lopes <sup>1</sup> , Anxo Carreira-Casais <sup>1</sup> , Franklin Chamorro <sup>1</sup> , Maria Fraga-Corral 1,2 , Paz Otero 1, Pascual Garcia-Perez <sup>1</sup> , Sergio Baamonde 3, Fermín Fernández-Saa 3, Hui Cao 1, Jianbo Xiao 1, Miguel A. Prieto 1,2,\* and Jesus Simal-Gandara 1,\***


**Abstract:** Macroalgae are regarded as a healthy food due to their composition and nutritional properties. In this work, nutritional composition of two green (*Ulva rigida*, *Codium tomentosum*) and two red (*Palmaria palmata*, *Porphyra purpurea*) edible seaweed was studied. Total lipids were measured gravimetrically as evaporated mass after petroleum-ether Soxhlet extraction of samples. In addition, fatty acid profile was determined by gas chromatography coupled to a flame ionization detector (GC-FID). Results showed that all studied species were accounted for very low levels of lipids (<1% dw), but levels of unsaturated fatty acids oleic, linoleic, and linolenic acids were present at high concentrations, with *P. palmata* displaying the highest quantities (>200 mg C18:1/g extract). In parallel, proteins were quantified following the macro-Kjeldahl method. In this analysis, red algae, especially *P. purpurea*, showed significant protein content up to 30% DW. Total organic acids were found by ultra-filtration liquid-chromatography coupled to an amperometry detector (UFLC-PAD) after an acid extraction, *P. purpurea* being the algae with the higher organic acid content (10.61% dw). Minerals were identified and quantified by inductively coupled plasma atomic emission spectroscopy (ICP-OES), suggesting that both algae groups are rich in K and Mg (>15 g/kg), but *U. rigida* also displayed a remarkable iron content (>1 g Fe/kg). Other detected minerals in minor concentrations were Ca, P or F. Altogether, results corroborate that these edible algae are a good source of nutrients in accordance with literature.

**Keywords:** macroalgae; nutrition; composition; chromatography; minerals

#### **1. Introduction**

Seaweeds (macroalgae) are common ingredients in East Asian cuisine, of which many species such as wakame (*Undaria pinnatifida*), sweet kelp (*Saccharina latissima*) or nori (*Porphyra purpurea*) are used in different dishes. Seaweeds have long been recognized as healthy foods owing to their low caloric index and high content in dietary fiber, minerals and antioxidant molecules such as their cell wall polysaccharides [1,2] Besides, in recent years, there has been an increasing consumer interest in vegetarian food sources. In this context, algae could be a valuable alternative source of essential macronutrients. Indeed, seaweeds have been proposed as an alternative ingredient for the formulation of nutritional supplements that could cover various dietary needs [3]. One of the key

**Citation:** Echave, J.; Lourenço-Lopes, C.; Carreira-Casais, A.; Chamorro, F.; Fraga-Corral, M.; Otero, P.; Garcia-Perez, P.; Baamonde, S.; Fernández-Saa, F.; Cao, H.; et al. Nutritional Composition of the Atlantic Seaweeds *Ulva rigida*, *Codium tomentosum*, *Palmaria palmata* and *Porphyra purpurea*. *Chem. Proc.* **2021**, *5*, 67. https://doi.org/10.3390/ CSAC2021-10681

Academic Editor: Huangxian Ju

Published: 14 July 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

nutritional components is protein and the amino acid composition of a food protein. It is known that red seaweeds account for a protein content between 20 and 47% of its dry weight (dw), while green algae generally contains about 9–26% and brown seaweeds 3 to 15% [4]. From a nutritional perspective, seaweed proteins are also valuable since their content in essential amino acids is generally higher (~50%) than legumes (~40%) [5]. Seaweeds have also been generally described to contain very low levels of lipids usually between 1 and 4% dw, but nonetheless rich in polyunsaturated fatty acids (PUFA) [6]. Considering their mineral composition, seaweeds tend to hold much higher content of potassium, magnesium or calcium than several terrestrial plants. However, they are also described to generally accumulate iodine in great amounts of which an excessive intake could be hazardous to thyroid function [7]. In some cases, hazardous levels of arsenic have also been reported, which requires monitoring and assessment upon consumption of certain species [8]. Nonetheless, the nutritional composition of several seaweed species considering their growing region remains to be described, especially considering traditional methods for determining proximate compositions. In this work, nutritional composition of edible seaweed *Ulva rigida* (UR), *Codium tomentosum* (CT), *Palmaria palmata* (PA) and *Porphyra purpurea* (PU) widely distributed in Atlantic shores was studied using standardized analytical methods.

#### **2. Material and Methods**

#### *2.1. Sample Preparation*

Algae samples were provided by Algas Atlánticas Algamar S.L company (www. algamar.es, accessed on 1 July 2021) located in Pontevedra, Spain. The algae samples were collected from the coasts of the Galician region, Pontevedra province (NW Spain), washed with distilled water, frozen at −80 ◦C and freeze-dried afterwards. The seaweed samples were then crushed and grinded to obtain a homogeneous matrix, which was stored at −20 ◦C for further analysis.

#### *2.2. Proximate Composition*

Proximate composition was studied following AOAC guidelines (1995) [9]. All thermogravimetric analyses were carried out with a SETSYS Evolution thermobalance (Setaram, Caluire-et-Cuire, France). Results are expressed as g per 100 g dw.

Humidity was determined by a gravimetric method. 1 g of sample was deposited in a previously dried (104 ◦C/24 h) ceramic crucible. Next, the crucibles with fresh sample were placed in an oven at 104 ◦C for 24 h. After that time, they were weighed again, and humidity was calculated as the weight difference.

#### 2.2.1. Inorganic Material

To find the content of inorganic material (Ash), 250 mg were placed in a porcelain crucible (previously weighed) and the samples were incinerated at 600 ± 15 ◦C for 5 h. The crucible was then weighed with the resulting sample content. The obtained difference in weight was calculated as the ash value for each sample.

#### 2.2.2. Protein Content

Protein content was determined according to the macro-Kjeldahl method [10]. Briefly, 500 mg of sample were placed in a Kjeldahl tube, then adding a catalytic tablet (Sigma Aldrich, St. Louis, MO, USA) and 20 mL of sulfuric acid. The tubes were placed in a digestive block and the temperature was gradually increased to 400 ◦C for 70 min. The tubes were then removed, allowed to briefly cool, and 25 mL of distilled H2O was added. The nitrogen (N) converted to ammonia was measured with a macro-Kjeldahl distiller. The resulting N value was multiplied by a correction factor of 6.25 to obtain the estimate of the protein content, an extensively used correction factor for algae N-to-protein determinations [11].

#### 2.2.3. Lipids

For total lipids determination, 3 g of sample were placed inside a paper cartridge. An extraction with petroleum ether was then conducted through a ST 243 SOXTEC Soxhlet extraction system (Foss, Hillerød, Denmark) at a constant temperature of 120 ◦C for 7 h. The resulting product was transferred to a ground test tube, previously weighed, and placed in the oven for evaporation of the solvent. After solvent evaporation, the tube was weighed again to obtain, by difference, the total lipids content.

#### 2.2.4. Fiber and Hydrocarbons

Fiber was determined following the gravimetric AOAC method [9]. Briefly, 1 g of dried sample was sequentially treated with α-amylase from *Bacillus licheniformis* (pH 6, 30 min, 37 ◦C), protease from *Bacillus licheniformis* (pH 7.5, 30 min, 37 ◦C) and amyloglucosidase from *Aspergillus niger* (pH 4.5, 30 min, 40 ◦C). The obtained residue was precipitated with 4 times its volume in ethanol and filtered through a 0.45 μm paper syringe filter. The obtained difference in weight was calculated as total fiber.

Total hydrocarbons were calculated as the difference of the rest of the components, following Equation (1) and the results expressed as % (g/100 g dw) [12]:

$$\text{Hydrocorborus} = 100 - \left(\text{Lipids} + \text{Proteins} + \text{Ash} + \text{Fiber}\right) \tag{1}$$

#### *2.3. Organic Acids*

To determine the organic acid content, 1 g of each sample was weighed, and an extraction was carried out with 25 mL of 4.5% metaphosphoric acid, while stirring for 20 min. It was then filtered through paper and nylon (0.22 μm) to be able to work in ultra-fast liquid chromatography coupled to a photodiode array detector (UFLC-PAD).

The analysis was performed using a Shimadzu 20A series UFLC (Shimadzu, Kyoto, Japan) Separation was achieved on a SphereClone (Phenomenex, Torrance, CA, USA) reverse phase C18 column (5 μm, 250 × 4.6 mm) at 35 ◦C. Sulfuric acid 3.6 mM was used as mobile phase with a flow rate of 0.8 mL/min. Detection was carried out using wavelengths between 215 and 245 nm. Detected organic acids were quantified by comparison of the area of their peaks with calibration curves obtained by comparison to an ascorbic acid standard (Sigma Aldrich, St. Luois, MO, USA).

#### *2.4. Mineral Content*

For mineral detection and quantification, 2 g samples were subjected to a metaphosphoric acid digestion for 10 min and analyzed afterwards with optic emission spectrometry with inductively coupled plasma (ICP-OES) using an Optima 4300 DV instrument (Perkin Elmer, Whaltman, MA, USA) [9]. Briefly, quantification of minerals was determined following detection in specific wavelengths for Ca (317.9 nm), Mg, (285.2 nm), Cl, (134.7 nm), Fe (248.6 nm), Mn (279.4 nm), Zn (206.2 nm), K, (769.9 nm), I (178.2 nm), F, (685.6 nm), As (188.9 nm), P (213.6 nm) with RF power of 1450 W and at an argon plasma flow of 15 L/min. Results are expressed as g/kg dw.

#### *2.5. Fatty Acid Profile*

To carry out this determination, the product resulting from the lipids Soxhlet extraction was used and a derivatization process was carried out to obtain fatty acid methyl esters (FAME). 5 mL of reagent A (MeOH, H2SO4 and C7H8) was added in a 2:1:1 ratio and they were kept in a bath at 50 ◦C while stirring at 160 rpm for 12 h. Afterwards, 3 mL of distilled H2O was added, then adding 3 mL of diethyl ether under vigorous and continuing stirring until a homogeneous sample was obtained. Later, the two phases separation was allowed to occur, and the supernatant was transferred to a vial with sodium sulfate. The contents of the vial were filtered through 0.22 μm nylon prior to their chromatographic analysis by gas chromatography coupled to an infrared detector (GC-FID). The GC system was an Agilent 7820A and an Agilent HP-88 (60 m, 250 μm × 0.25 μm) column was used (Agilent Technologies, Santa Clara, CA, USA). Helium was used as carrier; 1μL of sample was injected. Oven temperature program started at 120 ◦C, increasing to 175 ◦C at 10 ◦C pre min. rate and hold for 10 min. Then, temperature was further increased to 220 ◦C at 3 ◦C increase per min. and kept at 220 ◦C for 5 min.

Different fatty acid levels were determined by comparing the relative retention times of the FAME peaks of the algae samples with respect to a commercial standard of FAME mix (Supelco 37 Component FAME MIX, Sigma Aldrich, St. Luois, MO, USA).

#### **3. Results and Discussion**

#### *3.1. Proximate Composition*

All studied species were accounted for very low levels of lipids (<1% dw), which could be due to the harvesting season or an incomplete extraction, as other studies report generally higher lipid content in some closely related species [7].

Green seaweeds displayed protein levels around 15% dw (Figure 1). Red algae, especially *P. purpurea*, showed a significant protein content, up to 30% dw. This is in contrast with other analyses reported on both *P. purpurea* and *P. palmata*, which displayed higher protein content [7,13]. Organic acids content was significantly heterogeneous, with only *P. purpurea* showing a high content (10.61% dw), half of which was determined as citrate. No organic acids were detected in *U. rigida*. Almost all the analyzed species showed more than 40% dw of insoluble fiber, with similar results to those generally reported in literature [5,13]. On the other hand, inorganic matter was somehow homogeneous for all seaweeds except for *C. tomentosum*, which showed a 37% dw ash content.

**Figure 1.** Proximate composition of the studied seaweed species (% dw). UR, *Ulva rigida*; CT, *Codium tomentosum*; PA, *Palmaria palmata*; PU, *Porphyra purpurea*; Prot, proteins; Hdrc., hydrocarbons; Org. Ac., organic acids.

#### *3.2. Mineral Composition*

The main minerals detected were Cl, K and Mg, of which *P. palmata* significantly outstand with as much as 100 g/kg dw of Cl and K, followed by *U. rigida* (Figure 2A). Indeed, the latter was accounted for the highest levels of Mg (22.9 g/kg dw), whereas *C. tomentosum* displayed the lowest mineral concentrations for Cl, Mg or K. *P. purpurea* displayed 33.5 g/kg dw of K and 12.1 g/kg dw of Cl, but the rest of its minerals were found at levels below 10 g/kg dw.

**Figure 2.** Mineral composition of the studied seaweed species in: (**A**), in major concentrations (g/kg dw); and (**B**), in minor concentrations (g/kg dw). UR, *Ulva rigida*; CT, *Codium tomentosum*; PA, *Palmaria palmata*; PU, *Porphyra purpurea*.

Minerals detected at minor concentrations are displayed in Figure 2B. Ca levels for *C. tomentosum* and *P. palmata* were higher than those found in terrestrial plants. Conversely, out of the four studied species, *P. purpurea* displayed the highest P content, a feature that may be related to its higher protein content (Figure 1). It is worth noting the remarkably high Fe content of *U. rigida*, as it is several times higher than those reported in terrestrial plants. Considering this result, *U. rigida* may contain higher iron quantities than legumes, which are considered one of the main sources of this mineral [3]. On the other hand, iodine levels from the studied sampled species appear lower [14]. Although this could be due to experimental errors, it is plausible that it is related to the their growing region, since it seems that seaweeds from Galician waters accumulate less iodine [15]. It is also noteworthy that *C. tomentosum*, despite showing the highest ash content, did not display significantly high levels of any of the test minerals, except for Ca (5.1 g/kg dw). This could be due to the presence of other minerals or metals not tested in this work. On the other hand, other minerals and metals searched for like Mn, As and Zn were detected at trace amounts, below 0.01 g/kg dw (data not shown). F was also detected in all species, reaching up to 1.1 g/kg dw in *P. palmata*, although *C. tomentosum* had the lowest levels of it (0.1 g/kg dw).

#### *3.3. Fatty Acid Profile*

Regarding fatty acid profile (Figure 3), the proportion of PUFA was notably high, with *P. palmata* displaying the highest relative quantities. *P. palmata* levels of oleic acid were higher than 200 mg/g extract and more than 150 mg/g extract for palmitoleic acid. Moreover, *P. palmata* denoted more than 100 mg/g extract of eicosatetraenoic acid. *U. rigida* however, accounted for the most linoleic acid content (>150 mg/g extract).

**Figure 3.** Fatty acid profile of the studied species. UR, *Ulva rigida*; CT, *Codium tomentosum*; PA, *Palmaria palmata*; PU, *Porphyra purpurea*.

In contrast, *P. purpurea* showed higher proportions of saturated fatty acids, foremost palmitic (>320 mg/g extract) and stearic acids (>70 mg/g extract). Nonetheless, *C. tomentosum* accounted for the highest proportion of saturated fatty acids, like palmitic acid (>400 mg/g extract) and behenic acid (>170 mg/g extract). Altogether, results corroborate that these edible algae are a good source of nutrients and analytical methods are suitable, in accordance with literature [7,16].

#### **4. Conclusions**

Seaweeds have been traditionally consumed for their nutritional value and availability. Moreover, in recent years they have been proposed as an alternative source of proteins and minerals, in contrast to terrestrial plants. It is of interest to accurately determine their nutritional composition to critically assess their value. In this work, standardized and recognized AOAC analytical methods for nutritional composition were employed to determine the major elements of four edible Atlantic seaweed species. Additionally, their mineral and fatty acid profile was investigated. Results unveiled that their protein and mineral content makes them notable sources of these nutrients, especially in red algae species. In these red seaweeds, protein content reached more than 20% for *P. palmata* and even more than 30% for *P. purpurea*. Whereas lipid content was particularly low (possibly due to specific environmental growing conditions), fatty acid analysis denoted high proportions of unsaturated fatty acids, i.e., oleic and linoleic acids in both *U. rigida* and *P. palmata.* Considering the reported results, these seaweed species growing in NW Spain could be a potential food and/or feed ingredient, especially owing to their high contents in minerals. Among the studied species, *P. palmata* stands out due to its PUFA and mineral composition, as well as its mineral content. *P. porphyra* on the other hand, was shown to be especially rich in proteins and P. Taking present results, seaweeds could be proposed as alternative supplementation ingredients for food and feed instead of animal or terrestrial plant sources; since not only they are rich in valuable nutrients, but also currently underexploited for this purpose. Further research should analyze other nutritional aspects from these widespread seaweed species in more depth. This would allow to assess with more accuracy the nutritional value of these Atlantic seaweeds.

**Supplementary Materials:** The poster presentation is available online at https://www.mdpi.com/ article/10.3390/CSAC2021-10681/s1.

**Author Contributions:** Conceptualization, J.E., C.L.-L., A.C.-C., F.C., M.F.-C., P.O., P.G.-P., S.B., F.F.-S., H.C., J.X., J.S.-G. and M.A.P.; methodology, J.E., C.L.-L., A.C.-C. and F.C.; validation, M.F.-C., P.O., H.C. and P.G.-P.; formal analysis, M.F.-C., P.O., H.C. and P.G.-P.; investigation, S.B., F.F.-S., J.X., J.S.-G. and M.A.P.; writing—original draft preparation, J.E., C.L.-L., A.C.-C. and F.C.; writing—review and editing, M.F.-C., P.O., P.G.-P., H.C., J.X., J.S.-G. and M.A.P.; visualization, S.B., F.F.-S., J.X., J.S.-G. and M.A.P.; supervision, J.X., J.S.-G. and M.A.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** The research leading to these results was supported by MICINN supporting the Ramón y Cajal grant for M.A. Prieto (RYC-2017-22891) and the FPU grant for A. Carreira-Casais (FPU2016/06135); by Xunta de Galicia for supporting the program EXCELENCIA-ED431F 2020/12, the post-doctoral grant of M. Fraga-Corral (ED481B-2019/096), the program BENEFICIOS DO CONSUMO DAS ESPECIES TINTORERA-CO-0019-2021 that supports the work of F. Chamorro and the program Grupos de Referencia Competitiva that supports the work of J. Echave (GRUPO AA1-GRC 2018); by the Bio Based Industries Joint Undertaking (JU) under grant agreement No 888003 UP4HEALTH Project (H2020-BBI-JTI-2019) that supports the work of P. Otero, P. Gar-cia-Perez and C. Lourenço-Lopes; and by Ibero-American Program on Science and Technology (CYTED—AQUA-CIBUS, P317RT0003).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** The authors are grateful to Algamar (www.algamar.com, accessed on 1 July 2021) for their collaboration and algae material provision. The JU receives support from the European Union's Horizon 2020 research and innovation program and the Bio Based Industries Consortium. The project SYSTEMIC Knowledge hub on Nutrition and Food Security, has received funding from national research funding parties in Belgium (FWO), France (INRA), Germany (BLE), Italy (MIPAAF), Latvia (IZM), Norway (RCN), Portugal (FCT), and Spain (AEI) in a joint action of JPI HDHL, JPI-OCEANS and FACCE-JPI launched in 2019 under the ERA-NET ERA-HDHL (No 696295).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Proceeding Paper* **Drift Compensation of the Electronic Nose in the Development of Instruments for Out-of-Laboratory Analysis †**

**Anastasiia Shuba 1,\*, Tatiana Kuchmenko <sup>1</sup> and Dariya Menzhulina <sup>2</sup>**


**Abstract:** A technique was developed to evaluate and compensate for the drift of eight mass-sensitive sensors in an open detection cell in order to estimate the influence of external factors (temperature, changes in the chemical composition of the background) on the out-of-laboratory analysis of biosamples. The daily internal standardization of the system is an effective way to compensate for the sensor signal drift when the sorption properties of sensitive coatings change during their long-term, intensive operation. In this study, distilled water was proposed as a standard for water matrixbased biosamples (blood, exhaled breath condensate, urine, etc.). Further, internal standardization was based on daily calculation of the specific sensor signals by dividing the sensor signals for the biosample according to the corresponding averaged values obtained from three to five standard measurements. The stability of the sensor array operation was estimated using the theory of statistical process control (exponentially weighted moving average control charts) based on the specific signal of the sensor array. The control limits for the statistical quantity of the central tendency for each sensor and the whole array, as well as the variations of the sensor signals, were determined. The average times required for signal and run lengths, for the purpose of statistically substantiated monitoring of the electronic nose's stability, were calculated. Based on an analysis of the tendency and variations in sensor signals during 3 months of operation, a technique was formulated to control the stability of the sensor array for the out-of-laboratory analysis of the biosamples. This approach was successfully verified by classifying the results of the analysis of the blood and water samples obtained for this period. The proposed technique can be introduced into the software algorithm of the electronic nose, which will improve decision-making during the long-term monitoring of health conditions in humans and animals.

**Keywords:** piezoelectric sensor array; drift; electronic nose; stability; statistical process control chart; blood

#### **1. Introduction**

When designing a new methodology for the routine analysis of biological samples, significant attention is paid to reducing systematic errors linked to the changing properties of measuring devices. Control over external conditions (temperature and humidity) is determined by the nature of the working elements in the device. Thus, in gas analyzers of the sorption type, the signal is defined by the interaction of phases of modifiers with vapours and gases in the pre-sensory space; in reusable sensors, the stability of sorption surfaces, which defines the sorption–desorption processes, constitutes the main requirement. Since temperature changes the volatility of substances and the sorption properties of substances and sorbents, operating sorption-type sensors at a normal temperature and

**Citation:** Shuba, A.; Kuchmenko, T.; Menzhulina, D. Drift Compensation of the Electronic Nose in the Development of Instruments for Out-of-Laboratory Analysis. *Chem. Proc.* **2022**, *5*, 68. https://doi.org/ 10.3390/CSAC2021-10464

Academic Editor: Ye Zhou

Published: 1 January 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

relative humidity is a critical factor. The development of approaches to ensure the stability of measurements during long-term operation is paramount in portable and mobile analytical systems [1] as this provides time minimization from the moment of sample selection to that of measurement, which is vital for biosample analysis.

It is well known that the drift of sensor signals influences classification solutions and the quantitative determination of components in a mixture [2–4]. The short-term drift of basic lines in sorption-type piezoelectric sensors is caused by malfunctions of electric schemes and an abrupt change in the external conditions of the experiment. As a rule, such mistakes are fixed immediately during the measurement and discarded as outliers. Essentially, a long-term drift relates to a change in the sensitive coatings of sensors during operation. Currently, two ways of decreasing the drift impact of a sensor array on analysis results exist: applying extremely stable sensitive sensor coatings [5–8] and mathematical algorithms of drift compensation [9–13]. These approaches increase accuracy while classifying samples. The mathematical algorithms based on multivariate analysis of the drift correction of sensor signals are difficult to implement in practice for out-oflaboratory analysis, since they require additional computational costs, software changes, or the application of additional programs along with the operator's modelling skills. The most critical component in mathematical processing implementation is valid information selection, obtained over a long period of time, for a single measurement but not for a data set. The goal of the investigation, the nature of the biosample, and the operating conditions are the factors that determine the choice of drift compensation approach.

In this study, we present a technique for drift compensation for the signal of masssensitive piezoelectric quartz sensors during operation in the open detection cell and frontal input of the gaseous phase over biosamples with significant water content. The technique consists of piezoelectric sensor signal correction in a portable e-nose as part of a lengthy routine analysis of water and blood samples, combined with a simple algorithm for assessing system instability, in order to increase the quality of measurement and decrease first- and second-order classification errors. This technique can be used for real sample analysis, including biological samples (blood, secretions, mucus, exhaled breath condensate, perspiration, urine, and saliva), with minor a modification of the device software for out-oflaboratory analysis.

#### **2. Methods and Analysis**

A gaseous phase analysis of distilled water and blood samples was conducted on the gas analyzer Diagnost-Bio-8 (LLC SenSino, Kursk, Russia) using frontal analyte input mode [14]. The sensor array consisted of eight piezoelectrical quartz BAW-type resonators with a 10.0 MHz basic oscillation frequency, with coatings from solid nanostructured sorbents ('living system' set): Sensor 1 and 8 are carboxylated carbon nanotubes of different masses, marked in the tables and text as MCNT1, MCNT2; Sensor 2 and 7 are phases of nitrate of zirconium oxide of different masses (Zr1, Zr2); Sensor 3 is dicyclohexano-18 crown-6 (DCH18C6); Sensor 4 and 5 are biohydroxyapatite phases of different masses (HA1, HA2); and Sensor 6 is polyethylene glycol succinate (PEGsc). This study describes sensor manufacturing [15]. Moreover, the work [16] presents the sorption characteristics of the sensor array from the 'living system' set. The basic lines of the sensors were observed to remain stable (±1 Hz) during an 80 s blank measurement before the sample analysis.

The blood or water sample (volume: 0.5 cm3) was placed on a glass petri plate and then covered by a detection cell of Diagnost-Bio-8. The measurement mode involved combined 80 s sorption and 120 s spontaneous desorption [14].

For the analysis of work stability and drift correction of the sensor array signal, a 3 month period (October–December 2019) of device operations (656 measurements) was chosen, during which 75 water samples and 31 blood samples of somatically healthy people, with indicators of general and biochemical analysis of blood within normal limits, were tested. As part of the training set of the samples, sensor array data for 64 samples (19 blood samples and 45 water samples) were selected for the first 1.5 months of operation (October–November). The sensor array data for 42 samples (12 blood samples and 30 water samples) were selected as the test set of samples for the next period of operation (November–December). During the experiment, external factors showed variation: the room temperature changed from 20 to 25 ◦C, the humidity was 45–55%, and the slight changes in room smell (background) per day corresponded to disinfection and ventilation schedule of the laboratory. Measurements were not performed during disinfection and ventilation and 90 min after these processes. Regarding the original sensor data for each sample, a special software determined maximum sensor responses (Δ*Fmax,i*, Hz), which corresponded to 80 s of measurement (sorption time).

We suggested the correction of the sensor signals in an array based on daily internal standardization. The distilled water was chosen as the standard. We calculated the specific signals *Fi* via the following formula:

$$\overline{F\_i} = \frac{\Delta F\_{\text{max},i}(for\ sample)}{\Delta \overline{F\_{\text{max},i}}(for\ water)} \,\, ^\prime \,\, \tag{1}$$

where Δ*Fmax,i*, Hz are values of the original signal of *i*-th sensor during the biosample analysis and Δ*Fmax*,*iFmax*,*<sup>i</sup>* is the average signal of the *i*-th sensor for three or five analyses of water samples on the same day.

For specific signals calculated for the water samples, we applied a statistical process control method, namely the exponentially weighted moving average (EWMA) control chart, for uni- and multivariate data [17], in order to assess the stability of the sensor arrays. We investigated the statistical quantity T2 *<sup>i</sup>* in the sensor arrays, for each sensor–parameter z*i*, and exponentially weighted the mean square (EWMS) error s*<sup>i</sup>* during 3 months of operation. Before the computation of the statistical criteria, the specific signals obtained were standardized by the average value equal to 1.00 for all the sensor signals. The standard deviation for each sensor was determined according to the data obtained for the first 25 days of operation.

As an alternative method of drift compensation for sensor signals, component correction by principal component analysis (CC-PCA) was employed in order to remove a part of the information. This described the sensor signal drift corresponding to the first principal component from the data matrix [18].

As variables for designing the classification models, we used original maximum sensor responses and specific signals. The results of the water and blood gaseous phase analysis by sensor array were classified into two classes—'water' and 'blood'—using linear discriminant analysis with preliminary processing by principal component analysis (PCA-LDA) with a significance level of 0.05. A CAMOSoftware Unscrambler (v.10.0.0, Oslo, Norway) was employed to assess the effectiveness of the drift compensation techniques.

#### **3. Results and Discussion**

The following are the significant factors determining the drift of sensor signals in the device with an open detection cell: the changes in the sorption properties of sensitive coatings during operation and the external factors of the experiment (temperature, humidity, air composition in the laboratory [(background)). Frequently, these factors act together, which makes it difficult to estimate and predict the contribution of an individual factor to the total signal change. The other internal factors, such as the malfunctions in the oscillating scheme of the device and the change in the features of utilized quartz and electrodes, and the problems linked to signal transmission into software for our devices, lead to no more than 5% variation in sensor signals [19]. The algorithm of multivariate data processing for the drift compensation of the sensor signals was limited by the pilot experiment conditions for the model design. It cannot always contain all the factor combinations influencing the system's signals.

A daily internal standardization may perform a simple accounting of the external factors that influence sensor signals during long-term operation. As the standard, we proposed to use the substance whose content was dominant in the analyzed samples. For biosamples, such as blood, urine, exhaled breath condensate and saliva, distilled water is an appropriate standard. During the 3 months of the experiment, the value of the original signals of the sensors in water vapour varied from 20 to 27%; moreover, the dispersions of the original sensor signals were non-homogeneous, according to Cochran's C-test. The application of the daily standardization reduced the variation to 12%, and the dispersions of all the signals became homogeneous. The average specific signals of all the sensors throughout the total operation period equalled 1.00. The implementation of the specific signals considers the changes in biosamples as well as the sorption properties of sensor coatings, depending on external factors, and thereby reduces the errors of sample classification. However, more accurate classification results could be obtained using information about the stability of sensors.

#### *3.1. Stability Assessment of Piezoelectric Sensor Array Using Specific Signals*

The application of multivariate EWMA control charts by specific signals helps to discover only abrupt changes in external conditions (caused by the measurement of a contaminated standard, which can be classified as operator error). Therefore, detailed information about stability sensor arrays describes EWMA control charts for individual signals of the sensor array (Figure 1).

**Figure 1.** EWMA control chart for specific signals of four sensors from the array. The solid red lines indicate the upper and lower control limits for the third sensors in the array.

According to multiple excesses of control limits, the high instability of Sensor 6 with the PEGsc film was established because of the excellent water impact on the properties of polymer films. The control limits of the sensor with the DCH18C6 film (Sensor 3) featured the most significant values (Table 1). This is explained by the nature of the modifier, which consisted of solvated crown-ether macromolecules. This type of structure is less stable and varies more depending on external conditions.


**Table 1.** Maximum values of upper (+) and lower (–) control limits for parameter z*<sup>i</sup>* for each sensor.

The parameters z*<sup>i</sup>* for the sensors with hypersensitive nanostructured films of carboxylated nanotubes (Sensor 8) and hydroxyapatite (Sensor 4) were almost always within the control limits. The exceptions were the day of synchronous change of the z*<sup>i</sup>* parameters of all the sensors and the following three days (41–51 measurements); these exceptions were related to the changes in the coating properties during the sorption of the contaminated standard vapours and their subsequent reactivation by purging with laboratory air and repeated measurements of pure standard samples. The data from the biosample analysis obtained on these days may have been unreliable and may have contributed to error in the classification model.

To decide on the exclusion of the sensor signals from the classification data and to assess the sensor regeneration level after poisoning, an analysis of the changes to the signal variability was required. Here, the evaluation of the sample generalized variance for sensor array was not informative, since its values throughout the total studied period were either negative or extremely low (less than 10−60). Therefore, we considered the EWMS errors s*<sup>i</sup>* of specific sensor signals (Figure 2). On the day of the abrupt change in external conditions (measurements of the contaminated sample of the standard; measurement No. 41–43), variation for all the sensors and z*<sup>i</sup>* exceeded the upper limit control (UCL) of 1.45. Therefore, in the case of the synchronous increase in control limits by parameters z*<sup>i</sup>* and s*<sup>i</sup>* for all the sensors, it was necessary to terminate measurements, verify compliance with the conditions for the measurement procedure and reactivate the sensor array's coatings. When calculating the EWMS for each subsequent day after conducting the reactivation of coatings, EWMS s0 = 1 was accepted as the initial point for all the sensors. In the following eight days (measurement No. 44–65), the EWMS of all the sensors, except for Sensors 1 and 8, which featured carbon nanotube coatings, exceeded the UCL. This probably indicated the sorption quality changes in the hydrophilic modifiers following the sorption of contaminated standard vapor and their gradual recovery when measuring the pure standard or slight inertia of the surface reactivation process of nanostructured mineral coatings, such as hydroxyapatite.

**Figure 2.** Control chart for EWMS errors of four sensors.

The average run length (ARL) and time to signal (ATS) were calculated according to EWMA theory. The ARL was equal to three measurements to detect three σ shifts in signals and the ATS was 2.5 h. After a long break or recovering of the sorption properties of coatings, the ARL was five measurements, whereas ATS was 1.5 h.

General recommendations for sensor data analysis of biosamples, accounting for their stability, were formulated as a scheme that could be implemented in the device's software. The purpose was to monitor the stability sensor array with minimal control by the operator, which is essential for out-of-laboratory analysis. The proposed approach was implemented for different forms biosample analysis, such as blood analysis.

#### *3.2. Blood and Water Sample Classification*

To ensure the orthogonality of the data for the discriminant analysis, we performed the principal component analysis of original and specific sensor signals. For further analysis, we used the first four principal components with 98% explained variance, which allowed us to consider all the information from the sensor array during the sample classification. In the first stage, we estimated the possibility of water and blood sample classification using original signals of sensors. The accuracy of this classification model for the training set of samples was 90.4% (Table 2).

The misclassification of blood samples as 'water' class was observed. This signifies that during the analysis of the gaseous phase over the blood samples, the sample matrix influences sensors and that over a period of time, differences within the gaseous composition in the samples become less noticeable for the sensors. The highest value for the accuracy and correctness of the blood sample classification belonged to the model constructed by specific sensor signals, which accounted for sensor stability information (Table 2). Consequently, the drift compensation of the sensor signals using specific signals with combination sensor stability monitoring by EWMA control charts could be applied, along with the routine sensor array operation in the laboratory and out-of-laboratory analyses. Furthermore, the correct classification of the samples considering the impact of the dominant component allows us to suggest that this approach can provide a more accurate classification of slight differences in the composition of gaseous phases of blood with pathologies.

**Table 2.** Accuracy and correctness of sample division into two classes by PCA-LDA models for original and specific sensor signals.


1—for the model design, we used the whole data array; 2—the samples and sensor data were excluded according to the stability evaluation of sensor array operation.

#### **4. Conclusions**

We proposed a fast and efficient compensation method for sensor signal drift based on daily standardization (dividing sensor signals for biosamples into corresponding average signals for standard samples (measured thrice during the day)). Distilled water was suggested as a standard for blood samples and other biofluids (urine, perspiration, exhaled breath condensate, and saliva). EWMA control charts were applied for sensor array stability monitoring as an additional module in the device software for routine analysis. The coating of carboxylated carbon nanotubes of small mass (Sensor 8) was the most stable within the studied sensor array when measuring the water and blood samples based on the statistical quantities of the EWMA charts. The effectiveness of applying for drift compensation by daily standardization, with a combination of sensor stability monitoring, was proven by significantly improving the blood samples' classification accuracy. Similarly, it is possible to compensate for the drift of sensor signals when analysing other biosamples using appropriate standards. The additional parameters of sorption or features of sensor output curves after standardization could be used to improve the correctness of the classification during the long-term operation of the sensor array.

**Author Contributions:** Conceptualization, T.K.; methodology, T.K. and A.S.; validation, D.M. and A.S.; investigation, D.M.; data curation, A.S.; writing—original draft preparation, D.M.; writing review and editing, A.S.; supervision, T.K.; project administration, T.K.; funding acquisition, T.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was partly funded by the Russian Science Foundation, grant number 18-76-10015.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by Ethics Committee of Voronezh State University of Engineering Technologies (protocol code No. 2 from 21 February 2021).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study. Written informed consent has been obtained from the patients to publish this paper.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


### *Proceeding Paper* **The Performance of Organophosphate Pesticides Determination Using Biosensor Based on Small Device Potentiometer as a Transducer †**

**Mashuni Mashuni 1,\*, Halimahtussaddiyah Ritonga 1, Muhammad Jahiding 2, La Ode Ahmad Nur Ramadhan 1, Desy Kurniawati <sup>1</sup> and Fitri Handayani Hamid <sup>1</sup>**


**Abstract:** The need to control pesticide residues in foodstuffs in a fast and straightforward analysis for the field scale is required. Therefore, this research develops a transducer-based biosensor with a small device potentiometer (SDP) to produce a fast and accurate pesticide detection tool. The biosensor based on Au electrodes by immobilizing the acetylcholinesterase (AChE) enzyme coated membrane cellulose acetate (CA) 15% (*w*/*v*) cross-linked glutaraldehyde (GA) 25% (*v*/*v*) and SDP as a transducer that produces a potential value. The biosensor testing results on the organophosphate pesticide class, namely diazinon and profenofos, showed the sensitivity of 21.204 and 20.035 mV decade−1, Limit of Detection (LoD) 10−<sup>7</sup> mg L−1, selectivity coefficient <sup>−</sup>1 < *Ki***,***<sup>j</sup>* < 1 and accuracy of 99.497 and 94.765%, respectively. The results showed that the biosensor connected to an SDP transducer had an excellent performance in determining the presence of organophosphate pesticides.

**Keywords:** small device potentiometer; biosensor; acetylcholinesterase; organophosphate; pesticide; diazinon and profenofos

#### **1. Introduction**

Organophosphate pesticides are a group of pesticides that contain a phosphate group. The organophosphate pesticides in agricultural and plantation processing systems are widely used to tackle pests and diseases that attack plants, the leading cause of declining crop yields. However, the use of pesticides harms the health of living things and the balance of the environment [1,2]. Analysis of pesticide residues from crop yields is necessary to ensure food safety. One alternative method of pesticide analysis is a biosensor [3,4].

In recent decades, biosensors have become a popular research area capable of identifying pesticide residues and other chemicals. Biosensors are "self-standing devices" that record physical, chemical or biological changes, convert them into measurable signals from the sample and monitor the analyte of interest [5,6]. The sensor contains a recognition element that allows a selective response to a specific analyte or group of analytes, minimizing interference from other sample components. Another significant sensor component is a transducer or detection device that produces a signal [7].

Electrochemical biosensors are a subclass of chemical sensors that combine sensitivity such as low detection limit—electrochemical transducers with the high specificity of biological recognition processes. These devices contain biological recognition elements (enzymes, proteins, antibodies, nucleic acids, cells, tissues or receptors) that selectively react with the

**Citation:** Mashuni, M.; Ritonga, H.; Jahiding, M.; Ramadhan, L.O.A.N.; Kurniawati, D.; Hamid, F.H. The Performance of Organophosphate Pesticides Determination Using Biosensor Based on Small Device Potentiometer as a Transducer. *Chem. Proc.* **2021**, *5*, 69. https://doi.org/ 10.3390/CSAC2021-10604

Academic Editor: Xin Wu

Published: 5 July 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

target analyte and generate an electrical signal related to the measured analyte concentration [5,8]. Enzyme-based electrochemical biosensors have advantages over conventional methods due to their excellent sensitivity, selectivity, mini size and fast response [5,9,10].

Maintaining the catalytic activity on the efficient immobilization of the acetylcholinesterase (AChE) enzyme is an essential consideration when developing electrochemical biosensors that can be used for practical applications. The highlight of the electrochemical biosensor is its unique ability to generate digital signals that can measure by converting the catalytic signal with the help of microfabricated electronics [3,11,12]. The electrochemical method using measurement tools is based on potentiometric [12], amperometric [13] or conductometric [14] biosensor. Potentiometric biosensors are suitable for measuring the response value of pesticide detection measurements [15].

Potentiometric biosensors are efficient for in-field analysis because they are more straightforward and ideal for real-time analysis [16]. The potentiometric detection system developed by Timur, S. and Telefoncu, A., 2004 [17], has the underlying principle of inhibition of AChE activity due to its properties in identifying organophosphate compounds. The enzyme was immobilized on the surface of the electrode with the help of a chitosan membrane [18]. Without a pre-concentration step, in both aqueous and organic media, detection of organophosphates without the requirement of trained personnel proved advantageous for the proposed portable biosensor. Pesticides were effectively detected in the range of 0.1–100 mM for parathion-methyl and methamidophos and 0.6–600 mM for Malathion [17]. However, in the presence of higher pesticide concentrations, only partial regeneration of the enzymatic activity was regenerated [15].

The combination of potentiometric-based AChE enzyme biosensors as transducers with analytical techniques has been widely reported in the literature as a suitable method. In this work, we report the development of a small device potentiometric (SDP) based biosensor as a transducer for the determination of pesticide organophosphates, based on the AChE enzyme immobilized on cellulose acetate (CA) and glutaraldehyde (GA) membrane-coated Au electrodes.

#### **2. Materials and Methods**

#### *2.1. Meterials*

Acetylcholinesterase (AChE, from electrophorus, Sigma-Aldrich, St. Louis, MO, USA, 1.17 mg with activity 425.94 units per mg (EC. 3.1.1.7)) in 9 mL PBS pH 8 and 1 mL KCl 10−<sup>1</sup> M, cellulose acetate (CA, from Sigma-Aldrich, St. Louis, MO, USA, 15% *v*/*v* in acetone), glutaraldehyde (GA, from Sigma-Aldrich, St. Louis, MO, USA, 25% in H2O), potassium chloride (KCl, from Merck, Darmstadt, Germany, 10−<sup>1</sup> M in H2O) and acetone (C3H6O, from Sigma-Aldrich, St. Louis, MO, USA, ≥99.5%). Phosphate buffer solutions (PBS) with values of pH 8.0 were prepared by mixing standard stock solutions of 0.2 M Na2HPO4 (99% purity) and 0.2 M NaH2PO4 (99% purity). Standard solution of acetylthiocholine chloride (ATCl, A5626 from Sigma-Aldrich, St. Louis, MO, USA) substrate with concentrations of 10−1, 10−<sup>2</sup> and 10−<sup>3</sup> M in PBS solution. The pesticides used in the OP group are diazinon and profenofos was purchased from Merck, Darmstadt, Germany, as inhibitors, each made in concentrations of 10<sup>−</sup>9, 10−8, 10−7, 10−6, 10−5, 10−4, 10−3, 10−<sup>2</sup> and 10−<sup>1</sup> mg L−<sup>1</sup> in 5 mL ethanol (C5H6O, from Sigma-Aldrich, St. Louis, MO, USA, 99.8%) and H2O.

#### *2.2. Apparatus*

The potentiometer is used as an experimental tool for measuring the potential value of analyte detection [19]. The working electrodes used are Au and a platinum (Pt) as a cathode of the electrolysis process and an Ag/AgCl as a reference electrode.

#### *2.3. Electrolysis of Ag/AgCl*

Reference electrodes of Ag/AgCl were carried out by electrolyzing Ag wire (anode) and Pt wire (cathode) in 0.1 M KCl solution for ±20 min. The length of time electrolysis

will affect the thickness of AgCl on Ag wire, where the more extended the electrolysis process, the thicker it will be to a certain extent, and vice versa. Next, the Ag/AgCl wire that has been formed is then dried in the open air. Finally, Ag/AgCl wire that has been dried at room temperature is inserted into the electrode body as a comparison electrode for Ag/AgCl [20].

#### *2.4. Preparation of Au Electrode Biosensor*

The Au electrode tip was immersed in a 15% CA membrane solution. The CA membrane formed was rinsed with distilled water and then dipped in 25% GA solution for 6 h. Furthermore, the electrode was rinsed with distilled water and PB solutions pH 8, then an electrode membrane (Em) was formed. Then, Em was immersed in the AChE enzyme for 2 × 24 h at 4 ◦C. Before measuring the response to the biosensor, the components in the measurement, such as standard electrodes, coated wire type working electrodes, ATCl substrate and inhibitor solution, need to be left at room temperature for about 2 h until components are stable and produce a good response [20].

#### *2.5. Measurement of the Potential Value Biosensor*

Measurement of the potential value of the enzymatic biosensor electrodes with the pesticide inhibitors diazinon and profenofos in concentrations of 10−8, 10−7, 10−6, 10−5, 10<sup>−</sup>4, 10−3, 10−<sup>2</sup> and 10−<sup>1</sup> mg L−<sup>1</sup> using a potentiometer. Em was immersed in PBS pH 8.0 for 10 min and then used ME to measure the potential value of 10−<sup>3</sup> M ATCl substrate to obtain a constant value. The Em was removed and rinsed with distilled water, and then Em was immersed in a pesticide solution for 30 min, then removed and rinsed with PB solution pH 8.0 before being dipped again into the ATCl substrate solution. Furthermore, it made observations to obtain a constant potential value.

#### *2.6. The performance Test of Biosensor*

#### 2.6.1. Sensitivity

The sensitivity value (Nernst factor) is determined using a graph of the relationship between the potential value and –log of inhibitor concentration. Then, we can see the linear equation from the chart to obtain the sensitivity range of the diazinon and profenofos pesticide electrode.

#### 2.6.2. Limit of Detection (LoD)

LoD is the lowest limit of analyte concentration that can be measured by the instrument, which is statistically different from the blank. Determination of LoD was carried out by analyzing the potential response of a series of standard solutions of various pesticide concentrations of 10−8, 10−7, 10−6, 10−5, 10−4, 10−3, 10−<sup>2</sup> and 10−<sup>1</sup> mg L−1. The analysis results obtained a linear equation of the calibration curve, y = ax + b, then the measurement of the potential value of the blank. The equation for the value of y at the detection limit is based on equation of Christian et al. (2014) [21]. They suggest calculation for the LoD = 3 × (SD/S) based on the response's standard deviation, SD, and the slope or sensitivity, S, of the calibration curve at levels approaching the limit.

#### 2.6.3. Selectivity

Selectivity is expressed as the degree of bias of the primary analyte analysis data with interference compared to the analyte analysis data without interference [21]. The sample analysis used a concentration of 10−5, 10−<sup>4</sup> and 10−<sup>3</sup> mg L−1. Measuring the potential value using a potentiometer and calculating the selectivity coefficient (*Ki,j*). The value of the electrode selectivity calculated based on the Nicolsky–Eisenman equation [22]. The potential of an ion-selective electrode in the presence of an interfering ion follows an equation:

$$E\_{ISE} = k + \frac{\mathcal{S}}{z} \log \left( a\_i + \sum\_{j \neq i} \mathcal{K}\_{i,j} \ a\_j^{z\_i z\_j} \right),$$

where *S* is the slope (theoretically 2.303RT/F) and *k* is the ion charge including all contributions independent of activities *a* and *z* is the ion charge. The subscript *i* stands for the primary pesticides and subscript *j* stands for the interfering pesticides.

#### 2.6.4. Accuracy

Determination of the accuracy value is obtained by calculating the % recovery, where the sample used is mustard greens with the addition of inhibitor concentrations of 10−4, 10−<sup>3</sup> and 10−<sup>2</sup> mg L−1. Measurement of potential value using a potentiometer and calculated % recovery based on the equation of Christian et al. (2014) [21].

#### **3. Results and Discussions**

As shown in Figure 1, an analysis of the performance of the biosensor on the pesticide diazinon and profenofos using a small device potentiometric (SDP)-based biosensor as a transducer was carried out. SDP-based biosensor performance tests, including sensitivity, LoD, selectivity and accuracy, are essential parameters in biosensors.

**Figure 1.** Graph of the relationship of −log [inhibitors] with the potential value of biosensor-based SDP.

The sensitivity, or Nernst factor, is one of the general parameters of biosensor performance testing, using SDP as a transducer indicated by the slope resulting from the calibration of the electrode potential response to the analyte's activity. Analysis using potentiometric is based on the potential change in each variation of ion concentration [16]. The feasibility of a tool used in detecting an analyte is seen from how much sensitivity is in the measurement process. So, in this study, the Nernst factor value was determined to see how well the sensitivity of the tool and the measurement range of an electrode was suitable for use as a pesticide detection tool. The results of the measurement of the potential value of the biosensor performance are presented in Table 1.

Figure 1 shows the sensitivity of the performance of SDP-based biosensors to the detection of pesticide diazinon and profenofos of 21.204 and 20.035 mV decade−1, respectively. The sensitivity is the slope value of the linear regression equation from the graph of the relationship −log [inhibitor], with the potential value measured using a potentiometer. The value of the Nernst factor is more ideal if it is close to the value of 29.6 mV decade−<sup>1</sup> [23].

The characteristics of a biosensor are also determined by its ability to detect the concentration of an analyte. The smaller the concentration that can be seen, the better the biosensor features. Limit of Detection (LoD) is the minor analyte concentration that gives a sufficiently large signal and can be distinguished from the signal obtained from the blank with a 99% confidence level [15]. The optimum performance of a biosensor can be determined by its ability to detect the concentration of an analyte. The LoD was determined using the standard deviation of the intercept and the slope of the calibration line. Table 1 shows the LoD value of the SDP-based biosensor as a transducer, which is 10−<sup>7</sup> mg L−<sup>1</sup> instead of 10−<sup>8</sup> mg L<sup>−</sup>1, because the potential at that concentration is close to the potential blank value. So, the smaller the concentration that can be detected, the better the biosensor characteristics [24].


**Table 1.** Measurement of potential value, sensitivity value and LoD of biosensor.

The selectivity is carried out to determine the method's ability to measure the presence of pesticides carefully and thoroughly in interfering components. The ideal biosensor is expected to only respond to the primary analyte to be detected with selectivity coefficient −1 < *Ki***,***<sup>j</sup>* < 1. If the electrode is highly selective towards *i* rather than ion *j*, *Ki***,***<sup>j</sup>* < 1. Conversely, if the electrode is highly selective towards *j*, ion *i*, *Ki***,***<sup>j</sup>* > 1. Variations in the value of *Ki***,***<sup>j</sup>* depend on the electrode's response and the component environment in solution. The selectivity coefficient value obtained is smaller than 1 (Table 2). Based on the overall value received, the range of low concentrations of interfering components is still within tolerance. The average selectivity coefficient value received still meets the specified selectivity value standard, greater than −1 and more minor than +1, except for the primary analyte analysis data, diazinon 10−<sup>5</sup> mg L−<sup>1</sup> and profenofos interfering 10−<sup>5</sup> mg L<sup>−</sup>1, then the average selective electrode for pesticide detection compared to interfering compounds [25,26].

Accuracy is a measure that shows the degree of closeness of the analysis results to the actual analyte content. Accuracy is expressed as the % recovery of the added analyte. In general, the acceptance criteria for accuracy (% recovery) are 80–110% [27]. Accuracy analysis is carried out using the recovery method by sample spiking or standard addition to the sample to be analyzed. The method is carried out by adding a certain amount of analyte with a certain concentration to the analyzed sample. The data in Table 3 show that the average % recovery of the SDP-based biosensor as a transducer has an accuracy rate of 99.497 and 94.765% for diazinon and profenofos pesticide detection, respectively. The percent recovery value obtained is in accordance with the required standard. See Supplementary Materials for details.


**Table 2.** Selectivity of biosensor-based SDP.

*ai* is the concentration of primary pesticides, *aj* is the concentration of interfering pesticides, *Ki*,*<sup>j</sup>* is the selectivity coefficient, (1) for diazinon pesticide and (2) for profenofos pesticide.

**Table 3.** Accuracy of biosensor-based SDP.


C'A is the concentration of pesticides added, CA is the concentration of the sample, CF is the total concentration of the sample obtained from the measurement.

#### **4. Conclusions**

Based on the results and data obtained from the study of SDP-based biosensors as transducers in the detection of organophosphate pesticides, the sensitivity was 21.204 and 20.035 mV decade−1, LoD 10−<sup>7</sup> mg L−1, selectivity coefficient −1 < *Ki***,***<sup>j</sup>* < 1 and accuracy of 99.497 and 94.765%. Thus, potentiometric biosensors with CA and GA membranes immobilized by AChE enzymes have good sensitivity, selectivity and accuracy in detecting the presence of organophosphate pesticides in a sample and LoD from tiny biosensors are effective for detecting at low scale and concentration.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10 .3390/CSAC2021-10604/s1, presentation materials at the 1st International Electronic Conference on Chemical Sensors and Analytical Chemistry, 1–15 July 2021.

**Author Contributions:** Data curation, M.M., L.O.A.N.R. and D.K.; writing—original draft preparation, M.M. and F.H.H.; writing—review and editing, M.M., H.R. and F.H.H.; supervision, M.M., H.R. and M.J. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received funding from the Ministry of Education, Culture, Research, and Technology of the Republic of Indonesia.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data available on request due to restrictions, e.g., privacy or ethical.

**Acknowledgments:** The authors thank all the team for their contribution to this research, University of Halu Oleo (UHO) and the Ministry of Education, Culture, Research, and Technology of the Republic of Indonesia.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Proceeding Paper* **Optimization of Bioactive Compounds with Antioxidant Activity of** *Himanthalia elongata* **by Microwave-Assisted Extraction Using Response Surface Methodology †**

**Franklin Chamorro <sup>1</sup> , Lucia Cassani 1,2 , Catarina Lourenço-Lopes <sup>1</sup> , Anxo Carreira-Casais <sup>1</sup> , Maria Carpena <sup>1</sup> , Javier Echave <sup>1</sup> , Sergio Baamonde 3, Fermin Fernández-Saa 3, Paz Otero <sup>1</sup> , Pacual Garcia-Perez 1,4 , Jesus Simal-Gandara 1,\* and Miguel Angel Prieto 1,4,5,\***


**Abstract:** *Himanthalia elongata* is a brown alga used in applications in the food, pharmaceutical and nutraceutical industries due to its biological properties, such as antioxidant, anti-inflammatory, and antimicrobial, among others. These effects are attributed to the high content of nutrients and secondary metabolites, especially phenolic compounds. The objective of this study is to optimize the microwave-assisted extraction method to recover phenolic compounds and flavonoids, considering three extraction parameters: the concentration of ethanol in water, the extraction time and pressure. The total phenolic content and the total flavonoid content were evaluated, and two biological tests were performed to assess the antioxidant properties.

**Keywords:** macroalgae; microwave-assisted extraction; *Himanthalia elongata*; bioactive compounds; antioxidant

#### **1. Introduction**

Traditionally, algae have been used as food and for medicinal purposes, mainly in eastern countries. However, its popularity is increasing in western countries, due to the search for healthier and more natural products by consumers, including food, cosmetics, pharmaceutical products, etc. [1,2]. Numerous studies indicate the good nutritional value of algae: they provide proteins and essential amino acids, and they are rich in non-digestible carbohydrates and polyunsaturated fatty acids, vitamins, and minerals. Furthermore, they are a source of compounds with various biological activities (e.g., antioxidants, antivirals, antimicrobials, antifungals, etc.) [3–5], which has attracted the attention of researchers hoping to study them and develop new industrial applications [6–8]. The antioxidant activity of some species of algae has been attributed to the presence of phenolic compounds such as polyphenols, hydroquinones and flavonoids. *Himanthalia elongata* is a brown alga of the order Fucales, found mainly in the N-W Atlantic Ocean and the North Sea.

**Citation:** Chamorro, F.; Cassani, L.; Lourenço-Lopes, C.; Carreira-Casais, A.; Carpena, M.; Echave, J.; Baamonde, S.; Fernández-Saa, F.; Otero, P.; Garcia-Perez, P.; et al. Optimization of Bioactive Compounds with Antioxidant Activity of *Himanthalia elongata* by Microwave-Assisted Extraction Using Response Surface Methodology. *Chem. Proc.* **2021**, *5*, 70. https://doi.org/10.3390/ CSAC2021-10478

Academic Editor: Ye Zhou

Published: 30 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Its antioxidant properties have been described previously [4], and it is reported that the amount of polyphenolic content is higher than in other algae [9].

Bioactive compounds from algae were commonly extracted using organic solvents (methanol, ethanol, acetone) with application temperatures between 45 and 60 ◦C, for hours or days, which implies high energy and environmental costs [10]. In contrast, non-conventional or green extraction techniques, such as ultrasound-assisted extraction (EAU), high-pressure-assisted extraction (HPAE), microwave-assisted extraction (MAE), enzyme-assisted extraction (EAE), supercritical fluid extraction (SFE), pulsed electrified field extraction (PEF), pressurized-liquid-assisted extraction (PLE) and surfactant-assisted extraction (SAE), have proven to be a valid alternative in the recovery of bioactive compounds from algae [11–13]. Among them, microwave-assisted extraction (MAE) is an efficient and environmentally friendly technique, which reduces the extraction time and the amount of organic solvents and, in the best of cases, uses less polluting solvents such as water [10,13]. Different variables, such as the type of solvent, time and pressure, influence the recovery efficiency of bioactive compounds. Optimal extraction parameters can be estimated with statistical optimization methods. In this sense, the response surface methodology (RSM) uses quantitative data from an experimental design to solve the multivariate equation and maximize the results of the selected response variables. The objective of this study is to establish the most favorable conditions for MAE, in terms of the type of solvent, time and pressure required to produce extracts of *H. elongata* rich in bioactive compounds that present antioxidant activity.

#### **2. Material and Methods**

#### *2.1. Sample Preparation*

*H. elongata* samples were provided by the company Algas Atlánticas Algamar S.L located in Pontevedra, Spain. The algae were collected from the coasts of the province of Pontevedra, and they were washed with distilled water, frozen at −80 ◦C and later lyophilized. Next, the samples were crushed and ground to obtain a homogeneous matrix, which was stored at −20 ◦C until use.

#### *2.2. Microwave-Assisted Extraction (MAE)*

The process for obtaining bioactive compounds was carried out by MAE, using the multiwave-3000 equipment (Anton-Par). The extraction was carried out using 0.6 g of the lyophilized alga and 20 mL of solvent (solute/solvent ratio of 30 g/L). The variables studied were the ethanol concentration (*%Et*), pressure (*P*) and time (*t*), as critical extraction parameters. Specifically, the *%Et* varied between 0 and 100% *v*/*v*, the *P* from 2 to 20 bar and *t* from 3 to 25 min. The power and frequency of the microwave were fixed for this batch of experiments and set at the maximum value of 1400 W for power and 2.45 GHz for frequency. Once the extraction was completed, the samples were placed in an ice bath for 5 min in order to rapidly lower the temperature and avoid degradation of the thermolabile compounds. Finally, the samples were centrifuged at 9000 rpm for 15 min and filtered to separate the supernatant from algae debris. These extracts were stored in a freezer at −80 ◦C.

In order to study the influence of MAE conditions (*%Et*, *P* and *t*), the RMS was applied using circumscribed central composite design (CCCD), which allows one to identify the operating conditions for maximizing five response variables: extraction yield (EY), total phenolic content (TPC), total flavonoid content (TFC) and antioxidant activity of *H. elongata*. The interaction between the different variables generates a total of 28 experiments. The least squares regression method was used to fit the data obtained in the 28 experiments to a quadratic model shown in the following equation:

$$Y = b\_0 + \sum\_{l=1}^{n} b\_l X\_l + \sum\_{\substack{l=1 \\ l \neq 1 \\ j > i}}^{n-1} \sum\_{j=2}^{n} b\_{lj} X\_l X\_j + \sum\_{l=1}^{n} b\_{li} X\_l^2 \tag{1}$$

where *Y* is the predicted responses (*Y*1: EY, *Y*2: TPC, *Y*3: TFC, *Y*4: DPPH assay, *Y*5: ABTS assay), *b*<sup>0</sup> is the constant of the model, *bi* is the linear coefficient, *bii* is the coefficient quadratic, *bij* is the coefficient of the interaction and *Xi* is the dimensionless coded value of the independent variables (*X*1: *%Et*, *X*2: *P* and *X*3: *t*).

#### *2.3. Determination of Bioactive Compounds and Antioxidant Capacity*

The *EY* was evaluated based on the dry weight (dw) obtained according to Equation (2).

$$EY\ (\%) = (P\_2 - P\_1) / P\_0 \times 100\tag{2}$$

where *P*<sup>0</sup> is the mass of lyophilized algae prior to extraction (mg), *P*<sup>1</sup> is the mass of the empty crucible (mg), *P*<sup>2</sup> is the mass of the dry extract in the crucible (mg).

The TPC was determined using the Folin-Ciocalteu reagent, while the TFC was evaluated according to the methodology proposed by Cassani et al. [14]. The results were expressed as mg of phloroglucinol equivalents (PGE)/g of dw and mg of quercetin equivalents (QE)/g of dw, respectively. Regarding the antioxidant capacity, it was determined using two assays: the diphenyl-2-picryl-hydrazyl radical (DPPH) and 2,2 -azino-bis(3 ethylbenzothiazoline-6-sulfonic acid) (ABTS) scavenging assays. The results of both assays are expressed in mg of scavenged compound/mL of extract.

#### **3. Results and Discussions**

The experimental results of the RSM of CCCD for the optimization of *H. elongata*' MAE for the five considered response variables are presented in Table 1.


**Table 1.** Experimental parameters of the optimization process.

Abbreviations: extraction yield (*EY*), total phenolic content (TPC), total flavonoid content (TFC), antioxidant assays (DDPH and ABTS).

As can be observed, the five response variables were favored by different extraction conditions. Regarding *EY*, the most favorable conditions were 7.5 min, 16.4 bar and 20% *Et*. The TPC was favored by an extraction time of 14 min, 11 bars and a 50% *Et*. In contrast, TFC was favored under the same conditions of time and pressure, but differed with respect to TPC in the solvent, with 0% Et achieving the best results. Similar results have been reported previously [15]. On the other hand, higher TPC and TFC usually corresponded with higher antioxidant activity in ABTS and DPPH assays. In general terms, the time and pressure parameters with intermediate values favored the EY and the obtaining of TPC and TFC. On the other hand, the parameter with the greatest influence was the *%Et*, showing differences in obtaining bioactive compounds. This can be explained by considering the polarity of the solvent and the compounds [16].

In order to obtain a *H. elongata* extract rich in phenolic compounds and flavonoids, with the maximum antioxidant capacity, all the response variables were simultaneously optimized by means of RSM. The operational conditions that simultaneously optimize all the considered response variables are presented in Table 2. These optimal extraction conditions give rise to an *EY* of 502.28 ± 25.11 mg/g of dw, a TPC of 37.43 ± 3.74 mg PGE/g dw and a TFC of 9.93 ± 0.99 mg QE/g dw. Regarding the antioxidant assays, the radical elimination activity of DPPH and ABTS was 16.37 ± 0.82 and 65.77 ± 1.97 mg/mL, respectively (Table 2).

**Table 2.** Effect of *H. elongata* extract by MAE under optimal conditions on antioxidant activity.


The optimized operating conditions are consistent with the study presented by Magnusson et al. [16], who obtained the best TPC using water as solvent and an extraction time between 3 and 15 min, but very high temperatures were required (160 ◦C). In this sense, Zhang et al. [17] stated, using terrestrial plants, that water is a solvent with good solubility and has an excellent ability to absorb microwave energy and lead to efficient heating of the sample. Regarding TPC, the results of previous studies are variable. For example, Jiménez-Escrig et al. [9] reported a similar TPC around 30 mg PGE/g dw; however, when using aqueous methanol (50%) and extraction times longer than 2 h, TPC was around 10 mg PGE/g dw when using water but also with longer extraction times (1 h). Fernández et al. [18] reported values of 18 mg gallic acid equivalents/g dw, but again they required the use of organic solvents and extraction times longer than one hour and with different steps. Therefore, our data present a rapid, simple and green method to effectively extract a different kind of biomolecule from *H. elongata*. Nevertheless, further microwave parameters, such as temperature, power and frequency need to be further analyzed to obtain the most efficient extraction method. Furthermore, it is noticeable that the differences observed between studies could be due to the great variability of the content and phytochemical profile of algae, which can be larger and affected by different climatic and intrinsic factors, such as season, age, geographical location and environmental conditions [19].

#### **4. Conclusions**

*H. elongata* is an alga species with reported antioxidant activity, which has been attributed to the presence of phenolic compounds and flavonoids. In this study, MAE resulted in a suitable technique to extract those compounds and obtain extracts with antioxidant activity. Furthermore, the RSM was a suitable statistical method to determine the optimal conditions that maximize the content of polyphenols and total flavonoids, the antioxidant capacity and the extraction performance using microwaves. According to the optimization results, the best operational conditions that allowed us to produce extracts rich in bioactive compounds and displayed significant antioxidant effects on DPPH and ABTS assays were 0% Et, 20.00 bar and an extraction time of 16 min. Considering the growing interest in algae compounds, this extract could be used in the development of functional food, cosmetic and pharmaceutical applications.

**Author Contributions:** Conceptualization, F.C., L.C., S.B. and F.F.-S.; methodology, F.C., L.C., A.C.-C. and C.L.-L.; software, M.C., P.O. and P.G.-P.; validation, M.C., P.O. and P.G.-P.; formal analysis, F.C., M.C., P.O. and P.G.-P.; investigation, F.C., L.C., A.C.-C. and C.L.-L.; writing—original draft preparation, F.C. and J.E.; writing—review and editing, F.C. and J.E.; visualization, J.S.-G. and M.A.P.; supervision, J.S.-G. and M.A.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** The JU receives support from the European Union's Horizon 2020 research and innovation program and the Bio Based Industries Consortium. The project SYSTEMIC Knowledge hub on Nutrition and Food Security has received funding from national research funding parties in Belgium (FWO), France (INRA), Germany (BLE), Italy (MIPAAF), Latvia (IZM), Norway (RCN), Portugal (FCT), and Spain (AEI) in a joint action of JPI HDHL, JPI-OCEANS and FACCE-JPI launched in 2019 under the ERA-NET ERA-HDHL (n◦ 696295). The authors acknowledge the "Margarita Salas" grant awarded to Pascual Garcia-Perez, supported through the European Union by the "NextGenerationEU" program.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** The research leading to these results was supported by MICINN supporting the Ramón y Cajal grant for M.A. Prieto (RYC-2017-22891) and the FPU grant for A. Carreira-Casais (FPU2016/06135); by Xunta de Galicia for supporting the program EXCELENCIA-(ED481B-2021/152), the post-doctoral grant of L. Cassani, the program BENEFICIOS DO CONSUMO DAS ESPECIES TINTORE-RA-(CO-0019-2021) that supports the work of F. Chamorro, the EcoChestnut Project (Erasmus+ KA202) that supports the work of J. Echave; by University of Vigo for supporting the predoctoral grant of M. Carpena (Uvigo-00VI 131H 6410211) and the European Union through the "NextGenerationEU" program supporting the "Margarita Salas" grant awarded to P. Garcia-Perez. The authors are grateful to Ibero-American Program on Science and Technology (CYTED—AQUA-CIBUS, P317RT0003), to the Bio Based Industries Joint Undertaking (JU) under grant agreement No 888003 UP4HEALTH Project (H2020-BBI-JTI-2019) that supports the work of P. Otero, and C. Lourenço-Lopes, and to AlgaMar enterprise for the collaboration and algae material provision.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Proceeding Paper* **Semiconductor Oxide Gas Sensors: Correlation between Conduction Mechanisms and Their Sensing Performances †**

**Ambra Fioravanti 1,\* , Sara Morandi <sup>2</sup> and Maria Cristina Carotta <sup>1</sup>**


**Abstract:** In this work, a variety of semiconducting oxides were prepared and principally characterized by means of spectroscopic techniques (absorbance FT-IR, diffuse reflectance UV-Vis-NIR) to shed light on the electronic properties and defects involved at the roots of gas sensing capabilities. The thick films were obtained by screen printing technology on which electrical characterization and gas sensing measurements were performed. From the cross analysis of the results, a description of the specific sensing mechanism for each material is proposed.

**Keywords:** thick film gas sensors; nanostructured semiconductor oxides; UV-Vis-NIR and FT-IR spectroscopies; electrical characterization; sensing mechanisms

#### **1. Introduction**

The adsorption of a gas on the surface of a semiconducting oxide can induce a significant change in the electrical resistance of the material. This effect is at the basis of the development of chemiresistors for gas detection [1]. Due to their high sensitivity, tunable selectivity, easy production, small dimensions, and low cost, they are successfully used in a broad range of applications (pollutant monitoring, food quality control, industrial system control, and medical diagnosis) to detect a large number of gaseous compounds. Despite this, an increasing demand of gas sensors with high performances has been documented [2]. Many actions can be made to improve the sensing performances, such as the synthesis of nanostructures with a high specific surface area and the loading with noble metals, but the first issue is to understand the sensing mechanism of the materials and their sensing properties [3,4].

The IR and the UV-Vis spectroscopies are excellent experimental tools for investigating the electronic properties and surface chemistry of a large class of metal oxides used in the fabrication of solid state devices for gas sensing [5].

This work is aimed to determine the electronic properties for a variety of semiconducting oxides (single or combined, such as SnO2 MoO3, WO3, ZnO, TiO2, Ti-Sn, W-Sn, Mn-W mixed oxides, etc.) and to correlate them with the sensing mechanism and the sensor performances.

#### **2. Materials and Methods**

A selection of semiconducting oxides (single or combined, such as SnO2 MoO3, WO3, ZnO, TiO2, Ti-Sn, W-Sn, Mn-W mixed oxides, etc.) were prepared using wet chemistry methods [6,7]. The synthesis procedures were optimized to obtain nanopowders with a single phase, and a homogenous distribution in grain size (analyzed by X-ray diffraction and scanning electron microscopy, respectively) was presented.

**Citation:** Fioravanti, A.; Morandi, S.; Carotta, M.C. Semiconductor Oxide Gas Sensors: Correlation between Conduction Mechanisms and Their Sensing Performances. *Chem. Proc.* **2021**, *5*, 71. https://doi.org/ 10.3390/CSAC2021-10472

Academic Editor: Ye Zhou

Published: 30 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

The powders were characterized by means of spectroscopic techniques with the aim to evaluate the electronic properties and defects involved in the sensing processes. Absorbance IR spectra were run on a Perkin–Elmer System 2000 FT-IR spectrophotometer, equipped with a Hg–Cd–Te cryodetector, working in the range of wavenumbers 7800–580 cm<sup>−</sup>1. Diffuse reflectance spectra in the UV-Vis-NIR region were run on a Varian Cary 5 spectrophotometer, working in the range of wavenumbers 53,000–4000 cm<sup>−</sup>1. The spectroscopic responses were studied in the range from room temperature to 500 ◦C, both for the interaction with pure gases and for mixture gas/O2 at different concentrations depending on the oxide [8,9].

For obtaining thick films for electrical characterization and gas sensing measurements, the functional materials were added to an organic vehicle together with a small percentage of glass frit. Then, they were deposited on alumina substrates with interdigitated Au contacts and a heating element, finally fired at 650 ◦C. The flow-through technique was used maintaining a flow rate of 0.5 L/min, using synthetic air as carrier gas in dry conditions for the: (i) conductance measurements vs. temperature (room temperature of 650 ◦C); (ii) surface potential barrier height measurements to determine the intergranular energy barrier (Schottky barrier) versus temperature, following the method of stimulated temperature conductance measurements, as described by Clifford and Tuma [8,10]; and (iii) dynamical responses obtained in the presence of a mixture of different gases and operating temperature from 350 to 550 ◦C. The sensor response was calculated as the ratio between the conductance in the presence of the gas test and the conductance in air.

Finally, a sensing mechanism was proposed for each material by combining the results of spectroscopic and electrical characterization.

#### **3. Results and Discussion**

In this section, the main findings of some single (SnO2, WO3, and TiO2) and mixed (Ti-Sn and W-Sn) oxides, considering the carbon monoxide gas test, are summarized. More detailed characterizations are reported in [8,9].

The spectroscopic measurements on single oxides show that, on the one hand, SnO2 is characterized by the presence of mono-ionized oxygen vacancies (evidence in the medium IR region); on the other hand, WO3 is characterized by the presence of polarons, i.e., electrons trapped in deep levels in the band gap (evidenced in the Vis-NIR region). TiO2 shows both mono-ionized oxygen vacancies and polarons. The amount of defects increases upon CO interaction at increasing temperatures.

As for mixed oxide Sn1-xTixO2, the spectroscopic measurements in CO reveal two different behaviors: (i) samples with x = 0.1 shows absorption related to the photo-ionization of mono-ionized oxygen vacancies, as SnO2 shows; and (ii) samples with 0.3 ≤ x ≤ 0.9 show the increase in an absorption near to the VB–CB edge, as TiO2 shows. Samples with x = 0.2 is a borderline sample, showing behavior in between those of SnO2 and TiO2. This behavior was confirmed by electrical measurements.

Combining the results of spectroscopic and electrical characterization, two detection mechanisms emerge depending on the kind of chemical reaction involved. The first occurs between ionosorbed oxygen atoms and CO, with a consequent electron transfer from surface to bulk, resulting in a conductance increase and a consequent barrier height decrease. The second occurs between surface lattice oxygen atoms and CO; the bond electrons of the surface lattice oxygen atoms do not contribute to the formation of the spatial charge region and the Schottky barrier. To conclude, the prevalent gas detection mechanism in the materials with x < 0.3 is that usually occurs through Schottky barrier modification, whilst x ≥ 0.3 is based on bond electrons which, after the reaction, enter the conduction band without affecting the barrier height, but only bulk conductance.

Concerning the CO responses, the Ti-Sn solid solutions offer higher responses than those of pristine oxides, and the solution with x = 0.25 results in the best material to detect carbon monoxide.

The spectroscopic characterization of W-Sn mixed oxides highlighted the presence of polaron levels, with a position not affected by the Sn content, for mixed oxides with Sn molar content up to 33% (as for WO3) and mono-ionized oxygen vacancies for the mixed oxide with Sn molar content of 89% (as for SnO2).

The different positions of the defect levels in the band gap cause the formation of surface potential barriers significantly lower for WO3 and for mixed oxides with Sn molar content up to 33% (WO3-like samples) than for SnO2 and for the mixed oxide with Sn molar content of 89% (SnO2-like sample). This result allows to define the correlation between the electronic levels associated with the defects and the surface potential barriers in the air and in reducing atmospheres. In particular, the electrical measurements indicate that the changes in the already-low barriers of WO3 and WO3-like samples are almost negligible in the presence of a reducing gas, such as CO; otherwise, CO significantly decreases the barriers of SnO2 and SnO2-like sample. These results are completely in agreement with the low CO sensitivity of WO3 and WO3-like samples, and with the better CO sensitivity of SnO2 and SnO2-like sample.

#### **4. Conclusions**

In the electrical characterization, the main parameter typically measured is the conductance. All the operating characteristics of the sensors are derived from this measurement, considering the strength and the weakness of semiconductor sensors. On the one hand, it is simple and easily measured, but it is a second-order parameter that, although very sensitive to some reactions at the solid surface, is not a direct indicator of the exact processes taking place. For this reason, we investigated the behavior of different oxide materials by means of IR and UV-Vis spectroscopies to enlighten surface reactions and electronic properties and coupling the results to those of electrical characterization. We demonstrate the possibility to describe the processes involved in the detection mechanism with a method which can be applied to every functional material characterized towards every gas of interest.

**Author Contributions:** Conceptualization, characterizations, experiments conduction and data analysis A.F., S.M. and M.C.C.; writing—original draft preparation A.F. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


#### *Proceeding Paper* **Cross-Sensitive Potentiometric Sensors Based on Anti-Crown (C6HgF4)3 †**

**Ekaterina Yuskina 1, Kirill Tugashov 2, Vladimir B. Shur 2, Irina A. Tikhonova 2, Vasily Babain <sup>1</sup> and Dmitry Kirsanov 1,\***


**Abstract:** In this work, we explore the possibility of using anti-crown ether (C6HgF4)3 as a membraneactive component for potentiometric cross-sensitive sensors. Anti-crown ligands have already been employed as ionophores in plasticized polymeric membranes; however, the results of these studies are contradictory. In order to clarify the electrochemical sensitivity patterns of anti-crown-based sensors, we have studied plasticized polymeric membranes containing cation and anion-exchanging additives and various solvent-plasticizers. We explored the electrochemical sensitivity of these membranes in a wide variety of aqueous solutions of inorganic salts. Alkaline, alkaline-earth, and d-element salts with different anions were studied. It was found that the sensors based on anti-crown (C6HgF4)3 exhibit cationic sensitivity, and no considerable anionic responses were observed.

**Keywords:** anti-crown; potentiometric sensors; plasticized polymeric membranes

**Citation:** Yuskina, E.; Tugashov, K.; Shur, V.B.; Tikhonova, I.A.; Babain, V.; Kirsanov, D. Cross-Sensitive Potentiometric Sensors Based on Anti-Crown (C6HgF4)3. *Chem. Proc.* **2021**, *5*, 72. https://doi.org/ 10.3390/CSAC2021-10424

Academic Editor: Xin Wu

Published: 30 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

#### **1. Introduction**

In this work, we explore the possibility of using the so-called anti-crown compound (C6HgF4)3 as a membrane-active component for potentiometric cross-sensitive sensors. Anti-crown ethers are macrocyclic organometallic compounds consisting of mercury atoms or tin atoms separated by carbon atoms, including fluorinated macrocycles and mercurycarborands.

The chemistry of anti-crowns is quite extensive and it is a rapidly grown field of research. Recently, it was reported that (o-C6F6Hg)3 is readily coordinated with bromide and iodide anions to yield the following complexes: {o-[(C6F6Hg)3Br]−} and {o-[(C6F6Hg)3I]−}. When anti-crowns interact with halogens, they form polydecker wedge-shaped sandwiches: [(···(C6F6Hg)3···X···)n] <sup>n</sup>−, where X = chloride and bromide anions [1]. The ability of anticrown ethers to coordinate with single-charge anions has attracted the attention of analytical chemists. Several attempts have been made to create sensors based on anti-crown ethers to detect anions. For example, mercury-carborand was used as an ionophore for the determination of chloride ions [2]. The authors made a PVC-plasticized membrane consisting of NPOE as a plasticizer, TDDMA-Cl as an anion exchange additive, and mercury-carborand-3 as an ionophore. It was experimentally shown that the sensors exhibit a near-Nernstian response when interacting with Cl− in a wide range of concentrations, and detection limits were in the micromole range. It was also found that the presence of TDDMA-Cl in the membrane causes an increase in sensor sensitivity. Another study explored three mercury anti-crown ethers to develop sensors for the Me4N<sup>+</sup> cation [3]. The authors of the article synthesized three different PVC-plasticized membranes with various plasticizers: bis(2-ethylhexyl) sebacate (DOS), dioctyl phthalate (DOP), and dibutyl phthalate (DBP). Potassium tetrakis(4-chlorophenyl)-borate (KTpClPB) was used as a cation exchanger. A near-Nernstian response of 54.2 mV/dec towards the Me4N+ cation was observed when DOS was employed as a plasticizer. With other plasticizers, the responses were smaller. The electrode with the ionophore was more sensitive, with a response of 54.9 mV/dec. Figure 1 shows several typical anti-crown ether structures.

**Figure 1.** Structural formulas of typical anti-crown ethers based on fluorinated macrocycles and mercury-carborand: (**1**) Containing 3 mercury atoms; (**2**) containing 4 and 5 mercury atoms.

#### **2. Experimental Part**

#### *2.1. Reagents*

Three-mercury anti-crown ether was used as an ionophore in all membranes. The anticrown was synthesized at A.N. Nesmeyanov Institute of Organoelement Compounds. The polymer matrix of the membranes was made of PVC (poly(vinylchloride)). O-nitrophenyloctyl ether (NPOE) and dioctyl sebacate (DOS) were used as a solvent-plasticizers. NaTpClPB was used as an anionic additive, and TDDMA-NO3 was used as a cationic additive. All these substances were from SigmaAldrich, at Selectophore grade. We made three membranes of each type to assess repeatability of the results.

#### *2.2. ISE Preparation and Potentiometric Measurements*

The composition of the sensor membranes is shown in Table 1. The membranes were made by mixing the following components: 200 mg plasticizer, 109 mg PVC, 17 mg anticrown ether, 30 mg NaTpClPB (membranes 2 and 5), and 20 mg TDDMA-NO3 (membranes 3 and 6). The components where mixed with 5 mL THF using a magnetic stirrer until completely dissolved. After that, the mixture was transferred to a flat-bottom teflon beaker and left overnight for solvent evaporation. The disks (7 mm in diameter) were cut from the parent membrane and glued into electrode bodies. After the glue dried, the housing was connected to the electrodes and filled with 0.01 M NaCl solution. Finally, the electrodes were soaked for a day in a solution of sodium chloride of the same concentration. Between subsequent measurements, the sensors were stored in air. In total, we made 6 membranes and 18 sensors.


**Table 1.** Membrane compositions.

Solutions with a concentration of 1 M were prepared by weight, and less concentrated solutions were prepared by sequential volume dilution of the parent solution. All solutions were prepared with doubly distilled water.

The galvanic cell for the potentiometric measurements was the following:

Ag|AgCl, KCl(sat.)|analyzed solution|PVC membrane|NaCl, 0.01 M, AgCl|Ag

The reference electrode was a silver chloride electrode filled with a saturated solution of potassium chloride. A glass electrode was used to control the pH during the experiment. All measurements were carried out at room temperature.

#### **3. Results and Discussion**

#### *3.1. Anion Sensitivity*

To study the anion sensitivity of the sensors, a series of measurements was carried out in aqueous solutions of Li2SO4, NaCl, NaOAc, NaF, and Ca(NO3)2 in the concentration range of 10−6–10−<sup>2</sup> M. Based on the results of three parallel measurements, the slopes of the linear parts (10−5–10−<sup>2</sup> M) of the calibration curves were calculated. The sensor compositions 2 and 5 (Table 2) contain cation-exchange NaTpClPB, compositions 3 and 6-TDDMA-NO3 contain an anion-exchange additive.

**Table 2.** Response characteristics of the electrodes to anions (±2 mV/dec).


As can be observed, the developed sensors in most cases do not demonstrate pronounced sensitivity to anions. The sensitivity values obtained for chloride for the sensors

of composition 6 (−40.85 mV/dec with NPOE plasticizer) are the closest to the theoretical values. The sensors without any ion-exchanging additive showed cation sensitivity, thus suggesting that three mercury anti-crown ether promotes cation sensitivity.

#### *3.2. Cation Sensitivity*

We also studied the sensitivity of the electrodes to cations. A broad variety of inorganic cations was studied: Li+, Na+, Mg2+, K+, Ca2+, Co2+, Ni2+, Cu2+, Zn2+, Sr2+, Cd2+, Cs+, Pb2+, La3+, Eu3+, and Lu3+. The measurements were performed in the concentration range of 10−7–10−<sup>3</sup> M for lanthanides solutions; other salts were in the concentration range of 10−6–10−<sup>2</sup> M. The slopes were calculated for the concentration range of 10−5–10−<sup>2</sup> M (10<sup>−</sup>5–10−<sup>3</sup> M for La3+, Eu3+, and Lu3+). The results are provided in Table 3.


**Table 3.** Cation sensitivities of the electrodes (±2 mV/dec).

It can be seen that the sensors have pronounced cross-sensitivity to cations. The sensors of compositions 3 and 6 in most cases showed anion sensitivity, apparently due to the presence of the anion exchanger. The sensor with a cation-exchange additive based on the plasticizer DOS (composition 2) was more sensitive to all the cations than the sensor without an additive with the same plasticizer (composition 1). For example, for Cs<sup>+</sup> and Pb2+, the slopes for sensor 1 were 35.8 mV/dec and 2.1 mV/dec, and the slopes for sensor 2 were 51.2 mV/dec and 18.7 mV/dec, respectively. Comparing sensors based on different plasticizers (sensors 2 and 5; plasticizers DOS and NPOE), we can say that for most of the cations (Li+, Mg2+, Ni2+, Cu2+, Sr2+, and Cd2+), the values of the slope of the electrode function are higher for the sensor based on o-nitrophenyloctyl ether (composition 5). For example, for Cd2+, the slope for sensor 2 was 8.4 mV/dec, and the slope for sensor 5 was 12.9 mV/dec. At the same time, it is worth noting that the observed sensitivity values in most cases significantly differ from the theoretical values. The developed sensors did not show sensitivity to the La3+, Eu3+, and Lu3+ cations.

#### **4. Conclusions**

We studied the possibility of using a three-mercury anti-crown as an ionophore. Also, the electrochemical sensitivity of developed sensor membranes in solutions of inorganic anions and cations was studied. It was found that the presence of three-mercury anti-crown ether in the polymer plasticized membrane promoted the cation sensitivity of the sensors.

In general, the developed sensors may have potential in the development of potentiometric multi-sensor systems; however, further studies are needed to confirm that there is some benefit to using anti-crown compounds as ionophores, since the observed sensitivities followed the lipophilicity of the cations (the highest values were observed for cesium and lead).

**Author Contributions:** Conceptualization, D.K. and V.B.; methodology, E.Y., K.T., V.B.S. and I.A.T.; formal analysis, E.Y.; investigation, E.Y., K.T., V.B.S. and I.A.T.; resources, D.K.; writing—original draft preparation, E.Y.; writing—review and editing, D.K. and V.B.; visualization, E.Y.; supervision, D.K. and V.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

