Simultaneous Analysis of Paracetamol and Diclofenac Using MWCNTs-COOH Modified Screen-Printed Carbon Electrode and Pulsed Potential Accumulation

A differential-pulse adsorptive stripping voltammetric (DPAdSV) procedure with the use of pulsed potential accumulation and carboxyl functionalized multiwalled carbon nanotubes modified screen-printed carbon electrode (SPCE/MWCNTs-COOH) was delineated for simultaneous analysis of paracetamol (PA) and diclofenac (DF). The use of carboxyl functionalized MWCNTs and pulsed potential accumulation improves the analytical signals of PA and DF, and minimizes interferences from surfactants. After optimization of analytical conditions for this sensor, the peak currents of the two compounds were found to increase linearly with the increase in their concentration (5.0 × 10−9–5.0 × 10−6 mol L−1 with a detection limit of 1.4 × 10−9 mol L−1 for PA, and 1.0 × 10−10–2.0 × 10−8 mol L−1 with a detection limit of 3.0 × 10−11 mol L−1 for DF). For the first time, the electrochemical sensor allows simultaneous determination of PA and DF at concentrations of 24.3 ± 0.5 nmol L−1 and 3.7 ± 0.7 nmol L−1, respectively, in wastewater samples purified in a sewage treatment plant.


Introduction
Paracetamol (PA) is a very popular drug with an antipyretic effect, caused by inhibition of prostaglandin synthesis in the central nervous system. PA has potent antipyretic and analgesic effects, but no anti-inflammatory effect. Indications for administration of the drug include fever and acute and chronic pain. PA is recommended by the World Health Organization as one of the basic drugs in the treatment of pain during cancer. In addition, it is used for headaches, including migraine, earaches, toothaches, menstruation, and neuralgia, as well as rheumatic, myofascial, bone, postoperative, and other pains [1,2].
Diclofenac (DF) belongs to the group of nonsteroidal anti-inflammatory drugs (NSAIDs). Thanks to its chemical structure, it is classified as a phenylacetic acid derivative. DF exhibits activities characteristic of the NSAID group, that is, anti-inflammatory, analgesic, antipyretic, and inhibiting platelet aggregation. The basis of the mechanism of action is inhibition of cyclooxygenase, an enzyme involved in the synthesis of prostaglandins from cell membrane lipids. DF is used to treat inflammation available screen-printed carbon/carbon nanofibers electrode (SPCE/CNFs, DropSens, Llanera, Spain, Ref. 110CNF).
The microscopic images of the SPCE/MWCNTs-COOH sensor surface were recorded using an optical microscope and a high-resolution scanning electron microscope Quanta 3D FEG (FEI, Hillsboro, OR, USA).

DPAdSV Procedure
Under optimized conditions, differential-pulse adsorptive stripping voltammetric determinations of PA and DF were performed in 0.15 mol L −1 acetate buffer (pH of 4.0 ± 0.1) using pulsed potential accumulation ( Figure 1). The procedure consisting of a 1 s accumulation period at a potential of 0.1 V (the anodic pulse) and a 1 s accumulation period at a potential of −0.25 V (the cathodic pulse) was repeated 30 times. The differential-pulse scans from −0.25 to −0.254 V with an amplitude (A) of 150 mV, a modulation time (t m ) of 20 ms, and a scan rate (ν) of 150 mV s −1 were recorded after 29 accumulation cycles. In the last cycle, the differential-pulse scan from −0.25 to 1.5 V was recorded with the parameters described above.

DPAdSV Procedure
Under optimized conditions, differential-pulse adsorptive stripping voltammetric determinations of PA and DF were performed in 0.15 mol L −1 acetate buffer (pH of 4.0 ± 0.1) using pulsed potential accumulation ( Figure 1). The procedure consisting of a 1 s accumulation period at a potential of 0.1 V (the anodic pulse) and a 1 s accumulation period at a potential of −0.25 V (the cathodic pulse) was repeated 30 times. The differential-pulse scans from −0.25 to −0.254 V with an amplitude (A) of 150 mV, a modulation time (tm) of 20 ms, and a scan rate (ν) of 150 mV s −1 were recorded after 29 accumulation cycles. In the last cycle, the differential-pulse scan from −0.25 to 1.5 V was recorded with the parameters described above.

HPLC/PAD Procedure
The chromatographic analysis was based on literature data [20] with a slight modification of the eluent composition. A mixture of acetonitrile and water with 0.025% of trifluoroacetic acid in proportion of 60:40 v/v for DF and 15:85 v/v for PA was used in analysis. The flow rate of the mobile phase was 1.0 mL min −1 and the temperature of the thermostat was set to 25 °C. Injection volumes were 80 µL. All samples were analysed at a wavelength of 276 nm for DF and 248 nm for PA 9 (n = 3).

HPLC/PAD Procedure
The chromatographic analysis was based on literature data [20] with a slight modification of the eluent composition. A mixture of acetonitrile and water with 0.025% of trifluoroacetic acid in proportion of 60:40 v/v for DF and 15:85 v/v for PA was used in analysis. The flow rate of the mobile Materials 2020, 13, 3091 4 of 16 phase was 1.0 mL min −1 and the temperature of the thermostat was set to 25 • C. Injection volumes were 80 µL. All samples were analysed at a wavelength of 276 nm for DF and 248 nm for PA 9 (n = 3).

Direct Analysis of Water Samples
The Bystrzyca river water samples (Lublin, Poland) and waste effluents purified in a sewage treatment plant (Lublin, Poland) were analyzed using the voltammetric and chromatographic methods. The samples were directly analyzed without sample pretreatment procedure.

Screen-Printed Electrode Selection and Surface Studies
In order to compare the PA (2.0 × 10 −6 mol L −1 ) and DF (2.0 × 10 −8 mol L −1 ) signals at commercially available screen-printed carbon sensors (screen-printed carbon electrode, SPCE; carboxyl functionalized multiwalled carbon nanotubes modified SPCE, SPCE/MWCNTs-COOH; carbon nanofibers modified SPCE, SPCE/CNFs), the differential-pulse adsorptive stripping voltammetric curves were registered ( Figure 2). PA and DF were accumulated at a constant value of potential of −0.25 V (E acc. ) for 60 s (t acc. ). The results demonstrated the small peaks of PA (2.1 µA) and DF (1.0 µA) at the SPCE (curve a). When the surface of the working electrode was coated with carbon nanofibers (curve b), the PA peak current was grown to 5.8 µA, but the DF signal was ill-defined (0.12 µA). The CNTs blocked the active surface of electrode for the DF molecules. In the case of the SCPE modified with MWCNTs-COOH, two well-defined peaks of PA (5.0 µA) and DF (2.3 µA) are visible (curve c). Moreover, the lowest background current was obtained at the SPCE/MWCNTs-COOH. It is obvious that, in the case of simultaneous determination of PA and DF, the SPCE/MWCNTs-COOH should be chosen. However, for the individual PA determination, the SPCE/CNFs should be used. These results perfectly confirm our previous research already described in the literature [2,4]. In the next step of the experiments, attempts were made to explain these differences between the size of PA and DF signals at the electrodes.

of 17
The Bystrzyca river water samples (Lublin, Poland) and waste effluents purified in a sewage treatment plant (Lublin, Poland) were analyzed using the voltammetric and chromatographic methods. The samples were directly analyzed without sample pretreatment procedure.

Screen-Printed Electrode Selection and Surface Studies
In order to compare the PA (2.0 × 10 −6 mol L −1 ) and DF (2.0 × 10 −8 mol L −1 ) signals at commercially available screen-printed carbon sensors (screen-printed carbon electrode, SPCE; carboxyl functionalized multiwalled carbon nanotubes modified SPCE, SPCE/MWCNTs-COOH; carbon nanofibers modified SPCE, SPCE/CNFs), the differential-pulse adsorptive stripping voltammetric curves were registered ( Figure 2). PA and DF were accumulated at a constant value of potential of −0.25 V (Eacc.) for 60 s (tacc.). The results demonstrated the small peaks of PA (2.1 µA) and DF (1.0 µA) at the SPCE (curve a). When the surface of the working electrode was coated with carbon nanofibers (curve b), the PA peak current was grown to 5.8 µA, but the DF signal was ill-defined (0.12 µA). The CNTs blocked the active surface of electrode for the DF molecules. In the case of the SCPE modified with MWCNTs-COOH, two welldefined peaks of PA (5.0 µA) and DF (2.3 µA) are visible (curve c). Moreover, the lowest background current was obtained at the SPCE/MWCNTs-COOH. It is obvious that, in the case of simultaneous determination of PA and DF, the SPCE/MWCNTs-COOH should be chosen. However, for the individual PA determination, the SPCE/CNFs should be used. These results perfectly confirm our previous research already described in the literature [2,4]. In the next step of the experiments, attempts were made to explain these differences between the size of PA and DF signals at the electrodes. In the previously published papers [4], the electrochemical properties of SPCE and SPCE/MWCNTs-COOH were tested using CV studies in a solution of 0.1 mol L −1 KCl and 5.0 × 10 −3 mol L −1 K3[Fe(CN)6]. However, the electrochemical properties SPCE/CNFs were not studied. Therefore, the active surface of SPCE/CNFs was examined using CV in a solution of 0.1 mol L −1 KCl and 5.0 × 10 −3 mol  In the previously published papers [4], the electrochemical properties of SPCE and SPCE/MWCNTs-COOH were tested using CV studies in a solution of 0.1 mol L −1 KCl and 5.0 × 10 −3 mol L −1 K 3 [Fe(CN) 6 ]. However, the electrochemical properties SPCE/CNFs were not studied. Therefore, the active surface of SPCE/CNFs was examined using CV in a solution of 0.1 mol L −1 KCl and 5.0 × 10 −3 mol L −1 K 3 [Fe(CN) 6 ]. The cyclic voltammograms were recorded at different scan rates in the range from 5 to 500 mV s −1 ( Figure 3A). The peak-to-peak separation (∆E) for the SPCE/CNFs was estimated for the selected scan rate (175 mV s −1 ) as 169.0 ± 1.7 mV (n = 3). The results indicate the improvement of the reversibility process using CNFs-modified and especially MWCNT-COOH (∆E = 149.0 ± 1.5 mV) electrodes in comparison with the unmodified SPCE (189.0 ± 1.9 mV) [4]. The dependence between anodic peak currents (I p ) and square root of the scan rates (v 1/2 ) was plotted ( Figure 3B). On the basis of the Randles-Sevcik equation [21], the active surface area (A s ) of the SPCE/CNFs was calculated. It should be mentioned that the geometric surfaces of all electrodes are the same. For the unmodified SPCE and SPCE/MWCNTs-COOH, the A s equals 0.061 ± 0.00058 cm 2 (n = 3) and 0.10 ± 0.00097 cm 2 (n = 3) [4], respectively, while the area of SPCE/CNFs was calculated to be 0.08090 ± 0.0014 cm 2 (n = 3). The results show that the SPCE/MWCNTs-COOH has a greater number of active centers than the unmodified SPCE and the SPCE/CNFs. These results explain the enhancement of PA and DF signals in relation to the SPCE, and the DF signal in relation to the SPCE/CNFs. Moreover, DF may have a higher affinity to the SPCE/MWCNTs-COOH surface than SPCE/CNTs and SPCE owing to the surface functionalization with carboxyl (hydrophilic) groups. A slight difference in the peak current of PA at the SPCE/CNFs and SPCE/MWCNTs-COOH (5.8 µA vs. 5.0 µA, respectively) is owing to the fact that the SPCE/CNTs surface better facilitates the adsorption of PA. The electrochemical oxidation process of PA at the SPCE/CNFs surface is purely adsorption-controlled [2]. However, the goal of this work was to show the voltammetric procedure for the simultaneous analysis of DF and PA, and thus the SPCE/MWCNTs-COOH was chosen for further electrochemical study.
The cyclic voltammograms were recorded at different scan rates in the range from 5 to 500 mV s −1 ( Figure 3A). The peak-to-peak separation (ΔE) for the SPCE/CNFs was estimated for the selected scan rate (175 mV s −1 ) as 169.0 ± 1.7 mV (n = 3). The results indicate the improvement of the reversibility process using CNFs-modified and especially MWCNT-COOH (ΔE = 149.0 ± 1.5 mV) electrodes in comparison with the unmodified SPCE (189.0 ± 1.9 mV) [4]. The dependence between anodic peak currents (Ip) and square root of the scan rates (v 1/2 ) was plotted ( Figure 3B). On the basis of the Randles-Sevcik equation [21], the active surface area (As) of the SPCE/CNFs was calculated. It should be mentioned that the geometric surfaces of all electrodes are the same. For the unmodified SPCE and SPCE/MWCNTs-COOH, the As equals 0.061 ± 0.00058 cm 2 (n = 3) and 0.10 ± 0.00097 cm 2 (n = 3) [4], respectively, while the area of SPCE/CNFs was calculated to be 0.08090 ± 0.0014 cm 2 (n = 3). The results show that the SPCE/MWCNTs-COOH has a greater number of active centers than the unmodified SPCE and the SPCE/CNFs. These results explain the enhancement of PA and DF signals in relation to the SPCE, and the DF signal in relation to the SPCE/CNFs. Moreover, DF may have a higher affinity to the SPCE/MWCNTs-COOH surface than SPCE/CNTs and SPCE owing to the surface functionalization with carboxyl (hydrophilic) groups. A slight difference in the peak current of PA at the SPCE/CNFs and SPCE/MWCNTs-COOH (5.8 µA vs. 5.0 µA, respectively) is owing to the fact that the SPCE/CNTs surface better facilitates the adsorption of PA. The electrochemical oxidation process of PA at the SPCE/CNFs surface is purely adsorption-controlled [2]. However, the goal of this work was to show the voltammetric procedure for the simultaneous analysis of DF and PA, and thus the SPCE/MWCNTs-COOH was chosen for further electrochemical study. The selected three-electrode system surface consisting of an SPCE/MWCNTs-COOH (working electrode, a), an SPCE (auxiliary electrode, b), and an SPAgE (pseudo-reference electrode, c) was visualized by optical and scanning electron microscopes ( Figure 4). It is apparent that the MWCNTs-COOH adheres to the carbon and is distributed homogeneously on the surface [4]. The selected three-electrode system surface consisting of an SPCE/MWCNTs-COOH (working electrode, a), an SPCE (auxiliary electrode, b), and an SPAgE (pseudo-reference electrode, c) was visualized by optical and scanning electron microscopes ( Figure 4). It is apparent that the MWCNTs-COOH adheres to the carbon and is distributed homogeneously on the surface [4]. 6 of 17

Effect of pH
The supporting electrolyte pH influences the peak potential and current as well as the shapes of the signals of biologically active compounds. Therefore, choosing an appropriate pH value is an important step during the optimization procedure. Here, 0.  Figure 5B, the relationships between potential peaks of PA and DF and pH are shown. As can be seen, the slopes of −45.0 mV pH −1 (for PA) and −52.0 mV pH −1 (for DF) were close to the theoretical value of −59.0 mV pH −1 .
These results indicate that the number of protons and transferred electrons involved in the oxidation mechanism of PA and DF is equal [21]. As PA and DF oxidation is a two-electron process the number of protons was also predicted to be 2, indicating the 2e − /2H + process. DF is oxidized to 5-hydrohydiclofenac ( Figure 5C) and PA to N-acetyl-p-quinoneimine ( Figure 5D) [22,23].
Furthermore, it was observed that the peak current of PA and DF increased with increasing pH value to 4.0, and then the anodic peaks decreased ( Figure 5E). Therefore, the acetate buffer solution of pH 4.0 was chosen as the supporting electrolyte in the simultaneous PA and DF determination. Moreover, it was found that the highest values of PA and DF signals were attained at 0.15 mol L −1 concentration of acetate buffer solution of pH 4.0, and hence it was further used ( Figure 5F). Optical (left side) and scanning electron microscopic (right side) images of SPCE/MWCNTs-COOH surface.

Effect of pH
The supporting electrolyte pH influences the peak potential and current as well as the shapes of the signals of biologically active compounds. Therefore, choosing an appropriate pH value is an important step during the optimization procedure. Here, 0. The results indicate that the potential peaks of PA and DF shifted to less positive values as pH increased ( Figure 5A), indicating that protons were directly involved in the electrode reaction. Additionally, in Figure 5B, the relationships between potential peaks of PA and DF and pH are shown. As can be seen, the slopes of −45.0 mV pH −1 (for PA) and −52.0 mV pH −1 (for DF) were close to the theoretical value of −59.0 mV pH −1 . These results indicate that the number of protons and transferred electrons involved in the oxidation mechanism of PA and DF is equal [21]. As PA and DF oxidation is a two-electron process the number of protons was also predicted to be 2, indicating the 2e − /2H + process. DF is oxidized to 5-hydrohydiclofenac ( Figure 5C) and PA to N-acetyl-p-quinoneimine ( Figure 5D) [22,23].
Furthermore, it was observed that the peak current of PA and DF increased with increasing pH value to 4.0, and then the anodic peaks decreased ( Figure 5E). Therefore, the acetate buffer solution of pH 4.0 was chosen as the supporting electrolyte in the simultaneous PA and DF determination. Moreover, it was found that the highest values of PA and DF signals were attained at 0.15 mol L −1 concentration of acetate buffer solution of pH 4.0, and hence it was further used ( Figure 5F).

Accumulation of PA and DF at SPCE/MWCNTs-COOH and Sensor Selectivity
The   peaks at potentials around 330 and 550 mV, respectively, when the sweep was initiated in the positive direction. As can be seen, the oxidation peak potential of PA and DF shifted toward more positive values with the increasing scan rate. This confirms that PA and DF are irreversibly oxidized. Other peaks at less positive potentials are related to the formation of electrochemically active oxidation products of DF [4]. direction. As can be seen, the oxidation peak potential of PA and DF shifted toward more positive values with the increasing scan rate. This confirms that PA and DF are irreversibly oxidized. Other peaks at less positive potentials are related to the formation of electrochemically active oxidation products of DF [4].
As can be seen in Figure 6B, the linear relationships between the PA and DF peak current (Ip) and the square root of scan rate (v 1/2 ) indicated that the oxidation processes of PA (r = 0.9970) and DF (r = 0.9879) are controlled by diffusion at the SPCE/MWCNTs-COOH. However, the curve slopes of 0.67 (for PA) and 0.72 (for DF) observed in the plot of logIp versus logv ( Figure 6C) indicate that these processes are not purely diffusion-or adsorption-controlled [24]. Therefore, in the next step of the experiments, the effect of accumulation potential (Eacc.) was tested. In voltammetric procedures, even a low concentration of surface active substances can foul and passify the electrode, causing a decrease or total decay of the analytical signal. UV irradiation or microwave heating before determination are recommended for elimination of this type of interference. However, such a process makes the procedures lengthy, complicated, and more expensive; requires additional apparatus; and cannot be used in field analysis. The literature also lists different simple and cheap ways for solving the problem with the organic matrix of natural water samples, namely application of potential pulses for accumulation. In addition, this way for the minimization of interferences can be applied outside laboratories. The potential of cathode pulses was chosen in a way that made it represent the maximum adsorption of the determined element and the potential of anode pulses to desorb the interfering surfactants [25,26]. Therefore, the procedure consisting of a 1 s accumulation period at a potential of 0.1 V (the anodic pulse) and a 1 s accumulation period at a potential of −0.25 V (the cathodic pulse) was proposed for simultaneous determination of PA (1.0 × 10 −6 mol L −1 ) and DF (1.0 × 10 −8 mol L −1 ). The differential-pulse scans from −0.25 to −0.254 V with an amplitude (A) of 150 mV, a modulation time (tm) of 20 ms, and a scan rate (ν) of 150 mV s −1 were recorded after 59 accumulation cycles. In the last cycle, the differential-pulse scan from −0.25 to 1.5 V was recorded with the parameters described above. Additionally, the procedure with a constant value of accumulation potential of −0.25 V for 60 s as well as the procedure consisting of a 1 s accumulation period at a potential of −0.25 V (the cathodic pulse) and the anodic pulse with the differential-pulse scan from −0.25 to 0.1 V (ncycles = 60) were applied. As can be seen in Figure 7A, the application procedure with pulsed potential accumulation (60-times pulses of 0.1 V for 1 s and −0.25 V for 1 s) improves both PA and DF analytical As can be seen in Figure 6B, the linear relationships between the PA and DF peak current (Ip) and the square root of scan rate (v 1/2 ) indicated that the oxidation processes of PA (r = 0.9970) and DF (r = 0.9879) are controlled by diffusion at the SPCE/MWCNTs-COOH. However, the curve slopes of 0.67 (for PA) and 0.72 (for DF) observed in the plot of logIp versus logv ( Figure 6C) indicate that these processes are not purely diffusion-or adsorption-controlled [24]. Therefore, in the next step of the experiments, the effect of accumulation potential (E acc. ) was tested.
The effects of E acc. at the SPCE/MWCNTs-COOH surface were studied in the mixed solution of PA (1.0 × 10 −6 mol L −1 ) and DF (1.0 × 10 −8 mol L −1 ). Keeping the accumulation time (t acc. ) as 60 s, the dependence of stripping peak current on E acc. was evaluated over the potential range of 0.25 to −1.25 V. The peak current of PA and DF reached maximum at E acc. of −0.25 V. This value of potential was chosen for further experiments. However, the constant value of potential was changed to pulsed potential accumulation.
In voltammetric procedures, even a low concentration of surface active substances can foul and passify the electrode, causing a decrease or total decay of the analytical signal. UV irradiation or microwave heating before determination are recommended for elimination of this type of interference. However, such a process makes the procedures lengthy, complicated, and more expensive; requires additional apparatus; and cannot be used in field analysis. The literature also lists different simple and cheap ways for solving the problem with the organic matrix of natural water samples, namely application of potential pulses for accumulation. In addition, this way for the minimization of interferences can be applied outside laboratories. The potential of cathode pulses was chosen in a way that made it represent the maximum adsorption of the determined element and the potential of anode pulses to desorb the interfering surfactants [25,26]. Therefore, the procedure consisting of a 1 s accumulation period at a potential of 0.1 V (the anodic pulse) and a 1 s accumulation period at a potential of −0.25 V (the cathodic pulse) was proposed for simultaneous determination of PA (1.0 × 10 −6 mol L −1 ) and DF (1.0 × 10 −8 mol L −1 ). The differential-pulse scans from −0.25 to −0.254 V with an amplitude (A) of 150 mV, a modulation time (t m ) of 20 ms, and a scan rate (ν) of 150 mV s −1 were recorded after 59 accumulation cycles. In the last cycle, the differential-pulse scan from −0.25 to 1.5 V was recorded with the parameters described above. Additionally, the procedure with a constant value of accumulation potential of −0.25 V for 60 s as well as the procedure consisting of a 1 s accumulation period at a potential of −0.25 V (the cathodic pulse) and the anodic pulse with the differential-pulse scan from −0.25 to 0.1 V (n cycles = 60) were applied. As can be seen in Figure 7A, the application procedure with pulsed potential accumulation (60-times pulses of 0.1 V for 1 s and −0.25 V for 1 s) improves both PA and DF analytical signals. To reduce the analysis time, the effect of number of cycles (n cycles ) on the peak current of PA (1.0 × 10 −6 mol L −1 ) and DF (1.0 × 10 −8 mol L −1 ) was studied. Figure 7B shows the obtained results. For further experiments, the number of cycles was reduced to 30, as a compromise between the decrease in PA peak current and the increase in DF peak current. 10     According to the literature data, natural waters contain surfactants with the surface active effect similar to the effect induced by 0.2 to 2 ppm Triton X-100 [26]. Therefore, the effect of the use of pulsed potential accumulation of PA (1.0 × 10 −6 mol L −1 ) and DF (1.0 × 10 −8 mol L −1 ) on the minimization of interferences from surfactants was studied on the example of Triton X-100. As can be seen in Figure 7C,D, the application procedures with pulsed potential accumulation (b and c bars), compared with the application of a constant value of accumulation potential (a bars), contribute to the minimization of interferences from Triton X-100, particularly with regard to PA at a concentration of 2 ppm and upwards.
In summary, it can be stated that, in order to improve PA and DF analytical signals, as well as to minimize interferences from surfactants, pulsed potential accumulation can be applied. To our knowledge, this is the first time these two goals were achieved using pulsed potential accumulation. For further experiments, as a compromise between peak current and minimizing interference, the procedure consisting of a 1 s accumulation period at a potential of 0.1 V and a 1 s accumulation period at a potential of −0.25 V (n cycles = 30) was applied for the simultaneous determination of PA and DF.
Additionally, t m was tested in the range of 2-60 ms (A of 150 mV and ν of 150 mV s −1 ). The highest signals of PA and DF were obtained for t m of 20 ms ( Figure 8C).      Table 1. The limits of detection (LOD) and quantification (LOQ) obtained during simultaneous determination of PA and DF are 1.44 and 4.80 nmol L −1 and 0.030 and 0.1 nmol L −1 , respectively. These results demonstrate that the SPCE/MWCNTs-COOH can be applied to environmental water samples analysis in which PA and DF concentrations are in the range of 10 −9 -10 −8 and 10 −11 -10 −8 mol L −1 , respectively [7,8]. Table 2 shows the comparison techniques used for the simultaneous determination of PA and DF. It can be summarized that the DPAdSV with SPCE/MWCNTs-COOH allows the lowest LOD value to be obtained compared with all other electrochemical sensors and techniques [9][10][11][12][13][14][15][16][17][18][19].   Table 1. Characteristics of calibration plots of paracetamol (PA) and diclofenac (DF) attained at the commercially available carboxyl functionalized multiwalled carbon nanotubes modified screen-printed carbon electrodes (SPCE/MWCNTs-COOH). LOD, limit of detection; LOQ, limit of quantification.  Pharmaceutical, Lake water [18] AuNPs

Analytical Applications
Finally, the practical application of the proposed voltammetric procedure using SPCE/MWCNTs-COOH was illustrated by simultaneous determination of PA and DF in Bystrzyca river samples and wastewater samples purified in a sewage treatment plant. The voltammetric results were compared to those obtained by chromatographic method (HPLC/PAD) and summarised in Table 3. Figure 10 shows the DPAdSV curves obtained during simultaneous determination of PA and DF in the analysed samples. The results achieved by the voltammetric method show satisfactory agreement with those obtained by HPLC/PAD (the relative errors are in the range of 1.1-6.7%). In order to test the accuracy of the voltammetric procedure, the samples were spiked with standard solutions of PA and DF. The recovery values are between 96.5% and 104.8%, which corresponds to the satisfactory degree of accuracy. It needs to be highlighted that only the voltammetric procedure using the SPCE/MWCNTs-COOH allows simultaneous determination of PA and DF at concentrations of 24.3 ± 0.5 nmol L −1 and 3.7 ± 0.7 nmol L −1 , respectively, in wastewater samples purified in a sewage treatment plant. These results show that, after leaving the sewage treatment plant, the wastewater still contains PA and DF, which then end up in the natural environment. The concentrations of PA and DF in Bystrzyca river samples below the limit of detection of the DPAdSV technique confirm the dilution of analytes.

Conclusions
For the first time, in this study, carboxyl functionalized multiwalled carbon nanotubes modified screen-printed carbon electrode (SPCE/MWCNTs-COOH) was introduced for the simultaneous, direct analysis of low concentrations of paracetamol (PA) and diclofenac (DF). Moreover, for the first time, pulsed potential accumulation was used in order to improve PA and DF analytical signals and to minimize interferences from surfactants.
In this work, already published results regarding the electrochemical properties of SPCE and SPCE/MWCNTs-COOH [4] were compared with these obtained for SPCE/CNFs. The results show that the SPCE/MWCNTs-COOH has a greater number of active centers than the unmodified SPCE and the SPCE/CNFs, which explain the enhancement of PA and DF signals in relation to the SPCE, and the DF signal in relation to the SPCE/CNFs. The SPCE/MWCNTs-COOH was recommended for simultaneous analysis of PA and DF, but the SPCE/CNFs for the individual analysis of PA. Moreover, the electrochemical responses of PA and DF at the SPCE/MWCNTs-COOH in the 0.15 mol L −1 acetate buffer solution (pH 4.0) were characterized by the CV technique. The obtained results indicated that the oxidation processes of PA and DF at the SPCE/MWCNTs-COOH are not purely diffusionor adsorption-controlled.
Moreover, only the proposed voltammetric procedure using the SPCE/MWCNTs-COOH allows simultaneous determination of PA and DF at concentrations of 24.3 ± 0.5 nmol L −1 and 3.7 ± 0.7 nmol L −1 , respectively, in wastewater samples purified in a sewage treatment plant. These results show that, after leaving the sewage treatment plant, the wastewater still contains PA and DF, which then end up in the natural environment. It should be clearly emphasized that the samples were directly analysed without performing any special sample pretreatment procedure.
The proposed voltammetric procedure has the advantages of being much more sensitive, less time-consuming, and less expensive than HPLC. Moreover, the analysis of water samples can be carried out in the laboratory and at the place of sampling.