A direct and selective electrochemical hydrogen sulfide sensor

Abstract

Continuous, in situ detection of hydrogen sulfide (H₂S) in biological milieu is made possible with electrochemical methods, but direct amperometry is constrained by the generation of elemental sulfur as an oxidative byproduct. Deposition of a sulfur layer passivates the working electrode, reducing sensitivity and causing performance variability. Herein, we report on the use of a surface preconditioning procedure to deposit elemental sulfur on a glassy carbon electrode (GCE) prior to measurement and evaluate performance with common analytical metrics. The lack of traditional anti-poisoning techniques (e.g. redox mediators, cleaning pulses) also allowed for facile surface modification with electropolymerized films. For the first time, a series of electropolymerized films were characterized for their H₂S permselective behavior against common biological interferents. Highly selective, film-modified electrodes were then evaluated for their anti-biofouling ability in simulated wound fluid. The final optimized electrode was capable of measuring H₂S with a low detection limit (i.e., <100 nM) and ~80% of its initial sensitivity in proteinaceous media.

Graphical Abstract

1. Introduction

Hydrogen sulfide (H₂S) was for a century only known as a noxious environmental pollutant—a product of decaying matter with a distinct smell of rotten eggs [1–3]. Only recently (i.e., the last few decades) was H₂S adopted into the family of endogenous gasotransmitters that regulate essential processes of mammalian physiology [4, 5]. Nitric oxide (NO) and carbon monoxide (CO)—earlier members of this collection of gases—were also deemed poisonous gases at one point [6–8]. As the most recent addition, much attention around H₂S has been placed on interpreting its specific biological activity and related concentration-dependence.

Endogenous H₂S is either biosynthesized through enzymatic metabolism of cysteine or drawn from acid-labile/bound sulfide reservoirs [9, 10]. It is now established that H₂S plays key roles in the cardiovascular, endocrine, gastrointestinal, and central nervous systems, as it has been identified in tissues associated with each function [11, 12]. Like NO and CO, H₂S has been reported to modulate neuronal transmission [13, 14], relax smooth muscle [15], heighten/suppress the inflammatory response [7], and exert cardioprotective effects [16]. As a result, hydrogen sulfide-related research has grown to encompass therapeutic donors hoping to harness these effects [16–22]. Much investigation has also centered on the clinical utility of H₂S as a serum biomarker of cardiovascular disease, such as artherosclerosis [23], hypertension [24], coronary heart disease [25], and chronic obstructive pulmonary disease [12, 26], Given the breadth of its activity (either defined or ambiguous), the ability to accurately detect and quantify H₂S in biological milieu is essential to further understanding the pharmacology and physiology of this pervasive molecule.

An ideal detection platform would be capable of sensing biologically relevant H₂S concentrations (100 nM – 10 μM) in situ, continuously, and with minimal sample preparation [27–29]. Traditionally, iodometric titration and methylene blue colorimetric assays have been the most common means of quantifying H₂S [30, 31]. While relatively simple to execute, extensive sample preparation, high LODs (~1 μM), and potential sample damage (e.g., to cells, tissue) prevent the use of these assays for most biological applications [32]. Novel, but less conventional methods for H₂S measurement include chromatographic techniques, such as high pressure liquid chromatography (HPLC) and gas chromatography (GC) [9, 33]. However, column loading is not amenable to real-time measurement, and hypoxic/anoxic conditions are required to prevent conversion to sulfur oxides. Several fluorescent [34, 35] and electrochemiluminescent [36, 37] probes have been developed that are capable of highly sensitive and selective detection, but long incubation times and irreversible coordination of H₂S forgo the possibility of dynamic detection [12].

In addition to their simple and low-cost fabrication, electrochemical platforms allow for highly sensitive, in situ detection without the need for exogenous reagents [12]. Ion sensitive electrodes (ISEs) are able to measure sulfide ions (S^2-) with near-perfect selectivity and low LODs (~100 nM); ISEs have thus been used extensively for serum measurements [38, 39]. However, samples must undergo alkaline pre-treatment to shift proton dissociation equilibria from H₂S towards S^2- (pK_a1 = 6.6, pK_a2 = 13.8), draining acid-labile sulfide pools and yielding concentrations representative of the total sulfide content instead of the freely available H₂S [2]. Alternatively, direct amperometric detection of H₂S and hydrosulfide (HS⁻) is possible via a two-electron oxidation with elemental sulfur as a byproduct (Eq. 1 and 2, respectively) [40].

$${\text{H}}_2\text{S}\to {\text{S}}^0+2{\text{e}}^{-}+2{\text{H}}^{+}$$ 1

$${\text{HS}}^{-}\to {\text{S}}^0+2{\text{e}}^{-}+{\text{H}}^{+}$$ 2

Of note, growth of an insulating sulfur layer passivates the electrode surface, causing sensitivity reduction and performance variability [41–45]. Clark-type amperometric sensors have attempted to resolve this problem by incorporating a gas-permeable membrane and alkaline internal solution with a redox mediator (e.g., ferrocyanide) [46, 47]. The redox mediator accepts electrons from H₂S/HS⁻ and is regenerated at the working electrode to create a measurable current (as low as 10 nM LOD) [11]. Even with redox mediation, elemental sulfur is still produced, poisoning the internal solution and resulting in high background currents. Typical problems associated with internal solutions (i.e., leakage, evaporation, replacement, and poor miniaturizability) also remain as a concern [12].

Herein, we report the development of a direct amperometric sensor based on the oxidation of H₂S/HS⁻ at a glassy carbon electrode (GCE). Previous work in our laboratory demonstrated the utility of a high potential cleaning pulse (+1.5 V) on a GCE to convert elemental sulfur to water-soluble sulfate, thereby mitigating surface passivation [48]. The use of such high potentials produced significant residual background currents, alter the chemical structure of the GCE surface, and potentially cause oxidative damage to any modifying films [49]. As a result, we have chosen to investigate the use of a surface conditioning procedure to pre-poison the electrode surface with sulfur and stabilize sensor performance. To improve selectivity for H₂S over common biological interferents, we tested several common electropolymerized films previously used in the context of hydrogen peroxide (from glucose oxidase) [50–52] and NO [53, 54] sensing. Polymers produced from the following monomers were studied: 5amino-1-napthol, phenol, eugenol, and ortho-, meta-, and para-phenylenediamine. After extensive selectivity testing, film-modified electrodes were evaluated for anti-biofouling ability and long-term stability under continuous use in simulated wound fluid.

2. Experimental

2.1. Materials, reagents, and apparatus

5-Amino-1-naphthol (5A1N), phenol, eugenol, ortho-phenylenediamine (o-PD), meta-phenylenediamine (m-PD), para-phenylenediamine (p-PD), sodium sulfide nonahydrate (Na₂S·9H₂O), hydrochloric acid (HCl), sodium nitrite, L-ascorbic acid, acetaminophen, dopamine hydrochloride, L-cysteine, ethylenediaminetetraacetic acid disodium salt dihydrate (EDTA), and fetal bovine serum (FBS), were acquired from Sigma-Aldrich (St. Louis, MO). Sodium hydroxide (NaOH) and hydrogen peroxide (H₂O₂; 30 wt%) were obtained from Fisher Scientific (Hampton, NH). Nitric oxide (99.5%), carbon monoxide (99.3%), and nitrogen (99.998%; N₂) gases were obtained from Airgas National Welders (Durham, NC). Other solvents and chemicals were analytical-reagent grade and used as received without further purification.

All solutions were prepared with distilled water purified to a resistivity of 18.2 MΩ cm and a total organic content of ≤6 ppb with a Millipore Milli-Q UV Gradient A10 water purification system (Bedford, MA). Saturated solutions of NO (1.9 mM) and CO (0.9 mM) were prepared by purging ~20 mL of phosphate buffered saline (PBS; 10 mM, pH 7.4) on ice with N₂ for 25 min to remove oxygen, followed by purging with NO or CO gas for an additional 25 min [55]. Saturated gaseous solutions were stored at 4 °C and used the same day as preparation. Fresh hydrogen sulfide (H₂S; 3.0 mM) stock was prepared the day of use by dissolving Na₂S·9H₂O in deoxygenated PBS with 150 μM EDTA, sealing the container with a rubber septum, and purging the headspace with N₂. Stock concentration was determined by iodometric titration [30].

Electrochemical experiments were carried out on a CH Instruments 1030 8channel Electrochemical Analyzer (Austin, TX). The multi-electrode configuration consisted of 8 glassy carbon inlaid disc working electrodes (GCE; 3.0 mm dia.) sealed in Kel-F (6.0 mm total dia.; CH Instruments), a silver-silver chloride (Ag|AgCl) reference electrode (3.0 M KCl; CH Instruments), and a platinum (Pt) wire counter electrode. All potentials cited are versus the Ag|AgCl reference electrode. Unless otherwise specified, electrochemical measurements were collected in 20 mL deoxygenated PBS (pH 7.4) at room temperature (23 °C).

2.2. Preparation of electropolymerized film-modified electrodes

Glassy carbon working electrodes were polished consecutively with 1.0, 0.3, and 0.05 μm particle size deagglomerated alumina slurries (Buehler; Lake Bluff, IL) dispersed on a microcloth. Electrodes were then copiously rinsed with water and wiped with tissue to remove residual alumina. To ensure a stable background, electrodes were cycled in PBS between 0 and +1.0 V using cyclic voltammetry (CV) eight times at a scan rate of 10 mV s⁻¹ before being transferred to monomer solutions for electrodeposition. All monomers were dissolved at a concentration of 10 mM in PBS. To improve solubility, solutions of 5A1N were titrated to pH 1.0 with HCl, and eugenol to pH 13 with NaOH. Electropolymerized film depositions were carried out using CV, sweeping the potential between 0 and +1.0 V at a scan rate of 10 mV s⁻¹ for a total of 20 cycles. Film-modified electrodes were then rinsed with water to remove unbound oligomer from the surface.

2.3. Voltammetric measurements

The potential was scanned from −0.3 to +1.1 V in PBS eight times via differential pulse voltammetry (DPV) on bare glassy carbon electrodes to achieve a constant background. Differential pulse voltammograms were then collected in solutions of common, electroactive biological interferents by sweeping over the same potential window. The anodic peak potential(s) (E_a) of the following interferents were measured: L-ascorbate (AA; 1.0 mM), cysteine (CYS; 1.0 mM), dopamine (DA; 1.0 mM), acetaminophen (AP; 1.0 mM), hydrogen peroxide (H₂O₂; 1.0 mM), nitrite (NO₂⁻; 1.0 mM), and nitric oxide (NO; 95 μM). Of note, voltammograms of carbon monoxide (CO) were indistinguishable from background at its saturation limit (0.9 mM) with no measured E_a value. All DPV traces were collected using the following parameters: 50 mV amplitude, 4 mV step increase, 0.5 s period, 0.2 s pulse width, and 0.0167 s sample width. Bare glassy carbon electrodes were cycled between −0.3 and +1.1 V via CV (scan rate: 100 mV s⁻¹) in PBS eight times to achieve a constant background. Ten consecutive CV cycles were then collected over the same potential window in the presence of 30 μM H₂S (positive initial sweep; scan rate: 100 mV s⁻¹). All voltammetric data herein is presented in the polarographic convention after background subtraction.

2.4. Constant potential amperometric measurements

All electrodes (either bare or film-modified) were first polarized at the applied potential of measurement in PBS for 15 min via constant potential amperometry (CPA) to achieve a stable background before use. Hydrogen sulfide sensitivity was measured on bare GCEs in a stirred solution of 20 mL deoxygenated PBS with five injections of 3.0 μM H₂S spaced 50–100 s apart. This procedure will hereafter be referred to as a “standard calibration.” With the applied potential set to +0.1, +0.3, +0.5, or +0.7 V for these standard calibrations, the sensitivity and LOD were calculated and compared in order to determine the optimal applied potential for H₂S detection (+0.3 V in our evaluation). Once the optimal applied potential was determined, the effect of surface sulfur passivation was investigated to determine an appropriate surface conditioning procedure. With the applied potential set at +0.3 V, five consecutive standard calibrations were carried out. Analytical metrics, including H₂S sensitivity, LOD, and background current, were calculated and compared as a function of calibration number (N_calibration).

With the applied potential and surface conditioning procedure optimized (+0.3 V and N_calibration = 3, respectively), standard calibrations were carried out on both bare and film-modified GCEs. Stock solutions of interferents were prepared in PBS to test for H₂S selectivity. Separate staircase amperograms were collected for each interferent with successive aliquot injections of AA (0.50 mM), AP (0.50 mM), DA (0.10 mM), CYS (0.50 mM), H₂O₂ (2.5 mM), NO₂⁻ (2.5 mM), or NO (0.475 mM). All selectivity measurements were carried out after surface preconditioning.

2.5. Measurements in simulated wound fluid

A simulated wound fluid (SWF) solution was prepared via a 10-fold dilution of FBS with PBS. Bare and film-modifed GCEs were calibrated in PBS three times before being transferred and polarized at +0.3 V in SWF for 1 h. Standard calibrations were then repeated three times in SWF. Relative sensitivity retention, change in LOD, and response time (90% max Δi) were evaluated to compare performance in SWF versus PBS. Based on these data, the best overall performing film modification (poly-o-PD) was selected for long-term stability testing. Poly-o-PD-modified electrodes were calibrated three times in PBS before polarizing for 24 h at a constantly held potential of +0.3 V in SWF. During this extended polarization, electrodes were calibrated in SWF at selected time-points, and selectivity coefficients versus AA, DA, and AP were remeasured at 24 h.

2.6. Calculations and statistical analysis

Selectivity coefficients for H₂S against the interferent species were calculated according to Eq. 3, where S_j represents the sensitivity towards interferent j, and S_H2S the H₂S sensitivity.

$$\log k_{H2S,j}=\log \left(\frac{s_j}{s_{H2S}}\right)$$ 3

The permeability of H₂S on film-modified GCEs was calculated according to Eq. 4, where P_H2S is the permeability of the film with respect to H₂S, and S_H2S,film and S_H2S,bare represent the sensitivities of the film-modified and bare GCEs, respectively.

$$P_{H2S}=\frac{S_{H2S,film}}{S_{H2S,bare}}$$ 4

Sensitivities towards H₂S, LODs, sensitivity retentions, background currents, selectivity coefficients, permeabilities, and response times are all presented (either numerically or with error bars) as the mean ± the standard error of the mean. Comparisons between data sets were performed using two-tailed t-tests.

3. Results and Discussion

3.1. Voltammetry of H₂S and interferents

Cyclic voltammetry (CV) and differential pulse voltammetry (DPV) are useful techniques for determining the electrochemical properties of redox systems under observation. In the case of hydrogen sulfide (H₂S), interpretation of voltammetric data is confounded by elemental sulfur poisoning and resulting surface passivation. Low potentials give rise to sulfur production (E⁰ = +0.171 V) while more positive potentials convert this sulfur to water-soluble sulfur oxides, including sulfate (E⁰ ≥ +0.365 V) [40]. The balance of these half-reactions depends upon the applied potential, and so the electrode surface is in a continuously changing state of passivation while the potential is swept. This phenomenon was evident in the CV traces collected in 30 μM H₂S cycling between −0.3 to +1.1 V (Fig. 1). With the first sweep on a clean GCE, the low-potential oxidation wave feature at +0.16 V was easy to discern. By the second cycle, the surface had been poisoned with low-potential features noticeably suppressed. Previous findings from our group have confirmed the presence of elemental sulfur on GCEs via X-ray photoelectron spectroscopy after electrochemical oxidation of hydrogen sulfide [48]. Higher-potential features (e.g., the initial peak at ca. +0.8 V) were also distorted and shifted positively, meaning higher potentials were needed to clear passivating sulfur and draw equivalent currents. With repeated cycling, these trends continued, erasing all readily identifiable redox features. The shape of the trace had stabilized by the tenth cycle.

Fig. 1.

Representative cyclic voltammograms collected using a GCE (3.0 mm dia.) in the presence of 30 μM H₂S in 10 mM PBS (pH 7.4) with a positive initial sweep and a scan rate of 100 mV s⁻¹. Traces represent the first (solid), second (dash), and tenth (dot) cycles. Blank PBS backgrounds were subtracted from all traces.

The DPV traces of common, electroactive biological interferents were also collected on a clean GCE (Fig. 2). Without the convoluting effects of surface passivation (as with sulfur), anodic peak potentials (E_a) could be accurately measured (Table S1). Based on their relatively low E_a values and high signal response, it was evident that Lascorbate and dopamine were the most problematic interferents to sensor selectivity on a GCE.

Fig. 2.

Representative differential pulse voltammograms collected on a GCE (3.0 mm dia.) in the presence of electroactive, biological interferents in 10 mM PBS (pH 7.4). L-ascorbate (AA), cysteine (CYS), dopamine (DA), acetaminophen (AP), hydrogen peroxide (H₂O₂), and nitrite (NO₂⁻) were tested at a concentration of 1.0 mM, and nitric oxide (NO) at 95 μM. Parameters: 50 mV amplitude, 4 mV step increase, 0.5 s period, 0.2 s pulse width, and 0.0167 s sample width. Blank PBS backgrounds were subtracted from all traces.

3.2. Surface conditioning for continuous electrochemical H₂S measurement

Continuous electrochemical techniques (e.g., constant potential amperometry, chronocoulometry) are simpler and allow for greater temporal resolution than voltammetry, but are no less susceptible to surface poisoning by H₂S. Deposition of an insulating sulfur layer on the electrochemical transducer reduces sensitivity and alters the analytical performance of the sensor. Although it may be possible to account for such changes with controlled H₂S exposure, continuous measurement from novel or unpredictable systems would remain unfeasible. It is therefore critical to ensure reproducible sensor performance while developing any continuous H₂S sensor for biological applications. Unlike other methods that seek to mitigate surface passivation (e.g., redox mediators, cleaning pulses), a surface conditioning procedure “pre-poisons” the electrode until a steady-state performance is reached [46, 48]. While variability is greatly diminished, a repercussion is that the steady-state sensitivities may become so low that LODs are pushed to physiologically irrelevant levels (i.e., >100 nM) [28].

A suitable working potential was first determined to assess the utility of surface conditioning for a continuous electrochemical H₂S sensor. With the effects of poisoning, the only significant voltammetric feature of the H₂S trace was an increase in current with the increase in potential in the range of +0.1—1.1 V under CV (Fig. 1). As a direct result, the H₂S sensitivity increased with applied potential when measured at +0.1, +0.3, +0.5, and +0.7 V under constant potential amperometry (CPA; Fig. 3A). While potentials greater than +0.7 V are able to further oxidize deposited sulfur, a priori concerns with selectivity (particularly against NO and nitrite; Table S1) and electropolymerized film damage led to +0.7 V being selected as an upper bound to the applied potential. Increasing the applied potential also reduced the LOD significantly, particularly from +0.1 to +0.3 V (Fig. 3B). Subsequent increases to +0.5 and +0.7 V did not greatly affect the LOD due to intensification of the background noise. The plateau in LOD, in addition to the goal of remaining below interferent peak anodic potentials (Table S1), motivated our selection of +0.3 V as a suitable working potential.

Fig. 3.

(A) Hydrogen sulfide sensitivity and (B) LOD of a bare GCE (3.0 mm dia.) as a function of the applied potential used under CPA (n ≥ 8). ns = p ≥ 0.05, * = p < 0.05, ** = p < 0.01, *** = p < 0.001.

An appropriate surface conditioning procedure should provide enough exposure to the poisoning agent to ultimately elicit steady-state performance. In the context of electrochemical H₂S sensing, the target analyte and poisoning agent are the same molecule, so surface conditioning/exposure can simply be carried out with consecutive calibrations. The H₂S exposure within a single calibration will depend on the concentration range tested and duration of exposure. Based on these criteria, we have defined a “standard calibration” (see. and Fig. 4) to convey key performance metrics as a function of calibration number (N_calibration). As shown in Fig. 5, the H₂S sensitivity, LOD, and background current were each monitored and found to converge to values statistically insignificant from one another after three standard calibrations. More generally, the surface conditioning was completed upon the passage of 580 ± 110 μC cm⁻² (background charge density subtracted). While the sensitivity fell to approximately 80% of its original value (Fig. 5A) and the LOD increased slightly, both still allowed for measurement of biologically relevant concentrations of H₂S (Fig. 5B) [28]. An important clarification here is that while calibrations are discrete, poisoning occurs continuously throughout the calibration. The percent sensitivity retention will therefore depend upon the poisoning exposure (i.e., length and H₂S injection concentrations) occurring in the first calibration, to which all sensitivities are normalized. Abbreviated calibrations, though less accurate for measuring performance metrics, would foster greater resolution of poisoning dynamics and yield smaller retentions in sensitivity. For the purpose of surface conditioning an electrode quickly, the use of three standard calibrations was considered most favorable.

Fig. 4.

Example staircase amperogram collected on a bare GCE (3.0 mm dia.) in 10 mM PBS (pH 7.4) with successive injections of hydrogen sulfide under CPA (applied potential: +0.3 V). Inset: corresponding calibration curve from plateau currents.

Fig. 5.

Analytical performance metrics of a GCE (3.0 mm dia.) as a function of the number of standard hydrogen sulfide calibrations carried out. (A) Sensitivity retention, (B) LODs, and (C) background currents measured under CPA with an applied potential of +0.3 V (n ≥ 8). ns = p ≥ 0.05, * = p < 0.05, ** = p < 0.01, *** = p < 0.001.

3.3. Analytical performance of film-modified electrodes

The performance of a surface-conditioned, but otherwise unmodified GCE affords high H₂S sensitivity (−4.03 ± 0.91 nA μM⁻¹) and a low LOD (1 nM) but does not sufficiently block nonspecific oxidation of key interferent species (Table 1). The choice of electrochemical technique and associated parameters does lend some control over selectivity. For instance, the use of a relatively low applied potential (e.g., +0.3 V) virtually eliminates the response to NO and nitrite oxidation (selectivity coefficients <−5). Furthermore, without coordination to surface-confined metals (e.g., Pt, Sn) CO cannot be readily oxidized at a GCE [56]. Aside from these interferents, AA, AP, DA, CYS, and H₂O₂ are each oxidized at potentials near or below +0.3 V and would thus distort H₂S measurements in biological milieu (Table S1).

Table 1.

Analytical merits of electropolymerized film-modified GCEs for hydrogen sulfide detection.

Electrodea	LODb (nM)	SH2sc (nA μM−1)	PH2Sd	logkH2S,je
				AA	AP	DA	CYS	H2O2
GCE (bare)	1 ± 0	−4.03 ± 0.91	1.00 ± 0.23	−0.04 ± 0.11	−1.89 ± 0.14	0.76 ± 0.18	−1.77 ± 0.14	−4.58 ± 0.30
GCE/poly-5A1N	9 ± 6	−0.42 ± 0.08	0.10 ± 0.02	−1.92 ± 0.10	−2.97 ± 0.33	−1.81 ± 0.30	−3.23 ± 0.19	−4.72 ± 0.49
GCE/poly-phenol	79 ± 51	−0.05 ± 0.04	0.01 ± 0.01	−1.35 ± 0.34	−2.97 ± 0.52	−0.13 ± 0.42	−2.92 ± 0.46	−4.65 ± 0.62
GCE/poly-eugenol	10 ± 5	−3.92 ± 1.42	0.97 ± 0.35	−0.76 ± 0.14	−1.24 ± 0.08	0.42 ± 0.22	−1.07 ± 0.08	−4.10 ± 0.16
GCE/poly-o-PD	15 ± 12	−0.28 ± 0.06	0.07 ± 0.01	−2.39 ± 0.16	−4.49 ± 0.42	−2.50 ± 0.21	−2.78 ± 0.16	−5.01 ± 0.15
GCE/poly-m-PD	73 ± 37	−0.42 ± 0.06	0.10 ± 0.01	−3.09 ± 0.06	−3.11 ± 0.16	−2.23 ± 0.32	−3.36 ± 0.10	−4.57 ± 0.22
GCE/poly-p-PD	30 ± 10	−4.24 ± 0.58	1.05 ± 0.14	−2.40 ± 0.07	−4.28 ± 0.12	−3.09 ± 0.08	−2.28 ± 0.13	−4.79 ± 0.09

^aGlassy carbon electrode (GCE; 3.0 mm dia.).

^bS/N = 3.

^cSensitivity calculated from third standard calibration (applied potential: +0.3 V).

^dPermeability calculated using Eq. 4.

^eSelectivity coefficients calculated using Eq. 3 against interferent j as follows: L-ascorbate (AA; 0.25 mM), acetaminophen (AP; 0.50 mM), dopamine (DA; 0.10 mM), cysteine (CYS; 0.50 mM), hydrogen peroxide (H₂O₂; 2.5 mM). For all measurements, n ≥ 8.

In an effort to further enhance selectivity for H₂S, bare GCEs were modified with electropolymerized films via CV using a series of monomers known to undergo irreversible oxidation to cationic radicals in aqueous media under sufficiently positive potentials [53]. With radical coupling and secondary oxidations, oligomers whose size eventually exceeds solubility crash out from solution onto the electrode surface, building the electropolymerized film. The CV traces of the monomers reveal that the majority of the deposition occurs within the first sweep, as subsequent cycles draw greatly diminished current (Fig. 6). As the film approaches complete coverage, monomers in solution are blocked from the electrode surface, cutting off radical cation generation and self-terminating growth. Ultimately, the shapes of the CV traces are dictated by both the redox properties of the monomer and this insulating process, the latter capable of distorting higher-potential wave features. Nevertheless, peak oxidation potentials were measured from the first cycle with the 0 to +1.0 V window sufficient to oxidize all monomers (Table S2). The use of cyclic voltammetry at a slow scan rate (10 mV s⁻¹) creates more tightly packed films with greater adherence compared to CPA depositions [53, 57, 58]. The presence of multiple peaks in some cases (e.g., 5A1N and o-, m-, p-PD) was attributed to secondary/tertiary oxidations of soluble oligomers and/or over-oxidation of the film itself (which may retain a net positive charge) [51, 59]. The absence of reduction waves in the return traces is evidence of the insulating properties of these films, with the exception of 5A1N. The small, cathodic feature in the +0.3 to 0 V range demonstrates poly-5A1N’s conductive properties. Unlike other conductive films (e.g., poly-aniline), the polymerization mechanism of 5A1N is not autocatalytic and will proceed to self-terminate as would an insulating film [53, 54, 60].

Fig. 6.

The first (solid) and second (dashed) cycle of electrodepositions carried out on a 3.0 mm dia. GCE in 10 mM monomer solutions of (A) 5A1N, (B) phenol, (C) eugenol, (D) o-PD, (E) m-PD, and (F) p-PD in 10 mM PBS via cyclic voltammetry (0 to +1.0 V positive sweep; scan rate: 10 mV s⁻¹). The pH of the phenol, o-PD, m-PD, and p-PD solutions was 7.4. Solutions of 5A1N and eugenol were titrated to pH 1.0 with HCl and pH 13 with NaOH, respectively. Insets: monomer chemical structures.

With film modification, the permeability and sensitivity towards H₂S decreased by roughly one order of magnitude for poly-5A1N, -o-PD, and -m-PD, and by two orders for poly-phenol (Table 1). A previous comparison of these monomers in the context of NO sensing revealed that poly-5A1N and poly-phenol were more permeable towards NO than presently towards H₂S [53]. Although comparable in size to NO and H₂O in its undissociated form, at pH 7.4 hydrogen sulfide exists primarily (~80%) as dissociated hydrosulfide (HS⁻), whose hydration shell may impede film permeation in these films [61]. Both poly-eugenol and poly-p-PD retained H₂S permeability values close to unity (i.e., the bare GCE), suggesting that their polymeric structures may confer a unique advantage to H₂S detection (Table 1). As with bare GCEs, film-modified electrodes were surface conditioned with multiple H₂S calibrations (N_calibration = 3) to ensure performance stability (Fig. S1). Similar sensitivity reductions (20–30%) suggested that the presence of the film, regardless of monomer identity, did not greatly affect the surface conditioning process.

As a result of the reduced sensitivity, the LOD predictably increased (i.e., reciprocal changes in order of magnitude) for electrodes modified with poly-5A1N, -o-PD, and -phenol (Table 1). An intensification in the background noise resulted in polyeugenol, -m-PD, and -p-PD to have LOD values higher than permeability alone would predict. From lowest (poly-5A1N, 9 ± 6 nM) to highest LOD (poly-phenol, 79 ± 51 nM), each film modification studied would still permit measurement of biologically relevant concentrations of H₂S. Though a permselective barrier may nonideally reduce the permeability of the target analyte, the membrane will still confer a selectivity advantage if greater obstruction occurs against interferent species.

The primary blocking mechanism of electropolymerized films is size-exclusion, particularly in the case of neutrally charged molecules (reducing solvation and chargerepulsion interactions) [51, 52, 62, 63]. With an applied potential of +0.3 V, the bare GCE confers inherent electrochemical selectivity for H₂S over neutrally charged interferents acetaminophen (AP), cysteine (CYS), and hydrogen peroxide (H₂O₂) at pH 7.4 (Table 1). With film modification, the selectivity against H₂O₂ was maintained or marginally improved for each monomer studied, with the exception of eugenol, for which selectivity was slightly worse. In contrast to the larger interferents AP and CYS (against which selectivity was greatly improved over the bare GCE performance for each monomer except eugenol), the poor enhancement of H₂O₂ selectivity is rationalized by its comparatively small size. As each of these films has been studied previously with respect to H₂O₂ sensing (e.g., in the context of glucose biosensors), it was expected that H₂O₂ would be able to freely permeate [50–52]. The high overall H₂O₂ selectivity (coefficients <−4 for all films) is therefore chiefly attributed to the inherent electrochemistry for GCEs rather than differential permeation (Fig. 2). Against AP, GCEs modified by films of any of the phenylenediamine series performed exceptionally well with selectivity coefficients all below −3; and CYS was best rejected by poly-5A1N and poly-m-PD films (both <−3).

Dopamine (DA) and L-ascorbate (AA) are comparable in size to AP and CYS but at pH 7.4 retain positive (+1) and negative (−1) charges, respectively. The bare GCE has poor selectivity for H₂S over DA and AA given their low oxidation potentials (E_a < +0.3 V) and large signal responses (Table S1, Fig. 2). In fact, the bare GCE is even more responsive towards DA than H₂S, indicated by the positive selectivity coefficient (0.76 ± 0.18; Table 1). In order to isolate the selectivity benefits of the film modifications, changes in selectivity should be considered relative to the performance of the bare GCE. For each film modification investigated except poly-eugenol, selectivity for H₂S against both DA and AA was dramatically improved—most notably so with the phenylenediamine series (coefficients all <−2). Modifications with poly-5A1N, -o-PD, and -p-PD all had greater improvements in electrode selectivity against DA than compared to similarly-sized AA, AP, and CYS, suggesting that these films may retain a net positive charge for coulombic repulsion of DA. Improvements to selectivity against oppositely-charged DA and AA were within error of each other for poly-phenol and polym-PD films, indicating insignificant coulombic interactions.

Poly-eugenol films did not perform well in any measure of selectivity. Given the large measured H₂S permeability, the pores are likely large and inadequate for sieving out larger interferents [52, 64]. We have previously reported that poly-5A1N and poly-phenol have notable nitrite rejection capabilities [53]; given hydrosulfide’s similar size and charge, these membranes proved nonideal for H₂S/HS⁻ sensing. Overall, the phenylenediamine series of films performed best in terms of improving selectivity against all interferents, and of this series, poly-o-PD stood out for its low LOD (Table 1).

3.4. Performance in simulated wound fluid

In addition to interferents, a significant challenge for in situ biological H₂S measurements is biofouling. Protein accumulation on the transducer surface reduces sensitivity and degrades sensor performance [65]. Appropriate sensor characterization should be carried out in the media of measurement or closest simulant thereto to fully characterize utility. To this end, bare GCE and film-modified electrodes were polarized and calibrated in simulated wound fluid (SWF) to quantify any detrimental effects of biofouling (Fig. 7). The bare GCE retained 36 ± 5 % of its initial sensitivity in SWF relative to PBS trials (Fig. 7A) with an increase in LOD to 21 ± 9 nM (Fig. 7B). Modification with electropolymerized films significantly improved sensitivity retention, regardless of the electropolymerized film employed. An unmodified, positively-polarized GCE will plausibly attract more protein than one modified with an insulating film since most blood plasma proteins are negatively charged at pH 7.4 (e.g., albumin) [66]. Poly-oPD performed better than all other film modifications with statistical significance (p < 0.05 vs. 5A1N; p < 0.01 vs. others), retaining 79 ± 6 % of its sensitivity in SWF. Previous studies have also highlighted poly-o-PD for its anti-biofouling properties in proteinaceous media, possibly due to its greater compactness [53].

Fig. 7.

(A) H₂S sensitivity retention of bare and electropolymerized film-modified GCEs (3.0 mm dia.) in SWF relative to performance in PBS post-surface conditioning (significance with respect to the bare GCE). (B) Detection limit (S/N = 3) of electrodes in PBS (blue) and in SWF (red; significance with respect to corresponding PBS trials). (C) Response time (90% max Δi) of electrodes in SWF (significance with respect to the bare GCE). * = p < 0.05, ** = p < 0.01, *** = p < 0.001. For all measurements, n ≥ 8.

Although having better sensitivity retention compared to bare GCEs, the majority of the film-modified electrodes predictably experienced an increase in LOD with the reduction in sensitivity (poly-eugenol and poly-p-PD being the exceptions; Fig. 7B). Severely increased LODs for poly-5A1N and poly-phenol films resulted from an increase in the background noise. Along with poly-m-PD, the LOD values in SWF were too high to have any practical biological use. Of the remaining films with LOD < 100 nM, the H₂S response times of poly-eugenol- and poly-o-PD-modified electrodes were statistically insignificant from bare GCEs (15 ± 10 s; Fig. 7C). Protein adsorption did not greatly affect response time overall, since the bare GCE responded in 12 ± 10 s in PBS trials (i.e., identically to SWF trials). The >100 s response times of poly-m-PD- and poly-p-PDmodified electrodes were deemed prohibitive to in situ measurement.

Combined with the data from selectivity measurements, the best overall performing film modification was determined to be poly-o-PD, which was selected for further testing of biological interferent rejection and sensor stability with long-term continuous operation. The permselectivity traits of poly-o-PD were evaluated against other biologically relevant sulfur-containing compounds, including cysteamine, glutathione, and homocysteine. The results indicated that the poly-o-PD-modified sensor was at least an order of magnitude more sensitive towards H₂S than other sulfurcontaining compounds (Table S3). Continuous amperometric measurement in proteinaceous media may strain the electropolymerized film through protein adsorption and potential-induced degradation [65, 67]. With partial or complete stripping of the poly-o-PD film, the performance of the electrode would be expected to eventually approach that of the bare GCE [54]. On one hand, film destruction would increase the sensitivity towards H₂S in PBS (P_H2S = 0.07 ± 0.01 for poly-o-PD; Table 1), but the electrode would also be more susceptible to biofouling in SWF (Figure 7A). Combining these effects, an approximately 7-fold increase in sensitivity would be expected in the worst-case scenario of total membrane delamination. After 24 h continuous use (+0.3 V applied potential), the H₂S sensitivity of the poly-o-PD-modified electrodes increased by a factor of 1.35 ± 0.07, indicative of partial film wear. A study of the time-dependent sensitivity fluctuations in SWF revealed that the sensitivity equilibrated within the first 2 h of operation, demonstrating that the sensor only requires short polarization times before use in biological contexts (Fig. S2).

As further evidence of sensor wear, selectivity coefficients against DA and AP increased by one order; however, it is worth noting that the poly-o-PD film still furnished significant selectivity benefits over the bare GCE (Table 2). Interestingly, the selectivity coefficient against AA improved slightly after long-term use. The adsorption of negatively-charged proteins on the surface may confer additional permselectivity characteristics, or in light of poorer coefficients against DA and AP, the protein layer may repel negatively-charged AA. Overall, the poly-o-PD-modified electrode may be used for long-term continuous measurements without serious degradation in selectivity provided periodic recalibrations are used to account for changes in sensitivity.

Table 2.

Selectivity characteristics of poly-o-PD-modified GCEs with long-term use in SWF.

Electrode	Condition	logkH2S, ja
		AA	AP	DA
GCE (bare)	New	−0.04 ± 0.11	−1.89 ± 0.14	0.76 ± 0.18
GCE/poly-o-PD	New	−2.39 ± 0.16	−4.49 ± 0.42	−2.50 ± 0.21
GCE/poly-o-PD	After 24 h use in SWFb	−2.99 ± 0.12	−3.15 ± 0.08	−1.51 ± 0.03

^aSelectivity coefficients calculated with Eq. 2 for H₂S against interferent j as follows: L-ascorbate (AA; 0.25 mM), acetaminophen (AP; 0.50 mM), dopamine (DA; 0.10 mM).

^bUnder CPA with +0.3 V applied potential. For all measurements, n ≥ 8.

4. Conclusions

Direct, continuous electrochemical measurement of hydrogen sulfide is possible without a redox mediator or integrated cleaning pulse provided that transducer sulfur poisoning is accounted for carefully. Front-end surface conditioning through repeated H₂S calibrations may be used to deposit a sulfur layer on the electrode before continuous measurement. Although this procedure reduces sensitivity, performance variability is practically eliminated while LODs remain biologically relevant (i.e., > 100 nM). Of the electropolymerized film modifications studied herein, poly-o-PD best improved the overall H₂S sensing performance of the modified GCEs when considering LOD, selectivity versus key interferent species, sensitivity retention in proteinaceous media, and response time. Future work should focus on miniaturizing poly-o-PD-modified GCEs for continuous measurement in biological applications.