Electrochemical Behavior and Voltammetric Determination of Two Synthetic Aroyl Amides Opioids

Abstract

In the present work, we describe the electrochemical behaviour of two opioids structurally related to aroyl amides of forensic interest, namely U-47700 and AH-7921. The data obtained allowed for the mise-au-point of a voltammetric determination protocol, that relies on differential pulse voltammetry (DPV) at a glassy carbon electrode in ethanol/0.1 M lithium perchlorate/0.10 M 2,6-lutidine. To apply the method to the analysis of real samples (urines), a clean-up and a preconcentration strategy by solid phase extraction (SPE) using the adsorbent Florisil have been developed. LOQ of 0.2 μg·mL⁻¹ in urine samples with an enrichment factor of 20 and linear range from 5 to 150 μg·mL⁻¹ were obtained.

1. Introduction

Opioids are a class of widely used essential medicines whose abuse could lead to tolerance and dependence, followed by physical and psychological effects, including analgesia, miosis, reduced gastrointestinal mobility, sedation, respiratory depression, anxiety, and, in extreme cases, especially when combined with other substances of abuse, death [1,2,3]. The global-scale diffusion of New Synthetic Opioids (NSOs) on the black market, due to the easy availability of these compounds, is seen with concern by the world community, and is overtaking the diffusion of natural and semi-synthetic opiates (e.g., morphine, heroin, oxycodone), which are more difficult to obtain [4,5].

Among the compounds emerging as substances of abuse, two aroyl amides (AMs, Figure 1), namely 3,4-dichloro-N-(2-(dimethylamino)cyclohexyl)-N-methylbenzamide (also known as U-47700) and 3,4-Dichloro-N-[[1-(N,N-dimethylamino)cyclohexyl] methyl]benzamide (AH-7921), are particularly of concern [3,6]. Such compounds, that bear both a tertiary amine and a 1,2-dichlorobenzamide moiety, have been discovered in the 1970s, later patented but never commercialized as analgesics, thus only few and limited clinical studies are available. However, their abuse is increasing and, considering their potential toxicity, complex multi-drug poisoning has been often observed in emergency rooms [3,6].

From a pharmacokinetic point of view, AMs are subject to phase I and phase II metabolism, but a significant portion of the administered drug can be found unmodified in urine and plasma, in concentrations ranging from 0.15 μg∙mL⁻¹ up to μg∙mL⁻¹ [4,7,8] depending on the dose taken—usually between 5 mg and over 10 mg—and on the consumer’s metabolic profile.

Due to their diffusion, the development of quick analytical methods able to identify these drugs either in seized specimens or in biological matrices is mandatory [7,8]. Recently, different strategies—mainly based on chromatography—have been proposed, including, among others, GC-MS [9], LC-HRMS [10], SPE/LC–ESI-MS-MS [11], LC-QTOF [12], LC-MS/MS [13,14,15,16,17,18,19,20], and UHPLC-MS/MS [21,22] (see

Table 1. Alternative methods for the determination of aroyl amides opioids, with their corresponding figures of merit.

	AH-7920			U-47700
Quantification Methods	LOD ng∙mL–1	Linear Range ng∙mL–1	R2	LOD ng∙mL–1	Linear Range ng∙mL–1	R2
LC-MS/MS * [15]	5	1–2500	0.985	5	1–2500	0.985
LC-MS/MS [16]	-	-	-	1	1–1250	0.9983
LC-MS/MS [17]	0.02	0.01–20	0.9992	0.01	0.01–20	0.9990
UHPLC-MS/MS # [22]	-	-	-	0.6	0–250	0.998
LC-MS/MS [18]	-	-	-	0.5	1–100	0.9995
LC-MS/MS [11]	0.042	0.100−10.0	-	0.019	0.100−10.0	-
LC-MS/MS [19]	-	-	-	0.5	1–1000	0.9997
LC-MS/MS ° [20]	qualitative			qualitative

* Liquid chromatography-tandem mass spectrometry (LC-MS/MS). # Ultra-high-performance liquid chromatography-MS/MS. ° Quadrupole time-of-flight analyzer (LC–QTOF-MS/MS).

). On the other hand, electrochemistry offers several advantages when compared to chromatographic methods, such as low-cost and portability, ease of operation, as well as good sensitivity [23], so that the interest in developing electrochemical methods able to quantify NSOs at low concentration (<50–100 μM) in biological matrices is growing [24,25].

In recent years, our group has focused on the design of smart analytical strategies for the determination in biological and synthetic matrices of substances of abuse, such as synthetic cannabinoids [26], glaucine [27], and Lysergic Acid N, N-Diethylamide [28]; based on our previous proposals, we presented herein the electrochemical characterization by cyclic voltammetry (CV) and exhaustive coulometry of U-47700 and AH-7921. The data obtained have then been exploited for the development of a differential pulse voltammetric (DPV) method to quantify the analytes in capsules as well as in urine samples. In the latter case, a preconcentration/clean-up step was also added, obtaining results comparable to those already reported by previously reported methods [3].

2. Materials and Methods

2.1. Materials and Reagents

An Amel model 4330 module equipped with a standard three-electrode cell (25 mL) was used for the voltammetric techniques (CV, DPV). The experiments were performed using a glassy carbon (0.04 cm² geometrical area) working electrode, a Pt auxiliary electrode, and as a reference electrode, a Ag/AgCl/3 M NaCl. The described electrodes were purchased from BASi Electrochemistry. A 5-point Savitsky-Golay smoothing (included in Amel VA peak 2018 ver 6 software ) was applied when necessary to reduce the noise of the voltammogram.

Exhaustive coulometry, EC, was carried under nitrogen atmosphere using a single-compartment cell (10 mL, see Figure S1 in Supplementary Information) on a BASI PWR-3 power module equipped with three electrodes: a Ag/AgCl/3 Μ NaCl reference electrode, a Pt gauze working electrode, and a Pt wire as the counter electrode.

Standard solutions of the examined compounds used for the CV and DPV experiments were prepared by dissolving them in ethanol (1000 μg∙mL⁻¹) and diluting with the same solvents to obtain less-concentrated standards, as needed. The chemicals employed for the preparation of U-47700 and AH-7921 were commercially available and used as received. Ethanol used as solvent in the electrochemical analyses was ACS grade, purity >99.8% (Carlo Erba, Milan, Italy).

2.2. Synthesis of Aroyl Amides U-47700 and AH-7921

Synthesis of 3,4-dichloro-N-[2-(dimethylammino)ciclohexyl]-N-methyl-benzammide (U-47700). The examined compound was prepared according to a procedure previously described in the literature (see Scheme 1) [29].

Step a. 3,4-dichlorobenzoic acid (1.13 g, 5.92 mmol, 1.5 equiv) was refluxed in thionyl chloride (7 mL) overnight. The excess of SOCl₂ was eliminated under vacuum, and the resulting 3,4-dichlorobenzoyl chloride (I, 1.24 g, >99% yield) was used without further purification for the next step.

Step b. trans-N,N,N’-trimethyl-1,2-ciclohexandiammine (0.7 mL, 4 mmol) and triethylamine (1 mL, 7.2 mmol, 1.8 equiv) were dissolved in 50 mL anhydrous ethyl ether under nitrogen atmosphere, then a solution of 4-dichlorobenzoyl chloride (1.24 g, 5.92 mmol, 1.5 equiv) in anhydrous diethyl ether (10 mL) was added dropwise and stirred for 20 h. The solvent was evaporated under vacuum. The crude residue was purified by column chromatography over silica (hexane: ethyl acetate 7:3) to obtain 1.24 g of U-47700 (white solid, yield: 94%).

U-47700 1H-NMR (300 MHz, CD₃COCD₃) δ 7.69 (d, J = 8 Hz, 1H), 7.51 (s, 1H), 7.31 (d, J = 7 Hz, 1H), 3.37 (m, 1H), 2.85 (s, 3H), 2.64–2.58 (m, 1H), 2.28 (s, 3H), 2.08 (s, 3H), 1.83–1.69 (m, 2H), 1.29–1.19 (m, 2H). ¹³C-NMR (75 MHz, CD₃COCD₃) δ 170.1, 169.2, 140.1, 133.4, 131.9, 130.1, 128.0, 64.3, 55.1, 41.3, 40.8, 32.3, 27.8, 26.4, 23.1, 22.7.

Synthesis of 3,4-Dichloro-N-[[1-(N,N-dimethylamino)cyclohexyl] methyl]benzamide (AH-7921). The examined compound was prepared according to a procedure previously described in the literature (see Scheme 2) [30].

Step a. Synthesis of 1-cyano- 1-ciclohexy-dimethylamine. Methylammonium chloride (8.1 g, 0.120 mol) was dissolved in water (15 mL) and added one-pot to freshly distilled cyclohexanone (9.8 g, 0.099 mol), rapidly followed by the addition of potassium cyanide (6.8 g, 0.105 mol). The resulting solution was stirred at room temperature for 24 h. The obtained solid was filtered, washed with cold water (20 mL), and dissolved in dichloromethane (DCM, 15 mL). The DCM phase was washed with water (15 mL) and the aqueous phase was extracted with DCM. The organic phases were thus reunited, dried with Na₂SO₄, and the solvent evaporated at reduced pression. The resulting 1-cyano-1-ciclohexy-dimethylamine (II, colorless oil, 60% yield [31]) solidified at room temperature after few hours.

II: ¹H NMR (300 MHz, CDCl₃) δ 2.34 (s, 6H), 2.13–2.10 (m, 2H), 1.79–1.75 (m, 2H), 1.60–1.50 (m, 5H), 1.28–1.24 (m, 1H). ¹³C NMR (75 MHz, CDCl₃) δ 118.7, 62.4, 39.7, 34.4, 24.4, 22.2.

Step b. Synthesis of 1-(aminomethyl)- 1-ciclohexy-dimethylamine chlorohydrate. (1-cyano-1-ciclohexy-dimethylamine (2.3 g, 15 mmol) was dissolved in dry ether (20 mL) and added dropwise under stirring to a suspension of LiAlH₄ (1.2 g, 30 mmol) in dry ethyl ether (30 mL) under nitrogen atmosphere. The suspension was stirred at room temperature overnight and the excess of LiAlH₄ was quenched by carefully adding 22.1 mL of 30% w/v NaOH solution followed by a further 5 mL of water. The organic layer was thus separated, dried over Na₂SO₄, and the solvents were evaporated under vacuum. The residue was then dissolved in ether (10 mL) and a HCl solution in ethanol (10% w/v) was added until complete precipitation of crude 1-(aminomethyl)-1-ciclohexy-dimethylamine hydrochloride, which was purified by recrystallization from ethanol/ethyl ether (III, yield: 65%, white needles) [32].

III: ¹H NMR (300 MHz, CDCl₃) δ 2.65 (s, 2H), 2.17 (s, 6H), 1.19–1.52 (m, 10H). ¹³C NMR (75 MHz, CDCl₃) δ 77.4, 57.9, 42.5, 37.5, 28.6, 26.2, 22.1.

Step c. Synthesis of 3,4-Dicloro-N-[[1-(dimetilammino)cicloesil]metil]benzammide (AH-7921). 1-(aminomethyl)-1-ciclohexy-dimethylamine (250 mg, 1.6 mmol) and 3,4-diclorobenzoyl chloride (I, freshly prepared from 460 mg of 3,4-dichlorobenzoic acid, 2.4 mmol, and 3 mL of thionyl chloride as described for the synthesis of U-47700) were dissolved in 2 mL pyridine and kept under stirring at room temperature for 1 h. The crude AH-7921 was isolated, then recrystallized from ethanol/ethyl ether (colorless solid, 40% yield) [31].

AH-7921 ¹H NMR (300 MHz, CDCl₃) δ 11.86 (bs, 1H), 8.96 (s, 1H), 8.31 (s, 1H), 8.13–8.16 (d, 1H, J = 9 Hz), 4.00–4.02 (d, 2H, J = 6 Hz), 2.78 (s, 6H), 1.95–1.98 (m, 2H), 1.78–1.81 (m, 3H), 1.59.1.63 (m, 4H), 1.59–1.60 (m, 1H). ¹³C NMR (75 MHz, CDCl₃) δ 165.9, 136.1, 132.8, 132.7, 130.5, 130.4, 127.1, 77.1, 68.3, 38.5, 37.4, 27.5, 24.5, 21.9.

2.3. Glassy Carbon (GC) Electrode Pre-Treatment/Characterization

The GC electrode was mechanically cleaned using successively finer grades of alumina (from 1 μm to 0.05 μm), rinsed with HNO₃ 5%_aq, and ultimately washed with water. The electrode was further electrochemically cleaned performing 15 CV cycles CV, in a 0.5 M H₂SO₄ solution, in the potential range E_i = 0.0 mV to E_f = +1400 mV, using a fixed scan speed (ν) of 200 mV∙s⁻¹ [28,33]. The effective area of the electrode was measured by CV in a water solution containing 5.0 mM potassium hexacyanoferrate (II) and 0.1 M potassium nitrate as supporting electrolyte. The scans were performed in the potential range E_i = +200 mV to E_f = +800 mV, with a scan speed of 100 mV∙s⁻¹ applying the Randles–Sevcik equation. [34].

2.4. Electrochemical Analyses

2.4.1. Number of Electrons Involved in the Electrochemical Processes

The number of electrons involved in the redox reactions was determined by exhaustive coulometry (EC, see Table S1 in Supplementary Information) [35,36]. According to the literature, the electrolysis was carried out on an ethanolic 2.5 mM solution of the target compound (10 mL) containing lithium perchlorate (0.1 M) and 2,6-lutidine (0.10 M); the experiment was stopped when the current was <5% of the initial value.

The potential to be applied was the one of the first voltammetric wave observed by CV (E_p values reported in

Table 2. Peak potential and electrochemical behavior of the studied compounds evaluated by CV (2.5 in ethanol, containing 0.1 M lithium perchlorate and 0.10 M 2,6-lutidine, measured at scan speed 50 mV·s⁻¹). Note: i = irreversible. D = diffusion coefficient. K° and α refer to the process described by E_p1 (path a in Scheme 3).

Compound	D cm2∙s−1	E0′ mV	Ep1 mV	Ep2 mV	Peak 1 Intensity μA	Peak 2 Intensity μA	ln K°	α	Reaction Order
U-47700	1.40 × 10−5	+980 (i)	+980	+1300	41 ± 5	56 ± 7	−16.8	0.73	1
AH-7921	2.40 × 10−5	+1100 (i)	+1100	+1300	30 ± 6	31 ± 5	−15.4	0.70	1

) increased by +100 mV, to speed up the electrolysis and to compensate for the reduction in the concentration of the electroactive species during the experiment, which affects the potential needed as per the Nernst equation.

As the electrode used in the EC was a Pt gauze, we performed a CV experiment on a working Pt electrode. The potential measured and the waveform were comparable with the ones observed in the same conditions by GC, with a discrepancy <20 mV on the peak potentials.

2.4.2. Cyclic Voltammetry Measurements

These experiments, performed to understand the redox activity of the considered compounds, were performed in ethanolic solution of the analyte (2.5 mM), containing 0.1 M lithium perchlorate as supporting electrolyte and an excess (0.10 M) of a base (2,6-lutidine), that proved to improve the CV signals by favoring the deprotonation of the intermediates (see Scheme 3). A 0 V to +1.8 V and 0 V to +2.2 V range for U-47700 and AH-7921, respectively, was chosen. Different scan speeds were tested, from 10 mV∙s⁻¹ up to 1000 mV∙s⁻¹.

2.5. Electrochemical Determination of Aroyl Amides

A procedure based on DPV was developed for the quantification of the examined analytes. In a screening phase, a set of different solvents (acetonitrile, ethanol, N,N-dimethylformamide water) containing different supporting electrolytes, i.e., LiClO₄, KNO₃, and (nBu)₄ClO₄, according to the reciprocal solubility, was investigated; likewise, the effect of the presence of additives including water in non-aqueous solvents and hexafluoroisopropanol has been considered. In view of the results obtained, as a medium for the analysis a solution of LiClO₄ (0.10 M) and 2,6-lutidine (0.10 M) in ethanol was selected. The organic base guarantees the exclusive presence of the analytes in their deprotonated form during the analytical procedure.

The DPV instrumental parameters were as follows: wave frequency 12.5 Hz, wave amplitude 50 mV, wave period 80 ms, wave increment 5 mV, E_i = +500 mV, and E_f = +1400 mV. Before analyses, the solutions were purged with a stream of nitrogen for 3 min. The standard addition method was used.

2.6. Evaluation of LOD, LOQ, and Statistical Parameters

LOD and LOQ were calculated from the calibration curves according to ICH Topic [35].

$$\mathrm{L}\mathrm{O}\mathrm{D}=3.3\frac{Sy/x}{m} \mathrm{LOQ}=10\frac{Sy/x}{m}$$

where Sy/x is the residual standard deviation from the linear regression and m is the slope of the calibration curve.

The standard deviation is reported after the numerical value.

Interday and intraday precisions were calculated according to the same guidelines for LOD and LOQ. [37]

2.7. Analysis of Aroyl Amides Capsules

Capsules containing the considered drugs were prepared by mixing thoroughly in a porcelain mortar 10 mg of each active ingredient [3,37], 30 mg of calcium carbonate, and 70 mg of lactose, and grounding for 2–3 min to obtain a homogeneous solid suspension. For the analysis, the content of 2 capsules was placed in a beaker and was extracted with 10 mL ethanol by sonicating for 3–5 min The resulting suspension was filtered on a 2 cm diameter 0.22 μm syringe nylon filter (Whatman^®, purchased from Merck KGaA, Darmstadt, Germany). An aliquot of the filtrate—typically 150 μL—was diluted with 10 mL ethanol containing the supporting electrolyte and 2,6-lutidine in the concentration already described and analyzed by DPV by standard addition method (see above).

2.8. Analysis of Aroyl Amides in Water Samples, Synthetic and Natural Urines

Water samples were prepared by spiking distilled water with the proper amounts of the drugs. Synthetic urine was prepared as previously described [27,34,38], and its composition is reported in Supporting Information (SI Section S5). To analyze water sample and urine, a clean-up/preconcentration step was needed before DPV analyses. We used the solid phase extraction (SPE, Richardson, TX, USA) technique, employing home-made cartridges assembled with 0.5 g Florisil set on a 3 mL polypropylene tube, and interposed between two polyethylene frits of 20 μm porosity (Sigma-Aldrich, St. Louis, MO, USA). As a reservoir to dispense the water/urine solution to the cartridges, a polypropylene syringe (50 mL) was used, connected to the SPE cartridge by a proper adaptor (Aldrich, code 57020-U).

The assembled cartridges were preconditioned by washing with ethanol (10 mL) and then water (5 mL) before use. The water or urine samples, whose volume ranged from 10 mL to 200 mL, according to the specific experiment, were spiked with the considered drugs at concentrations in the range 0.5 μg·mL⁻¹ to 20 μg·mL⁻¹. The solutions were then brought to pH = 6 by 10% w/v HNO₃. At this point, to remove any possible solid precipitation that could clog the SPE column, the solutions were filtered toward a 2 cm diameter 0.45 μm nylon syringe filter. The clear solution was charged onto the SPE column, previously washed with 50 mL water at a flow rate of 1 mL·min⁻¹, by using a commercially available manifold (Visiprep™ SPE manifold, Sigma-Aldrich, St. Louis, MO, USA) connected to vacuum, with a flux of 1.0 mL∙min⁻¹. The drugs were eluted with 10 mL of a 0.1 M LiClO₄ solution in ethanol (flow rate: 1 mL·min⁻¹) and after the addition of 2,6-lutidine (0.10 M), they were analyzed by DPV as described. Recovery experiments were performed in triplicate.

3. Results and Discussion

3.1. Electrode Characterization

The GC electrode’s electrochemically active area resulted to be 0.051(3) cm², which is in good accordance with the geometrical area and proof that the surface has low roughness.

3.2. Electrochemistry of the Examined Aroyl Amides

The number of electrons involved in the redox process was calculated from the charge Q consumed in the EC experiment. It needed 35 min at +1100 mV (for U-47700) and 65 min at +1200 mV (for AH-7921) to completely electrolyze the solution, consuming, respectively, 4.3 coulomb (U-47700) and 5.6 coulomb (AH-7921). Considering that we used 5 × 10⁻⁵ moles of substrate, we calculated that a one-electron process was involved (see Table S1). Obviously, the number of electrons per mole involved in the rate-determining step must coincide with this value (see in Figures S2 and S3 the comparison between the CV of the solutions before after the EC experiments).

3.2.1. Cyclic Voltammetry and Oxidation Mechanism

Among the various combinations of supporting electrolytes and solvents tested, the best results in terms of waveform and peak intensity were obtained using a polar protic solvent (ethanol) and a supporting electrolyte which combines a small cation (Li⁺) and an intermediate size anion (ClO₄^-) [35,36]. In fact, by using an aprotic solvent (such as acetonitrile or DMF), significant fouling is found from scan to scan, and the electrochemical signal is highly irreproducible (i.e., a change >20% in successive scans is found, see Figure S4), while when using tetrabutylammonium perchlorate as a supporting electrolyte higher background signals are obtained, as can be seen from Figure S4 in which results for AH-7921 are reported (the same behavior was observed with U-7700).

The results obtained with the two opioids at scan speeds from 10 mV∙s⁻¹ up to 1000 mV∙s⁻¹. (see Figures S5 and S6) are comparable to each other and are described below.

The CV curves showed a two-peak irreversible oxidation wave in the range 0 mV to +2000 mV for scan speed 50 mV∙s⁻¹ (see

Table 3. Curves used for the determination of the electrochemical parameters exposed in the text. C = concentration (μg∙mL⁻¹); v = scan speed (V·s⁻¹); i= current intensity (A). The standard deviation is reported after the numerical value.

	U-47700
E vs. log v	E/V = (0.091 ± 0.013)·log (v/V·s−1) + (1.144 ± 0.047)
log i vs. log v	log (i/A) = (0.504 ± 0.010)·log (v/V·s−1) − (4.103 ± 0.032)
i vs. v1/2	i/A = (1.003 ± 0.021) × 10−4 v1/2/V1/2·s−1/2 + (3.004 ± 0.032)∙10−6
log i vs. log C	log (i/A) = (3.153 ± 0.032)·log (C/mg∙L−1) + (1.753 ± 0.019)
i·v−1/2 vs. v1/2	i·v−1/2/A·V−1/2·s1/2 = −(1.004 ± 0.042) × 10−5/V1/2·s−1/2 + (1.42 ± 0.28) × 10−4
	AH-7921
E vs. log v	E/V = (0.093 ± 0.014)·log (v/V·s−1) + (1.080 ± 0.032)
log i vs. log v	log (i/A) = (0.531 ± 0.023)·log (v/V·s−1) − (3.941 ± 0.043)
i vs. v1/2	i/A = (7.003 ± 0.041) × 10−5 v1/2/V1/2 ·s−1/2 + (4.002 ± 0.053)∙10−4
log i vs. log C	log (i/A) = (0.972 ± 0.031)·log (C/mg∙L−1) + (1.821 ± 0.040)
i·v−1/2 vs. v1/2	i·v−1/2/A·V−1/2·s1/2 = −(1.002 ± 0.050) × 10−5/V1/2·s−1/2 +(9.001 ± 0.033) × 10−5

for details and Figure 2), with the first oxidation wave being well defined and the second one only poorly defined and with no re-reduction peaks. Results reported below relate to the first oxidation peak, unless otherwise specified.

The E_p vs. log ν graph was linear, thus suggesting that an electrochemical step (E) is followed by an irreversible chemical step (C) (see Table 3 and Figure 3) [22,28,39,40,41,42]. This assumption is confirmed by the “current function”, as described in the “Materials and Methods” section, which is flat, with a slope < 10⁻⁵ V·s⁻¹ (see Table 2 and Figure 4 and Figure 5) [40]. Concerning AH-7921, the slope of the current function changes drastically for low scan speeds, meaning that in these conditions the rate of the chemical step is competitive with respect to the scan speed [43,44].

The redox process is diffusion-controlled and not limited by adsorption of the electroactive species or its products, as confirmed by the linearity of the log i_p vs. log ν plot, that has a slope around 0.5 units/decade (See Figure 4) [27,29].

To clarify the EC mechanism, we performed CVs on model redox compounds related to the two opioids considered. First, we proved that amides (DMF) and an ester related to the target analytes (methyl 3,4-dichlorobenzoate) are not redox-active in the explored potential range (see Figures S2, S3 and S5 in Supplementary Information), leading us to the conclusion that the monoelectronic oxidation involves the tertiary aminic nitrogen (first wave in CV). To further confirm this, we evidenced that the CV of the two precursor amines 1-(N-aminomethyl)-1-ciclohexy-dimethylamine and trans-N,N,N’-trimethyl-1,2-ciclohexandiammine showed a first peak congruous with those observed with AH-7921 and U-7700, respectively, that can be unambiguously assigned to the tertiary amine moiety [43,44,45,46]. Along with these two compounds, further oxidation peaks were observed and can be attributed to the oxidation of the secondary amine moiety and to the further oxidation of the degradation products formed after the electron transfer [44].

The presence of a base (2,6-lutidine) had a significant effect on the oxidation peak definition and intensity of the two opioids but not on their position, as expected for an E step followed by the loss of a proton [47,48].

The described results along with literature data [24,45] lead us to propose the redox mechanism reported in Scheme 3 for U-47700. According to this view, the one-electron oxidation of the aminic N of the analytes (E_p1, path a) leads to the formation of the radical cation intermediate I^•+, followed by a chemical step, identified as a deprotonation (path b) to form the α-amino radical II^•. The further oxidation of this intermediate [49] (E_p2, path c) afforded iminium ion III⁺ and accounted for the second voltammetric wave observed in the voltammogram. Once formed, III⁺ can undergo different competitive degradation pathways, among which are the addition of the nucleophilic solvent [50] and hydrolytic processes where the formation of an aldehyde prevails [43]. Analogues processes are involved with AH-7921.

3.2.2. Cyclic Voltammetry Plots

Plots of the CV data—namely, peaks potential (E_p) vs. log ν (see Figure 3) [51,52], log i_p vs. log ν (Figure 4) [53,54]—were done to further understand the redox mechanism. Moreover, to assess if the electrochemical processes were limited by adsorption or were diffusion-controlled, current intensity (i_p) was plotted vs. the square root of the scan speed (ν^1/2) (see Figure 5) and vs. scan speed (ν), Finally we plotted log i vs. log C (μg∙mL⁻¹) (Figure 6) and i_p∙ν^−1/2 vs. ν (Figure 7).

3.2.3. Determination of $E{°}^{’}$

As the electrochemical processes involving the studied compound were proved to be irreversible in the described conditions, an accurate determination of the formal potential is not possible, nevertheless it can be roughly estimated by using the E_p vs. ν graph, as $E°’$ corresponds to the value of E_p extrapolated on the vertical axes at ν = 0 [27,50]. Estimated $E{°}^{’}$ values are reported in Table 2. Values found are around +1000 mV, in agreement with the oxidation potential of tertiary amines [44].

3.2.4. Determination of the Charge Transfer Coefficient (α)

As is already known [39], α can be estimated via the Tafel slope, indicated as b in Equation (1), and obtained by the already described CV experiments from the E_p vs. log v plot.

b = ∂E_p/∂logν = 2.303RT/αn_αF

(1)

By substituting the values of T, R, and F, and taking into consideration that from literature data and our coulometric experiments the number of electrons involved in the rate determining step (n_α) is equal to 1 [35,36], Equation (2) can be obtained from which α can be directly calculated. The calculated values of α for the two opioids are comparable and around 0.7, as reported in Table 2. These values are consistent with irreversible redox processes in which the curves describing the activation energy barrier are not symmetric [55,56].

α = 0.059/b

(2)

3.2.5. Determination of the Diffusion Coefficient (D)

The already described CV curves can be used to evaluate D (cm²∙s⁻¹) from the slope of the graph i_p vs. v^1/2 applying the modified Randles–Sevcik equation for irreversible processes to the first oxidation peak, according to Equation (3) [55,56]:

$$i_p=2.99·{10}^{5}n{\left(1-\alpha \right)}^{1/2}A{\left[C\right]}_{\mathrm{\infty }}{D}^{1/2}{v}^{1/2}$$

(3)

where A is the electrode area (cm²), n is the number of electrons involved in the redox process (n = 1), i_p is the peak current (Ampere), [34] α is the charge transfer coefficient described above, [C] is the concentration of analyte (chosen as 5 × 10⁻⁶ mol∙cm⁻³ in the present case), and ν is the scan speed (V∙s⁻¹).

In our experiments, we found diffusion coefficients in the range 10⁻⁵ cm²∙s⁻¹ and 10⁻⁶ cm²∙s⁻¹ (see Table 2), in line with literature data for compounds of similar size and polarity [34].

3.2.6. Determination of the Kinetic Constant K°

Matsuda and Ayabe’s equation (Equation (4)) can be used to calculate the value of K° when irreversibility prevails [39,50]:

$$E_{pirrev}=\frac{RT}{a{n}_{\alpha }F}\left[0.78-ln\frac{K°f_{red}}{D^{\frac{1}{2}}}+\frac{1}{2}ln\frac{\alpha n_{\alpha }Fv}{RT}\right]$$

(4)

where $f_{red}$ is the activity coefficient that can be approximated to 1, and the other symbols have the usual meanings reported above. Peak potential (E_p) is expressed in V.

3.2.7. Determination of the Reaction Order

Differential pulse voltammetry (DPV) was used to evaluate the reaction order, using the experimental conditions described in Section 2.5 regarding the analytical determination of the drugs. The curve log i_p vs. analyte concentration [26,35,36], in the 5 μg∙mL⁻¹ to 150 μg∙mL⁻¹ concentration range (eight data points), was constructed, and its slope was used for the evaluation of the reaction order.

The reaction order for the two opioids was one (see Table S1 in Supplementary Information). These data helped prove the reaction mechanism (Section 3.2.1) and showed that no dimerization of the intermediates occurred.

3.3. Electrochemical Determination of Aroyl Amides

When shifting from CV to DPV to develop a quantitative analytical method, we found that in the considered oxidation potential range the two opioids showed two peaks, the first one of both analytes was used for the quantitative determination. Concerning U-7700, the second peak, it appeared only at higher concentrations of the analyte (>100 μg∙mL⁻¹).

Linearity was found with concentrations of the opioids ranging from 5 μg∙mL⁻¹ up to 150 μg∙mL⁻¹ (see Figure 8, Figure 9 and Figure 10 for the calibration curves). Details concerning the parameters associated with the DPV method are presented in

Table 4. Calibration curves (DPV) for the considered compounds in ethanol/lithium perchlorate 0.1 M, and associated parameters. Standard deviation is reported after the numerical value. Each curve was obtained with 9 data points, from 5 μg∙mL⁻¹ to 100 μg∙mL⁻¹. Peak potentials observed in DPV are also reported.

Compound	Calibration Curve	Correlation Coefficient R2	LOD μg∙mL−1	LOQ μg∙mL−1	Intraday Precision %	Interday Precision %	Ep1 mV	Ep2 mV
U-47700	i/μA = (0.01212 ± 0.00043)·(C/μg·L−1) + (0.012 ± 0.020)	0.993	0.2	0.6	7	6	+880	+1150
AH-7921	i/μA = (0.01533 ± 0.00083)·(C/μg·L−1) + (0.083 ± 0.02)	0.990	0.2	0.5	9	8	+970	+1150

. LOD and LOQ, calculated from 10 data points calibration curves, were 0.2 μg∙mL⁻¹ (LOD) for both opioids and 0.5 μg∙mL⁻¹ for AH-7921 and 0.6 μg∙mL⁻¹ for U-47700 (LOQ). The voltammograms for the two compounds are reported in Figures S8 and S9. The considered redox process is irreversible and characterized by the exchange of a proton. These phenomena are responsible for a change in the peak potential by changing the concentration of the substrate (visible in the calibration curves reported in Figure 8, Figure 9, Figure S8 and Figure S9). We think that the reason for why the phenomena are much more evident with U-47700 with respect to AH-7921 resides in the different kinetics of the redox process (as evidenced by the values reported in Table 2 and by the fact that the ΔE between the first and second peak is 200 mV in one case and 320 mV in the other).

As the voltammetric peaks of the two compounds have barely different potentials, we explored the possibility of determining one in the presence of the other. As can be seen from the voltammograms of Figures S8 and S9, this is feasible if the two compounds are both present at similar concentrations; on the contrary, when one is much greater than the other (i.e., 10×), a single broadened peak is obtained. As can be seen from Figure S9, increasing the concentration of AH-7921, the peaks tend to approach and overlap, due to the fact that their ΔE_p is <100 mV (see Table 4) and to the fact that they present a second peak that has the same peak potential (+1150 mV, see Table 4). For these reasons, the peaks overlap, giving rise to a broadened wave; nevertheless, as we measure the height of the peak and not their area, the errors are minimal if the concentration of the two drugs is in the same order of magnitude as reported in the text. Nevertheless, the co-administration of these two equipotent opioids is a rare event, and even in this case their dose in formulations is in the same order of magnitude.

3.3.1. Analysis of the Drug Formulations in Presence of Possible Interfering Substances

The excipients and drugs that can commonly co-present in the formulations of the two opioids were investigated as potential interferences in the DPV method, by adding them directly to the solution in the voltammetric cell in which a fixed concentration of the active compound is present. Several sugars were considered, e.g., glucose, dextrose, galactose, mannitol, and lactose (up to 5000 μg∙mL⁻¹). We also considered as possible excipients EDTA, amphetamine (up to 50 μg∙mL⁻¹), and dextrin (up to 70 μg∙mL⁻¹), without any significant interference (change in peak height and position <10%). The higher tested concentrations of the potentially interfering substances were chosen according to their solubility in the working medium. Water, up to 20% with respect to ethanol, did not interfere. DPV scans in presence of the various interfering substances are reported in the Supplementary Information, Figures S10 and S11.

3.3.2. Analysis of the Capsules

The capsules were prepared and extracted as described in the “Material and method” section. Recovery factors of 86–99% for U-47700 and 83–99% for AH-7921 were obtained (see

Table 5. Percentual recovery factor (Rc %) in water, synthetic urine, and real urine samples (number of replicates, n = 4) spiked with U-47700 and AH-7921 and in capsules containing 10 mg of one of the two compounds.

	Water		Synthetic Urine		Urine		Capsules
Compound	Concentration μg∙mL−1 a	Rc %	Concentration μg∙mL−1 a	Rc %	Concentration μg∙mL−1 a	Rc %	Amount mg	Rc %
U-47700	0.5 (100)	111	0.5 (100)	102	0.5 (100)	73	10	86–99
	0.5 (200)	111	0.5 (100)	102	0.5 (200)	97
	1 (50)	115	1 (50)	101	2 (50)	86
	20 (10)	87	2 (50)	105	5 (50)	95
AH-7921	0.5 (100)	98	0.5 (100)	76	0.5 (100)	104	10	83–99
	0.5 (200)	115	0.5 (200)	89	0.5 (200)	96
	1 (50)	100	1 (50)	85	2 (50)	103
	20 (10)	110	2 (50)	81	5 (50)	106

^a the volume of solution preconcentrated is reported between parentheses.

).

3.3.3. Analysis of Water and Natural and Synthetic Urines Spiked with the Title Compounds

As expected, the concentrations of the unmetabolized drugs in urine samples from real cases are in the order of 0.5 μg∙mL⁻¹ to 5 μg∙mL⁻¹ [57,58], i.e., in some cases below the LOQ of the method, and due to severe interferences caused by this complex matrix of partially unpredictable composition, a preconcentration and clean-up step is mandatory before the DPV analysis.

We already found that Florisil is a highly attractive stationary phase for the SPE of aminic compounds, as it efficiently retains this class of substances in a pH-dependent way [27,34]. For this reason, we used this approach for the clean-up/preconcentration of the two opioids, as described in the Materials and Methods section. The recovery factors were between 87% and 115% (water), 76% and 105% (synthetic urine), and 75% and 106% (real urine), depending on the analyte concentration (see Table 5).

The preconcentration factor obtained in our experimental conditions with the SPE phase was up to twenty times. Concentrations <0.5 μg∙mL⁻¹were not considered because of poor clinical significance, and urine volumes >200 mL were not routinely used due to the long percolation time on the SPE cartridge.

4. Conclusions

The characterization of the electrochemical oxidative behavior of two synthetic opioids of the category of aroyl amides highlighted that the electroactivity of the compounds is related to the presence of the aliphatic tertiary amine group. Upon oxidation, a radical cation is formed whose reactivity is responsible for a branched reaction path. A DPV method for the analysis of the two compounds in water, synthetic urine, and urine samples was developed by relying on their oxidative behavior. For both synthetic opioids, at a concentration range suitable for its application in clinical and forensic analyses, possible interference from common substances present in these samples has been verified and excluded. Solid-phase extraction clean-up and preconcentration allowed us to reach the quantification limits compatible with the amounts of unmetabolized compounds that can be found in urine. The DPV method here described is timesaving and has analytical performances comparable to the methods already published in the literature. The procedure has also been applied to the analysis of capsules containing 10 mg of the aroyl amides, with no significant interferences due to the excipients.