Development of the New Sensor Based on Functionalized Carbon Nanomaterials for Promethazine Hydrochloride Determination

Promethazine hydrochloride (PM) is a widely used drug so its determination is important. Solid-contact potentiometric sensors could be an appropriate solution for that purpose due to their analytical properties. The aim of this research was to develop solid-contact sensor for potentiometric determination of PM. It had a liquid membrane containing hybrid sensing material based on functionalized carbon nanomaterials and PM ions. The membrane composition for the new PM sensor was optimized by varying different membrane plasticizers and the content of the sensing material. The plasticizer was selected based on calculations of Hansen solubility parameters (HSP) and experimental data. The best analytical performances were obtained using a sensor with 2-nitrophenyl phenyl ether (NPPE) as the plasticizer and 4% of the sensing material. It had a Nernstian slope (59.4 mV/decade of activity), a wide working range (6.2 × 10−7 M–5.0 × 10−3 M), a low limit of detection (1.5 × 10−7 M), fast response time (6 s), low signal drift (−1.2 mV/h), and good selectivity. The working pH range of the sensor was between 2 and 7. The new PM sensor was successfully used for accurate PM determination in a pure aqueous PM solution and pharmaceutical products. For that purpose, the Gran method and potentiometric titration were used.


Introduction
Promethazine hydrochloride (PM) or N,N-dimethyl-1-phenothiazin-10-yl-propan-2amine hydrochloride is a widely used phenothiazine derivative. It has antihistaminic, anti-emetic, sedative, antipsychotic, and analgesic properties, so it is mainly used to treat mental illness, prevent motion sickness, and as an allergy medication [1,2]. Additionally, it can be used in a broad range of health conditions such as asthma, pneumonia, respiratory tract infections, hemorrhoids, intracranial hypertension, and neoplastic disorders, or as a melanogenesis-suppressing agent, skin-aging prevention agent, and sperm-killing agent [3]. Recently, PM has been recommended in the treatment of COVID-19 [4,5]. However, PM can cause some serious adverse effects, including severe breathing problems, depression of the central nervous system, cardiac problems, endocrine disorders, gastrointestinal effects, immunoallergic reactions, venous thrombosis, sudden infant death syndrome, tissue damage, etc. [3,6]. Even U.S. Food and Drug Administration (FDA) issued an alert in 2006, with notification that PM causes fatal respiratory depression in children younger than two years [7], and in 2009, considering the risk of severe tissue damage after intravenous injection [8]. The widespread nonmedical abuse of PM has also been observed in last few decades [9,10]. Due to its wide use, the monitoring of PM is very important. PM can be determined using different methods, including liquid chromatography-tandem

Sensor Preparation
The new sensing material (MWCNT−OSO 3 PM) was prepared by functionalization of MWCNTs (MWCNT−OH, 95+%, IoLiTec, Germany, o.d. of 20 to 30 nm and a length of 10 to 30 µm) with chlorosulfonic acid (Acros Organics B.V.B.A., Belgium) according to [31], and then mixing with the solution of PM (c = 1 × 10 −2 M). PM solution was added dropwise until the formation of a flaky precipitate. After magnetic stirring for 2 h, the formed precipitate was washed with water and dried under low pressure in a water bath at 35 • C using a rotary evaporator.

Apparatus
The Thermo Nicolet Avatar 380 FTIR with Smart Orbit Diamond ATR (Thermo Scientific, USA) was used for ATR-FTIR spectra recording. An ultrasonic bath (BANDELIN RK-100, Berlin, Germany) was used to prepare liquid membranes and solutions. The 794 Basic Titrino, 806 Exchange unit, 826 mobile pH meter, and 728 stirrer (all from Metrohm, Herisau, Switzerland) were used for measurements. Devices were controlled with Tiamo software (Metrohm, Switzerland) and in-house software.

Procedure
FTIR characterization was conceived by Thermo Nicolet 380 FTIR spectroscope using Smart Orbit diamond ATR sampling attachment. FTIR spectra were recorded with 4 cm −1 wavenumber resolution. In the range of 4000-400 cm −1 , 100 scans were recorded and averaged. The baseline was taken under ambient conditions on a cleaned ATR diamond surface. Samples of MWCNT-OSO 3 H, PM, and MWCNT-OSO 3 PM, were dried in a vacuum desiccator for 24 h before the examination (p = 0.1 mbar).
For all measurements, the new sensor was used as the indicator electrode, and Ag/AgCl electrode (Metrohm, Switzerland) as the reference electrode. The measurement conditions were: no pH and ionic strength adjustment, room temperature, and magnetic stirring.
Every day before measurements, the sensor was left for 15 min in PM (c = 1.0 × 10 −2 M) for conditioning. It was followed by calibration to check the sensor's reliability.

ATR-FTIR Characterization of MWCNT-OSO 3 PM
MWCNT−OSO 3 H was prepared and characterized by FTIR in previous study [31]. The FTIR spectrum of PM ( Figure 1) shows all characteristic peaks for PM [33]. A peak at 449 cm −1 is assigned to the stretching modes of C-S bonds. Stretching modes of some C-N bonds are observed at 755 and 1230 cm −1 , while peaks at 1100 and 1455 cm −1 are assigned to aromatic out-of-plane vibrations. These PM characteristic modes are observed in MWCNT−OSO 3 PM with small intensities, but they are not observed on the MWCNT−OSO 3 H sample, which is indirect proof of PM bonding to MWCNT−OSO 3 H. Due to the high IR absorbance of MWCNT, the small changes in FTIR spectra due to PM bonding cannot be identified or observed. The measuring devices were controlled with in-house software for all measurements except for titrations, where Tiamo software was used.

ATR-FTIR Characterization of MWCNT-OSO3PM
MWCNT−OSO3H was prepared and characterized by FTIR in previous study [31]. The FTIR spectrum of PM ( Figure 1) shows all characteristic peaks for PM [33]. A peak at 449 cm -1 is assigned to the stretching modes of C-S bonds. Stretching modes of some C-N bonds are observed at 755 and 1230 cm -1 , while peaks at 1100 and 1455 cm -1 are assigned to aromatic out-of-plane vibrations. These PM characteristic modes are observed in MWCNT−OSO3PM with small intensities, but they are not observed on the MWCNT−OSO3H sample, which is indirect proof of PM bonding to MWCNT−OSO3H. Due to the high IR absorbance of MWCNT, the small changes in FTIR spectra due to PM bonding cannot be identified or observed.

Response of the Sensor
The newly developed sensor is a solid-contact ISE. Its liquid membrane, which is responsible for the response of the sensor, contains MWCNTs covalently functionalized with a sulfate group and PM ion as an ionophore. The sensor has a potentiometric response to PM according to the Nernst equation: where E is measured electrode potential, E 0 is standard electrode potential, S is the slope of the sensor, and is the activity of the PM ion. The Nernstian slope representing the theoretical slope for PM ion is 59.2 mV/decade of activity at 25 °C.
The MWCNT−OSO3PM in the sensor membrane dissociates according to the Equation (2): MWCNTs are especially convenient to use as part of the ionophore considering their good electrical properties that makes MWCNT−OSO3 − sensitive to changes in a solution of the analyte. Additionally, their hydrophobicity and structure lead to the immobility of MWCNT−OSO3 − in the membrane and the formation of an electrostatic barrier for anions penetration. PM ions in the membrane close to the MWCNT−OSO3 − regulate their charge. As a result, the ion-to-electron transduction between the electrons in the MWCNT−OSO3 − wall and PM ions in the membrane occurs. The electrical double layer on MWCNT−OSO3 − near the interface between membrane and solution is related to the concentration of PM ions in solution. Subsequently, an electromotive force is generated

Response of the Sensor
The newly developed sensor is a solid-contact ISE. Its liquid membrane, which is responsible for the response of the sensor, contains MWCNTs covalently functionalized with a sulfate group and PM ion as an ionophore. The sensor has a potentiometric response to PM according to the Nernst equation: where E is measured electrode potential, E 0 is standard electrode potential, S is the slope of the sensor, and a PM + is the activity of the PM ion. The Nernstian slope representing the theoretical slope for PM ion is 59.2 mV/decade of activity at 25 • C. The MWCNT−OSO 3 PM in the sensor membrane dissociates according to the Equation (2): MWCNTs are especially convenient to use as part of the ionophore considering their good electrical properties that makes MWCNT−OSO 3 − sensitive to changes in a solution of the analyte. Additionally, their hydrophobicity and structure lead to the immobility of MWCNT−OSO 3 − in the membrane and the formation of an electrostatic barrier for anions penetration. PM + ions in the membrane close to the MWCNT−OSO 3 − regulate their charge. As a result, the ion-to-electron transduction between the electrons in the MWCNT−OSO 3 − wall and PM + ions in the membrane occurs. The electrical double layer on MWCNT−OSO 3 − near the interface between membrane and solution is related to the concentration of PM + ions in solution. Subsequently, an electromotive force is generated [34][35][36]. The described type of sensor can be successfully applied for the potentiometric determination of PM.

Selection of the Plasticizer
The plasticizer is a predominant component of the ISE membrane. Its main function is to reduce the viscosity and improve the physical and mechanical properties of the membrane, thus improving the mobility inside the membrane. Additionally, it is wellknown that its nature, especially the polarity, significantly impacts the ISE's selectivity, slope, and LOD [30,37].
The plasticizer's insolubility in water and mixability with the PVC matrix determines the physical properties of the membrane and the stability of the sensor. At the same time, its molecular compatibility with the analyte determines its sensing properties. Several theoretical methods can calculate molecular interaction properties. For the plasticizer selection, calculations of Hansen solubility parameters (HSP) were performed. HSP splits a liquid's total cohesion energy into contributions from hydrogen bonding (δh), atomic dispersion (δd), and polar interactions (δp), which forms a 3D Hansen parameter space [38]. The logarithm of the partition coefficient (logP), logarithm of the solubility (logS), and HSP for plasticizers and PM were calculated by Hansen Solubility Parameters in Practice (HSPiP) software version 5.2.02., using simplified molecular-input line-entry system (SMILES) obtained from the open chemistry database PubChem. The calculated data are shown in Table 1. Materials with logP > 3 and logS < −1 are considered water-insoluble, which is the requirement that all selected plasticizers meet. The plasticizer's affinity toward the analyte (PM) can be estimated from R a , the HSP distance of two materials in the 3D Hansen parameter space. It shows molecular similarity. The smallest value shows the largest similarity and likely the best solubility of PM in a given plasticizer. From this point of view, NPPE is predicted to be the best plasticizer for PM selective membranes. Experimental measurements were also performed to select the best plasticizer for the new PM sensor. Responses of five ISEs containing five different plasticizers (DBP, o-NPOE, DS, NPPE, and DOP) were measured. All ISEs investigated contained 2% of sensor material (MWCNT−OSO 3 PM), and the weight ratio of plasticizer and PVC was 2:1. The response characteristics of ISEs towards PM were investigated in a concentration range between 2.5 × 10 −8 M and 5.0 × 10 −3 M. The results are presented in Table 2. The statistical data were based on five repeated measurements and calculated using linear regression analysis. The IUPAC recommendations [39] were used for the estimation of LOD. Regardless of the plasticizer used, all sensors revealed a sub-Nernstian response. However, the sensor with NPPE revealed the slope value closest to Nernstian (55.8 mV/decade of activity). The LOD was lowest for a sensor with DS as a plasticizer, but it was not selected for further investigation due to its lower slope value. Considering the results and theoretical calculations obtained using HSPiP, the sensor with NPPE as a plasticizer was chosen for additional research.

Optimization of the Sensor Material Content
After selecting the most suitable plasticizer for the new PM sensor, the influence of the content of the sensor material on the response characteristics of the new sensor was investigated. It was based on the results of measuring the response to PM of three ISEs containing NPPE as a plasticizer and different content of MWCNT−OSO 3 PM (2%, 4%, and Sensors 2023, 23, 2641 6 of 11 6%). The weight ratio of the plasticizer and PVC was 2:1. The resulting response curves and their statistics, based on five repeated measurements, are shown in Figure 2 and Table 3.

Optimization of the Sensor Material Content
After selecting the most suitable plasticizer for the new PM sensor, the influence of the content of the sensor material on the response characteristics of the new sensor was investigated. It was based on the results of measuring the response to PM of three ISEs containing NPPE as a plasticizer and different content of MWCNT−OSO3PM (2%, 4%, and 6%). The weight ratio of the plasticizer and PVC was 2:1. The resulting response curves and their statistics, based on five repeated measurements, are shown in Figure 2 and Table  3. It can be seen that the sensor with 4% of sensor material in its membrane exhibited the Nernstian slope (59.4 mV/decade of activity), while the other two sensors investigated exhibited the sub-Nernstian slope. Additionally, it has the lowest LOD and the widest measuring range, so it was selected for further research.  It can be seen that the sensor with 4% of sensor material in its membrane exhibited the Nernstian slope (59.4 mV/decade of activity), while the other two sensors investigated exhibited the sub-Nernstian slope. Additionally, it has the lowest LOD and the widest measuring range, so it was selected for further research.

Dynamic Response
After immersion of the sensor in the analyte solution, it takes some time for the potential to stabilize. The time that is taken for the sensor to reach 90% of the final value of the potential after a sudden increase in the analyte concentration represents the dynamic response time of the sensor [40]. The response time of the new PM sensor was determined by measuring its potential in an analyte solution where PM concentration was suddenly changed every 30 s in a concentration range between 1.0 × 10 −7 M and 1.0 × 10 −3 M. The results are presented in Figure 3, where it can be seen that the new sensor has a very fast response. Its average response time was only 6 s.

Dynamic Response
After immersion of the sensor in the analyte solution, it takes some time for the potential to stabilize. The time that is taken for the sensor to reach 90% of the final value of the potential after a sudden increase in the analyte concentration represents the dynamic response time of the sensor [40]. The response time of the new PM sensor was determined by measuring its potential in an analyte solution where PM concentration was suddenly changed every 30 s in a concentration range between 1.0 × 10 −7 M and 1.0 × 10 −3 M. The results are presented in Figure 3, where it can be seen that the new sensor has a very fast response. Its average response time was only 6 s.

Signal Drift
ISEs suffer from the change of the potential over time (signal drift). It can affect the analytical performance of the sensor. Considering the above, the signal drift of the new PM sensor was measured in PM solution (c = 4.0 × 10 −3 M) for five hours. Based on the calculation obtained using the linear regression analysis, it can be described by the equation E (mV) = −0.0006·t(s) + 179.61. The signal drift for the new PM sensor amounted to -1.2 mV/hour.

The Influence of the pH
The effect of pH value on the potentiometric response of the new PM sensor was examined in the pH range between 2 and 10 in PM solution (c = 4.0 × 10 −3 M). The results are presented in Figure 4. It can be concluded that the measuring pH range of the new PM sensor is between 2 and 7, due to the potential stability. At pH 8 or above, a significant potential decrease was detected because of the decrease of the ionic form of PM in test

Signal Drift
ISEs suffer from the change of the potential over time (signal drift). It can affect the analytical performance of the sensor. Considering the above, the signal drift of the new PM sensor was measured in PM solution (c = 4.0 × 10 −3 M) for five hours. Based on the calculation obtained using the linear regression analysis, it can be described by the equation E (mV) = −0.0006·t(s) + 179.61. The signal drift for the new PM sensor amounted to −1.2 mV/h.

The Influence of the pH
The effect of pH value on the potentiometric response of the new PM sensor was examined in the pH range between 2 and 10 in PM solution (c = 4.0 × 10 −3 M). The results are presented in Figure 4. It can be concluded that the measuring pH range of the new PM sensor is between 2 and 7, due to the potential stability. At pH 8 or above, a significant potential decrease was detected because of the decrease of the ionic form of PM in test solution. It is a result of the hydrolysis reaction and consequently, formation of the free promethazine base [19,20,41].

The Selectivity
Successful determination of the analyte mostly depends on the selectivity of the sensor for the analyte. For that purpose, potentiometric selectivity coefficients (K pot A, B ) were determined using the fixed interference method [39] and Solver (Microsoft Excel) for mathematical adjustment of the Nikolskii-Eisenman equation (Equation (3)) to experimental data.
In Equation (3), a A and z A represent activity and charge of the analyte ion, respectively, and a B and z B represent activity and charge of the interfering ion, respectively. solution. It is a result of the hydrolysis reaction and consequently, formation of the free promethazine base [19,20,41].

The Selectivity
Successful determination of the analyte mostly depends on the selectivity of the sensor for the analyte. For that purpose, potentiometric selectivity coefficients ( , ) were determined using the fixed interference method [39] and Solver (Microsoft Excel) for mathematical adjustment of the Nikolskii-Eisenman equation (Equation (3)) to experimental data.
In Equation (3), and represent activity and charge of the analyte ion, respectively, and and represent activity and charge of the interfering ion, respectively.
The experiments were performed in solutions containing potential interferent (c = 1.0 × 10 −2 M) and PM (its concentration varied between 2.0 × 10 −6 M and 2.0 × 10 −3 M). The potential interferents were chosen due to their presence in composition of pharmaceutical products and body fluids. The calculated , values are presented in Table 4. It can be seen that the new PM sensor is selective for PM, so it could be applicable for PM determination in real samples. Although results suggest that the new PM sensor could be applicable for PM determination in pharmaceutical products and body fluids, in this study, the new sensor was used for PM determination in pharmaceutical products.   Table 4. It can be seen that the new PM sensor is selective for PM, so it could be applicable for PM determination in real samples. Although results suggest that the new PM sensor could be applicable for PM determination in pharmaceutical products and body fluids, in this study, the new sensor was used for PM determination in pharmaceutical products.

Determination of PM
A study on the applicability of the new sensor for PM determination was performed using potentiometric titration and the Gran method [42]. Two commercial drops containing PM as the active pharmaceutical ingredient in a concentration of 20 mg/mL were used as test solutions. The solutions were prepared by dissolving drops in distilled water. For the titration method, NaTPB solution (c = 4.0 × 10 −3 M) was used as the titrant. For the Gran method, increments of known concentration (c = 2.0 × 10 −3 M) of PM were added to the solution eight times. After each addition, the potential was measured. Using (V 0 + V s ) against c s V s , the linear graph was obtained. The negative x-intercept represents the negative amount of PM in the initial sample. Knowing the amount of PM, it was easy to calculate its concentration.
In Equation (4), E 1 and E 0 represent measured potential after each addition and before addition, respectively; V 0 and V s represent the volume of the solution before standard addition and volume of each standard addition, respectively; c x and c s represent PM concentration before standard addition and concentration of each standard addition solution, respectively.
The comparison of the results with the manufacturer's claimed values was used to check the accuracy of PM determination. The results are presented in Table 5. It can be seen that all results are acceptable, and there is no influence of the matrix components on the accuracy of PM determination. However, the results obtained using potentiometric titration are more precise. Additionally, recoveries for PM determination using potentiometric titration are in the range ±5%, while recoveries for PM determination using the Gran method are in the range ±10%. It was expected because the Gran method is the direct potentiometry method.

The Lifetime
The lifetime of the sensor is a time in which the sensor has constant analytical characteristics and can be used for accurate determination of the analyte. In order to determine the lifetime of the new PM sensor, every day before measurements, the calibration in a concentration range between 5.0 × 10 −6 M and 5.0 × 10 −3 M was performed. With daily measurements, the lifetime of the new PM sensor is at least four months, without significant deviations in the performance of the sensor.

Conclusions
The new potentiometric solid-contact sensor, with an ionophore based on functionalized MWCNTs and PM ions, was developed for PM determination. The membrane of the sensor was optimized using five different plasticizers and different content of sensing material (ionophore) in the membrane. The sensor with NPPE and 4% of sensing material was chosen for further characterization due to theoretical calculations and the best analytical performances. It had a Nernstian slope, low LOD (1.5 × 10 −7 M), a measuring range between 6.2 × 10 −7 M and 5.0 × 10 −3 M, good selectivity, and very fast response. The introduction of MWCNTs in the membrane of the sensor resulted in its low signal drift and good stability. The applicability of the new PM sensor was demonstrated using the Gran method and potentiometric titration for accurate PM determination in a pure aqueous solu-tion of PM and pharmaceutical products. Due to the results of the selectivity investigation, the new sensor also has the potential for PM determination in biological fluids.