The development of a fully integrated 3D printed electrochemical platform and its application to investigate the chemical reaction between carbon dioxide and hydrazine

Highlights

• •An integrated electrochemical platform was manufactured by bi-material 3D printing.
• •It was applied to investigate the reaction between hydrazine and carbon dioxide.
• •Experimental results were supported by finite-element method numerical simulations.

Abstract

The combination of computer assisted design and 3D printing has recently enabled fast and inexpensive manufacture of customized ‘reactionware’ for broad range of electrochemical applications. In this work bi-material fused deposition modeling 3D printing is utilized to construct an integrated platform based on a polyamide electrochemical cell and electrodes manufactured from a polylactic acid-carbon nanotube conductive composite. The cell contains separated compartments for the reference and counter electrode and enables reactants to be introduced and inspected under oxygen-free conditions. The developed platform was employed in a study investigating the electrochemical oxidation of aqueous hydrazine coupled to its bulk reaction with carbon dioxide. The analysis of cyclic voltammograms obtained in reaction mixtures with systematically varied composition confirmed that the reaction between hydrazine and carbon dioxide follows 1/1 stoichiometry and the corresponding equilibrium constant amounts to (2.8 ± 0.6) × 10³. Experimental characteristics were verified by results of numerical simulations based on the finite-element-method.

Graphical abstract

1. Introduction

The recent decade has witnessed an expansion of 3D printing (3DP) technologies into each and every field of science involving electrochemistry [1,2]. Approaches of 3DP create objects by positioning target material on previously formed layers. The combination of 3DP with computer assisted design (CAD) allows rapid, inexpensive, safe and environmentally friendly manufacture of customized ‘reactionware’ (cells and electrodes) directly in an electrochemical laboratory.

Electrochemical cells manufactured by 3DP offer several advantages over their glass counterparts: (I) cells may be printed directly in a chemical laboratory without specially trained personal, (II) CAD allows rapid prototyping and redesign of cells, (III) cells with complicated inner structure may be created, (IV) 3DP cells are significantly less breakable than glass cells, (V) 3DP offers considerable reduction of material and energy consumption. The first 3DP electrochemical cell was constructed by Symes et al. [3] in 2012 by extruding acetoxysilicone polymer through a syringe and casting it onto a glass substrate. Presently, the most widely used 3DP technologies are selective laser melting (SLM), stereolithography (SLA) and fused deposition modelling (FDM) [4]. The SLM technique creates objects by fusing atomized metallic powders melted locally by a laser beam and was applied to manufacture metallic electrodes for supercapacitors [5] and water electrolysers [6], [7], [8]. Recently, multi-material mode of SLM technique employing stainless steel, titanium and graphite was utilized to manufacture integrated proton exchange membrane electrolyzer cells [9]. The SLA technique forms objects from layers of resins polymerized locally by a laser-induced photochemical process. This technique was employed to develop cells for water electrolysis [10], spectroelectrochemical measurements [11] and electroanalytical applications [12]. Though the first conductive resins have been recently synthesized [13], values of their electric conductivity are insufficient to manufacture electrodes. The FDM as the most common and versatile 3DP technique relies on extruding filament through a heated nozzle. Low expenses and wide availability of printers render the FDM particularly suitable for the manufacture of cells. Most frequently used filaments are polylactic acid (PLA), [8,7,[14], [15], [16], [17], [18], [19] acrylonitrile butadiene styrene (ABS), [20], [21], [22], [23], [24], [25], [26], [27], [28], [29] polypropylene [30–32], and polyesters [33]. Cells manufactured by FDM 3DP involve multitude of functionalities and have found use in a broad range of electrochemical processes and applications, which are summarized in Table 1 . However, none of these cells allows a fundamental analysis of electron transfer processes and related phenomena which requires strictly oxygen-free environment, separated compartments for working, counter and reference electrode and an interface to introduce reaction components. The goal of this work is the development, manufacture and inspection of such cell design.

Table 1.

Summary of works emplyoying cells manufactured by fused deposition modelling 3D printing.

Material	Cell functionality	Working electrode	Application	Ref.
PLA	liquid/gas cells with a 3D printed separator	stainless steel (SLM technique)	water electrolysis	[7]
	simple cell, no functionality	stainless steel (SLM technique)	water electrolysis	[8]
	flow cell	PLA/graphene	determination of catechol	[14]
	flow cell with a gas purging system	fluorine-doped tin oxide on glass	wastewater cleaning	[15]
	flow cell	iron	electrosynthesis of magnetite particles	[16]
	simple cell, no functionality	PLA/carbon black	determination of Hg2+ ions	[17]
	simple cell, no functionality	PLA/carbon black	determination of Hg2+ions, caffeine and H2O2	[18]
	two compartment cell with gas purging system, devided by a membrane	Pd nanocubes-modified glassy carbon	electrooxidation of glycerol	[19]
ABS	flow capillary cell	Prussian blue/ gold	determinantion of H2O2	[20]
	flow cell for batch injection analysis	PLA/graphene	determination of tert-butylhydroquinone, dipyrone, dopamine and diclofenac	[21]
	flow cell separated by a membrane	nickel foil	mass transport characterization	[22]
	cell for batch injection analysis	Prussian blue/carbon	determination of glucose	[23]
	spectroelectrochemical cell	PLA/graphene	Raman spectroelectrochemistry	[24]
	flow cell	copper	characterization of current distribution	[25]
	Y-shaped flow cell	pencil (graphite)	enzymatic glucose biofuel cell	[26]
	simple cell, no functionality	PLA/carbon black	determination of organic acids and dopamine	[27]
	two compartment cell connected by a salt bridge	PLA/graphene, PLA/carbon black	water electrolysis	[28]
	Y-shaped flow cell	reduced graphene oxide	determination of dopamine	[29]
polypropylene	flow cell separated by a membrane	silver	water electrolysis	[30]
	packed-bed scalable electrochemical reactor	boron-doped diamond	electrosynthesis of graphene oxide	[31]
	flow cell	carbon	methoxymethylation of alcohols	[32]
co-polyester	electrochemical biosensor	screen-printed gold electrodes modified by antibodies	determination of tumor cells	[33]
polyamide 12	three compartment cell connected by Luggin capillaries, dedicated to liquid/liquid electrochemistry	platinum	determination of ephedrine	[34]

Recent advancements in the preparation of printable conductive composite materials enabled the use of FDM to manufacture electrodes. In the pioneering work of Rymansaib et al. [35] polystyrene was blended with graphite and carbon nanofibers to extrude a conductive composite filament further employed to print electrodes for the electroanalytical determination of Pb²⁺ ions. Following works employed commercial composites of PLA with graphene, [14,21,24,28,36,37] carbon black [17,18,28,[38], [39], [40], carbon nanotubes [41] and copper [42] as conductive fillers. Electrodes manufactured from in-lab synthesized PLA-based composites containing MoSe₂, carbon/platinum [43] and graphite [44] used as conductive additives were reported recently. Furthermore, commercial carbon-loaded ABS filaments were employed [17,45]. Bi-material FDM 3D printing combining conductive composites for electrodes and insulating filaments for cells enabled the manufacture of integrated platforms dedicated to electroanalysis [14,17,18,21,27,35,45] and Raman spectroelectrochemistry [24].

In this work we employ CAD combined with bi-material FDM 3DP to devise and manufacture a fully integrated electrochemical platform based on a three-compartment polyamide (PA12) cell and a PLA-carbon nanotube (PLA-CNT) composite working electrode. Employing cyclic voltammetry with the reversible electroactive probe we have recently demonstrated that PLA-CNT electrodes activated by an anodic oxidation exhibit area of the electrochemically active surface very close to the geometrical surface area. The ratio of the two quantities, further reffered to as relative area of the electrochemically active electrode surface, is thus close to unity. Cyclic voltammetric experiments with the electroactive probe further revealed that activated PLA-CNT electrodes show interfacial electron transfer rate very similar to conventionally used metallic or carbon electrodes [41]. Electrodes prepared by the reported manufacture/activation protocol are integrated with the 3DP electrochemical cell developed in this work. Importantly, the presented design allows reactants to be introduced and investigated in the oxygen free environment. The platform is designed for a systematic analysis of electron transfer processes coupled with bulk chemical reactions, which is not yet reported in the literature.

It is generally accepted that global warming is caused by rising levels of carbon dioxide largely generated by anthropogenic sources [46]. One of the strategies to mitigate the impact of human activities on the environment is the capture of carbon dioxide and its subsequent conversion to added-value chemicals [47,48]. Aqueous solutions of amines are seen as promising agents for the CO₂ capture [49], [50], [51], [52] Hydrazine, as the smallest diamine, potentially offers higher absorption capacity compared to organic amines and is thus considered as a particularly attractive CO₂ sorbent [53,54]. Furthermore, hydrazine was utilized to eliminate other environmental pollutants including halogens [55], heavy metal ions [56,57], or NO_x [58,59]. Importantly, hydrazine may be generated electrochemically [60] and if renewable energy sources are utilized in this process, it is produced without generating carbon dioxide. Motivated by the environmental potential of hydrazine we apply the electrochemical platform developed in this work to investigate the electrochemical oxidation of aqueous hydrazine coupled to its bulk reaction with carbon dioxide. We utilize cyclic voltammetry to determine the equilibrium concentration of hydrazine in its reaction mixtures with carbon dioxide and employ obtained results to determine the equilibrium constant of the reaction considering several possible stoichiometric scenarios. Experimentally obtained characteristics are corroborated by the results of finite-element-method numerical simulations.

2. Experimental section

2.1. Materials and chemicals

In all experiments performed in this work MilliQ deionized water (Millipore, France, maximum total organic carbon 3 ppb, minimum electric resistivity 18.2 MΩ cm) was used. All glassware (U-cell for the electrochemical activation of electrodes, volumetric flasks and beakers) and PTFE parts (magnetic stirrer) were cleaned by boiling in diluted (25 %) nitric acid (Lach-Ner, Czech Republic). Upon cooling down the glassware to the ambient temperature it was boiled in water four times and dried overnight at 105 °C. This purification protocol is known to guarantee cleanliness required to successfully perform sensitive electrochemical experiments that include desorption of self-assembled monolayers at monocrystalline surfaces [61] or charge transport studies performed at a single molecule level [62]. Ruthenium acetylacetonate (97 %, further abbreviated as Ru(acac)₃) was obtained from Sigma-Aldrich. Sodium sulfate (99.99 %, Suprapure, anhydrous) was obtained from Merck. Potassium sulfate (p.a.) was purchased from Lach-ner. Hydrazine (reagent grade, 50 – 60 % aqueous solution) was obtained from Sigma-Aldrich. Compressed gaseous argon (200 bar, 99.996 %) was purchased from Messer. Solid CO₂ (dry ice) was obtained from Linde. Natural polyamide 12 (PA12) filament was purchased from Devil Design, Poland. Electrically conductive polylactic acid-carbon nanotube (PLA-CNT) composite filament (brand name Alfaohm) was obtained from Filoalfa, Italy. Both PA12 and PLA-CNT filaments have the diameter of 1.75 mm. All 3D printed objects used in this work (the electrochemical cell including the lid, electrodes) were manufactured by FDM 3DP employing Prusa I3 MK3S 3D printer, E3D V6 brass nozzle (diameter 0.40 mm) and smooth spring steel sheet printing surface coated with polyetherimide (all three items purchased from Prusa Research, Prague, Czech Republic).

2.2. Cell design and manufacture

The electrochemical cell is dedicated to experiments in a three-electrode arrangement – involving working (WE), reference (RE) and counter (CE) electrode. The cell was devised by computer assisted design (CAD) employing the environment of Autodesk Fusion 360 (Autodesk Inc., USA). CAD drawing of the cell and its cross section is visualized in Fig. 1 A and 1B, while a photograph of the cell under operating conditions is displayed in Fig. 1C. Further details on the cell design, manufacture and inspection are in the Supplementary Information.

Fig. 1.

(A) Full and (B) cross-sectional CAD drawing of the integrated electrochemical platform devised, manufactured and inspected in this work. (C) Photograph of the platform under operating conditions.

2.3. Electrode design, manufacture, activation and inspection of interfacial electron transfer properties

Working electrodes have cylindrical shape (diameter 1.75 mm, height 40 mm) and were printed employing a 0.40 mm nozzle with layer height set to 0.05 mm. All other printing parameters were identical to those reported previously [41]. Just before the experiment the WE was activated by anodic polarization in aqueous 0.3 M Na₂SO₄ at +6.0 V for 10 s (for details see Ref. [41]). Upon the activation, the WE was briefly rinsed in water, dried in the air and masked by a PTFE tape leaving a part with height of 12 mm exposed (geometrical electrode/electrolyte interface area 0.84 cm² obtained by calculations employing CAD, procedure not shown). The performance of manufactured electrodes was inspected by measuring their electron transfer properties by cyclic voltammetry performed in deaerated aqueous 0.1 M K₂SO₄ supporting electrolyte containing 5 × 10⁻⁴ M Ru(acac)₃ electroactive probe in the PA12 cell designed and manufactured as discussed in the Supplementary Information. For details on the cell assembly and operation, see Electrochemical measurements part that follows. The Faradaic peak-to-peak separation and the relative area of the electrochemically active electrode surface were evaluated as discussed in Ref. [41].

2.4. Electrochemical measurements

All electrochemical measurements were carried out in the PA12 cell designed and manufactured as discussed in the Supplementary Information. The WE compartment of the cell was filled with 0.1 M K₂SO₄ supporting electrolyte containing 3.85 × 10⁻³ M NH₂NH₂. The RE and CE compartments were filled by applying a vacuum generated by a syringe/piston/needle assembly operating through a rubber septum. Subsequently, RE and CE were introduced. The solution left in the WE compartment was discarded and a fresh solution with volume of 17.0 mL (denoted as V _NH2NH2 in the Results and Discussion section) was added. A magnetic stirrer was introduced. The lid equipped with freshly activated WE was inserted, closing the cell. The cell assembly was finalized by connecting electrodes to the potentiostat (PGSTAT Autolab, Metrohm, Czech Republic), attaching the inlet/outlet tubing system (Fig. 1C) and purging the WE compartment by a gentle stream of argon gas saturated with the identical aqueous hydrazine solution for 5 minutes. Electrochemical measurements were performed in the staircase cyclic voltammetry regime at electrode polarization rate of 25 mV s⁻¹ and potential step of 1 mV. Before each measurement, electrochemical impedance spectroscopic measurement was performed in the double-layer capacitance region of the inspected system at 5 mV AC peak-to-peak amplitude and AC frequency ranging from 10⁵ to 10⁰ Hz. The value of the electric resistance was extracted graphically from the real part of the impedance in the Nyquist plot (data not shown). Such obtained value multiplied by a factor of 90 – 95 % was employed as the input parameter for positive feedback electronics to compensate voltage losses between the WE and the RE.

The cyclic voltammetry was first carried out in the absence of CO₂. Subsequently, desired amount of CO₂ was introduced as follows. Saturated aqueous solution of CO₂ was prepared by bubbling sublimated dry ice through water for at least one hour while continuously monitoring the gas pressure and the solution temperature. The concentration of CO₂ in its saturated solution (denoted as c _{0, CO2}) was calculated employing Henry`s law and solubility-temperature profiles reported for CO₂ at standard pressure [63]. The pressure of 1.01 bar and temperature ranging from 294 to 297 K were applied. Obtained values of c _{0, CO2} range from 4.05 to 3.68 × 10⁻² M. The saturated solution of CO₂ was introduced to the WE compartment of the cell by an air-tight syringe/piston/needle assembly through the septum in the lid. The volume of the solution V _CO2 ranged from 0.5 to 2.0 mL. The resulting NH₂NH₂/CO₂ reaction mixture was homogenized by a magnetic stirrer at 500 rpm for 5 minutes. The concentration of unreacted NH₂NH₂ was determined by cyclic voltammetry (see Results and Discussion section for details). In a control experiment, the entire procedure was precisely repeated with saturated solution of Ar instead of CO₂.

3. Calculation section

3.1. Finite-element-method numerical simulations

The interfacial electron transfer process (irreversible oxidation of hydrazine involving four electrons) and the bulk chemical reaction (synthesis of NH₂NH₂ and CO₂) studied experimentally in this work were further modelled theoretically in COMSOL Multiphysics 5.2a software [64]. The reaction scheme is summarized in Fig. 6. The electrolyte phase is represented by a 2D computational domain of a rectangular shape with length of 1 × 10⁻² m and height of 5 × 10⁻⁵ m (Fig. 2 ). In the domain, governing continuity equations were formulated for concentration values of considered system constituents – NH₂NH₂, CO₂, the product of their bulk synthetic reaction NH₂NH₂CO₂ (denoted as P, assuming 1/1 reaction stoichiometry, see Results and Discussion for details) and products of the hydrazine oxidation N₂ + 4H⁺ (cumulatively denoted as P´). The computational domain was divided to triangular non-uniform mesh with the highest density of points near the electrode (Fig. 2, left, shown in red). The electrode boundary involved 500 points. The maximum element size was set to 5 × 10⁻⁵ m, the minimum element size to 3 × 10⁻⁵ m, the maximum element growth rate to 1.2 and the curvature factor to 0.3.

Fig. 6.

Simulated response of the electrode/electrolyte interface in the system initially containing 3.85 × 10⁻³ M NH₂NH₂ in the entire computational domain (black solid curve). The response of the system with additionally introduced 3.28 × 10⁻³ M of CO₂ at t = 0 employing kinetic (coloured dotted curves) and thermodynamic (black dashed curve, ∞) model. Numbers denote values of $k_f^{bulk}$ (1, 5, 7 and 10) in 10⁹ M⁻¹ s⁻¹ employed in the kinetic model.

Fig. 2.

Schematic depiction of the computational domain. Γ_N denotes the Neumann boundary condition at the electrode (red) and insulating walls (blue). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

3.2. The kinetic model

It formulates the bulk reaction between NH₂NH₂ and CO₂ employing reaction rate (source/sink) terms in continuity equations. Altogether four continuity equations are formulated and solved for individual system constituents

$$\begin{array}{ccc}& & \partial \left[N{H}_{2}N{H}_2\right]/\partial t+\nabla .\left(-D_{NH2NH2}\nabla \left[N{H}_{2}N{H}_2\right]\right)\hfill \\ & & +k_f^{bulk}\left[N{H}_{2}N{H}_2\right]\left[C{O}_2\right]=k_b^{bulk}\left[P\right]\hfill \end{array}$$ (1)

$$\partial \left[C{O}_2\right]/\partial t+\nabla .\left(-D_{CO2}\nabla \left[C{O}_2\right]\right)+k_f^{bulk}\left[N{H}_{2}N{H}_2\right]\left[C{O}_2\right]=k_b^{bulk}\left[P\right]$$ (2)

$$\partial \left[P\right]/\partial t+\nabla .\left(-D_P\nabla \left[P\right]\right)+k_b^{bulk}\left[P\right]=k_f^{bulk}\left[N{H}_{2}N{H}_2\right]\left[C{O}_2\right]$$ (3)

$$\partial \left[P^{\prime }\right]/\partial t+\nabla .\left(-D_{P^{\prime }}\nabla \left[P^{\prime }\right]\right)=0$$ (4)

Values of $k_f^{bulk}$ and $k_b^{bulk}$ were calculated as delineated in the Results and Discussion section Eqs. (20) and ((21)). The D _NH2NH2 value of 2.37 × 10⁻⁵ cm² s⁻¹ was taken from Ref. [65] and the same value was assumed for $D_{P^{\prime }}$. The D _CO2 value of 1.97 × 10⁻⁵ cm² s⁻¹ was taken from Ref. [66] and the same value was used for D_P. Employing Stokes-Einstein equation and related concepts one can easily demonstrate that these two assumptions generate the relative error in the theoretically obtained Faradaic current of less than 10 %. Values of D and all other constants and parameters employed in numerical simulations are summarized in Table 2 .

Table 2.

Physical constants and parameters used in numerical simulations.

Constant/parameter and its symbol	Value and unit
standard redox potential E0	0.7 V
number of transferred electrons n	4
standard heterogeneous rate constant k0	1 × 10−2 m s−1 a
electrode polarization rate v	±0.025 V.s−1
electron transfer coefficient α	0.5
diffusion coefficient DCO2	1.97 × 10−9 m2 s−1 b
diffusion coefficient DP	1.97 × 10−9 m2 s−1
diffusion coefficient DNH2NH2	2.37 × 10−9 m2 s−1 c
diffusion coefficient $D_{P^{\prime }}$	2.37 × 10−9 m2 s−1
equilibrium constant K	2800 d
forward homogeneous rate constant $k_f^{bulk}$	1, 5, 7 and 10 × 109 M−1 s−1
ideal gas constant R	8.314 J mol−1 K−1
Faraday constant F	9.6485 × 104 C mol−1
absolute temperature T	298 K

[a] taken from Ref. [67]

[b] taken from Ref. [66]

[c] taken from Ref. [65]

[d] obtained experimentally in this study.

3.3. The thermodynamic model

It postulates that the bulk reaction between NH₂NH₂ and CO₂ is infinitely fast i.e. the state of the chemical equilibrium is reached in every time step of the simulation. This allows summing up Eqs. (2) and (3) obtaining the expression for $c_{CO2}=\left[C{O}_2\right]+\left[P\right]$

$$\partial c_{\mathit{CO}2}/\partial t+\nabla ·(-\overline{D_{\mathit{CO}2P}}\nabla c_{\mathit{CO}2}-c_{\mathit{CO}2}\nabla \overline{D_{\mathit{CO}2P}})=0$$ (5)

where $\overline{D_{CO2P}}=D_{CO2}{h}_{CO2}+D_P{h}_P$ is the effective diffusion coefficient that depends on molar fractions $h_{CO2}=\left[C{O}_2\right]/c_{CO2}$ and $h_P=\left[P\right]/c_{CO2}$. These fractions may be expressed as a function of K and [NH ₂ NH ₂]

$$\overline{D_{CO2P}}=D_{CO2}\left[1/\left(1+K\left[\mathrm{N}{\mathrm{H}}_2\mathrm{N}{\mathrm{H}}_2\right]\right)\right]+D_P\left[1-1/\left(1+K\left[\mathrm{N}{\mathrm{H}}_2\mathrm{N}{\mathrm{H}}_2\right]\right)\right]$$ (6)

In this way, the chemical equilibrium equation of the bulk reaction (Eq. (15)) is included in the continuity equation formulated for c _CO2 (Eq. (5)). In the chemical equilibrium state reaction rate terms in Eq. (1) cancel each other leading to the set of homogeneous expressions

$$\partial \left[\mathrm{N}{\mathrm{H}}_2\mathrm{N}{\mathrm{H}}_2\right]/\partial t+\nabla .\left(-D_{\text{NH}2\text{NH}2}\nabla \left[\mathrm{N}{\mathrm{H}}_2\mathrm{N}{\mathrm{H}}_2\right]\right)=0$$ (7)

$$\partial \left[P^{\prime }\right]/\partial t+\nabla .\left(-D_{P^{\prime }}\nabla \left[P^{\prime }\right]\right)=0$$ (8)

Altogether three continuity equations are formulated and solved for c _CO2 (Eq. (5) including Eq. (15)), [NH ₂ NH ₂] (Eq. (7)) and [P′] (Eq. (8)).

3.4. Boundary conditions

For both kinetic and thermodynamic computational model the left boundary of the domain represents the electrode (Fig. 2, red line). Here, non-zero flux Neumann boundary condition is formulated for continuity equations governing [NH ₂ NH ₂] and [P′]

$$\begin{array}{ccc}& & -n·(-D_{\mathit{NH}2\mathit{NH}2}\nabla \left[\mathrm{N}{\mathrm{H}}_2\mathrm{N}{\mathrm{H}}_2\right])=-k_f^{\mathit{el}}\left[\mathrm{N}{\mathrm{H}}_2\mathrm{N}{\mathrm{H}}_2\right]+k_b^{\mathit{el}}\left[P^{\text{’}}\right]\hfill \\ & & =-k^0\left[\mathrm{N}{\mathrm{H}}_2\mathrm{N}{\mathrm{H}}_2\right]\exp [(n-\alpha )F(E-E^0)/\mathit{RT}]\hfill \end{array}$$ (9)

$$\begin{array}{ccc}& & -n·(-D_{P^{\text{’}}}\nabla \left[P^{\text{’}}\right])=k_f^{\mathit{el}}\left[\mathrm{N}{\mathrm{H}}_2\mathrm{N}{\mathrm{H}}_2\right]-k_b^{\mathit{el}}\left[P^{\text{’}}\right]\hfill \\ & & =k^0\left[\mathrm{N}{\mathrm{H}}_2\mathrm{N}{\mathrm{H}}_2\right]\exp [(n-\alpha )F(E-E^0)/\mathit{RT}]\hfill \end{array}$$ (10)

A Butler-Volmer kinetic term was employed to express the rate constant $k_f^{el}$ [67]. We set $k_b^{el}=0$ as the NH₂NH₂ oxidation process is irreversible. On other three boundaries representing insulating walls (blue lines) a zero flux Neumann boundary condition was applied. As clearly demonstrated by cyclic voltammetric experiments (see Results and Discussion for details), CO₂ and P are both electrochemically inactive in the electrode potential region relevant to this work. Therefore, a zero flux Neumann boundary condition was set to all four boundaries in equations governing [CO₂] and [P].

The initial concentration value of the hydrazine oxidation product [P′] was uniformly set to zero in the whole domain. Initial values of [NH₂NH₂], [CO₂] and [P] were uniformly set to values calculated respecting the equilibrium condition $K/c^0=\left[P\right]/\left[\mathrm{N}{\mathrm{H}}_2\mathrm{N}{\mathrm{H}}_2\right]\left[\mathrm{C}{\mathrm{O}}_2\right]$ (taking experimentally determined K value) and satisfying values of the total concentration of NH₂NH₂ and CO₂ (expressed as $\left[\mathrm{N}{\mathrm{H}}_2\mathrm{N}{\mathrm{H}}_2\right]+\left[P\right]$ and $\left[\mathrm{C}{\mathrm{O}}_2\right]+\left[P\right]$) reconstructing initially introduced amounts of reactants in experiments.

4. Results and discussion

4.1. Electrochemical measurements

Functional electrochemical devices require electrodes with high and well-defined area of the active surface and minimal kinetic hindrance for the charge transfer between the electrode and the electrolyte. In our recent publications [40–42] we demonstrated that properly activated electrodes manufactured from conductive composites by FDM 3DP may approach charge transfer characteristics typical for conventionally prepared metallic or carbon electrodes. We apply the printing-activation protocol developed for the PLA-CNT composite [41] to manufacture electrodes that are further integrated to the electrochemical platform devised in this work (see Experimental Section and Supplementary Information for details).

Charge transfer characteristics of such manufactured PLA-CNT electrodes were inspected by measuring cyclic voltammetry response in the aqueous electrolyte containing Ru(acac)₃ as the electroactive probe as discussed in the Experimental Section. A detailed description of data analysis has been reported previously [41]. The black curve in Fig. 3 shows a characteristic cyclic voltammogram of the activated PLA-CNT electrode prepared in this work. Both positive and negative potential scans show a well-developed Faradaic feature. The separation between the cathodic (-550 mV) and anodic (-470 mV) peak is 80 mV indicating that the 3DP electrode causes almost no kinetic hindrance for the interfacial charge transfer. The comparison of the cathodic Faradaic peak current value with that calculated theoretically from the Randles-Sevcik equation for given conditions reveals that 91 % of the electrode/electrolyte interface is electrochemically active. Importantly, peak-to-peak separation and relative area of electrochemically active electrode surface are well comparable to those obtained previously in a conventional glass electrochemical cell (72 mV and 93 %) [41]. For comparison, the grey curve in Fig. 3 additionally depicts a cyclic voltammogram obtained for a non-activated PLA-CNT electrode. The peak-to-peak separation is 90 mV and the relative area of electrochemically active electrode surface amounts to 80 %. This indicates that the electrochemical activation improves both charge transfer characteristics of PLA-CNT electrodes.

Fig. 3.

Characteristic cyclic voltammogram of non-activated (grey) and activated (black) 3DP PLA-CNT electrode (geometrical area of 0.84 cm²) obtained in deaerated aqueous 0.1 M K₂SO₄ containing 5 × 10⁻⁴ M Ru(acac)₃ electroactive probe at 25 mV s⁻¹. Electrode potential values are stated against Ag/AgCl/3 M LiCl RE.

The devised electrochemical platform was further employed to investigate the oxidation of aqueous NH₂NH₂ coupled to its chemical reaction with CO₂. At the interface of aqueous electrolytes and platinum electrodes, NH₂NH₂ undergoes an irreversible oxidation [68]

$$\mathrm{N}{\mathrm{H}}_2\mathrm{N}{\mathrm{H}}_2\to {\mathrm{N}}_2+4{\mathrm{H}}^{+}+4{\mathrm{e}}^{-}$$ (11)

In our work, costly and environmentally problematic platinum has been replaced by electrodes manufactured by FDM 3DP from the composite filament based on inexpensive and biodegradable PLA polymer. First, the ability of 3DP electrodes to detect and quantify NH₂NH₂ was explored in argon-saturated 0.1 M K₂SO₄ supporting electrolyte in the absence of CO₂. Grey curve in Fig. 4 A shows a characteristic cyclic voltammogram obtained in the solution of 3.85 × 10⁻³ M NH₂NH₂. This concentration level is denoted further as c _{0, NH2NH2} and serves as a reference for subsequent analysis of NH₂NH₂/CO₂ reaction mixtures (vide infra) by cyclic voltammetry. The potential scan starting at -0.50 V (all potential values are stated against Ag/AgCl/3 M LiCl RE) shows a pronounced anodic feature at 0.45 V attributed to the oxidation of NH₂NH₂. Results of measurements performed in solutions with varied NH₂NH₂ concentration c _NH2NH2 indicate that the Faradaic anodic peak current i_FA is linearly proportional to c _NH2NH2 (Fig. 4B) and for c _{0, NH2NH2} it amounts to $i_{FA}^0$ = (1.8 ± 0.1) × 10⁻⁴ A. The observed proportionality between the concentration and the resulting Faradaic response confirms the applicability of activated 3DP PLA-CNT electrodes for the quantification of NH₂NH₂ in the aqueous solution by cyclic voltammetry.

Fig. 4.

Characteristic cyclic voltammogram of an activated 3DP PLA-CNT electrode in deaerated aqueous 0.1 M K₂SO₄ supporting electrolyte containing 3.85 × 10⁻³ M NH₂NH₂ prior to (grey) and upon (black) introducing 1.5 mL of saturated aqueous solution of (A) CO₂ and (C) Ar. (B) Faradaic anodic peak current i_FA as a function of hydrazine concentration c_NH2NH2 (without gaseous solutions introduced). (D) Response obtained in the supporting electrolyte (SE) prior to (grey) and upon (black) introducing 1.5 mL of saturated aqueous CO₂ solution (in the absence of NH₂NH₂). All cyclic voltammograms were obtained at electrodes with geometrical area of 0.84 cm² at 25 mV s⁻¹. Electrode potential values are stated against Ag/AgCl/3 M LiCl RE.

The response detected upon reversing the potential scan following the oxidation of NH₂NH₂ involves a small cathodic wave below -0.40 V. The same feature was observed also in the absence of hydrazine (grey curve in Fig. 4D). This control experiment additionally depicts a small anodic wave above +0.95 V. Results of experiments performed with varied WE potential limits (cyclic voltammograms not shown) confirm that these two baseline features are related. The anodic feature is ascribed to the oxidation of redox-active, presumably oxygen groups on the surface of CNTs and the cathodic feature to their subsequent reduction. Importantly, the absence of cathodic features attributable to NH₂NH₂ indicates that its oxidation is followed by a fast decomposition of electrogenerated products. We assume that nitrogen is the final product of the hydrazine oxidation following the reaction pathway reported for platinum electrodes (Eq. (11)) [68].

Values of i_FA obtained at varied c _NH2NH2 were further compared to theoretical predictions. For a chemically irreversible process (backward charge transfer reaction absent) realised in the voltammetric regime the theoretical Faradaic peak current is

$$i_F^{theor}=kn{\alpha }^{1/2}Ac{D}^{1/2}{v}^{1/2}$$ (12)

where k = 2.99 × 10⁵ A s mol⁻¹ V^−1/2, α is the charge transfer coefficient, A is the active electrode/electrolyte interface area, c is the bulk concentration of electroactive species, D is its diffusion coefficient, n is the number of transferred electrons and v is the electrode polarization rate [67]. For $c=c_{0,NH2NH2}$, values of v and A as used in experiments (see Experimental Section) and assuming that n = 4 (Eq. (11)), α = 0.5 and D = 2.37 × 10⁻⁵ cm² s⁻¹ (determined for NH₂NH₂ in its aqueous solution in Ref. [65]) the $i_F^{theor}$ value amounts to 2.1 × 10⁻³ A. Comparison of this value and the experimentally obtained $i_{FA}^0$ value gives the $i_{FA}^0/i_F^{theor}$ ratio of 8.6 % i.e. far below unity. We note that such low value cannot be rationalized by a low electrochemical activity of the electrode surface as the $i_{FC}^0/i_F^{theor}$ value of 91 % was found for the Ru(acac)₃ electroactive probe at the same electrode. To further explore this behaviour, we performed control measurements employing conventional glassy carbon (GC) electrodes (cyclic voltammograms not shown) and obtained the ratio of $i_{FA}^0/i_F^{theor}$ = 9.7 %. We additionally inspected the data reported for the electrochemical oxidation of NH₂NH₂ on GC electrodes in Ref. [65] and obtained $i_{FA}^0/i_F^{theor}$ value in the range of 10 – 20 %. Our analysis of the data reported for the NH₂NH₂ oxidation on platinum ultramicroelectrodes [68] leads to the $i_{FA}^0/i_F^{theor}$ value of 40 %. In summary, $i_{FA}^0/i_F^{theor}$values obtained for the NH₂NH₂ oxidation at different electrode surfaces are all well below unity and the observed anomaly is thus not specific to 3DP PLA-CNT electrodes. This observation suggests that a considerable portion of NH₂NH₂ molecules does not undergo the oxidation at the electrode/electrolyte interface, which is most likely caused by a deactivation or blocking of the electrode surface by reaction intermediates. Importantly for our current work, experimentally found linear proportionality between c _NH2NH2 and i_FA at activated 3DP PLA-CNT electrodes (Fig. 4B) allows their employment for the quantification of NH₂NH₂ in its reaction mixtures with CO₂.

Subsequently, aqueous electrolyte containing dissolved CO₂ was investigated by cyclic voltammetry as follows. Varied volumes of water saturated by CO₂ (0.5 to 2.0 mL, see Experimental Section for details) were introduced to deaerated 0.1 M K₂SO₄ supporting electrolyte in the absence of NH₂NH₂. The black curve in Fig. 4D displays a characteristic cyclic voltammogram obtained in the electrolyte containing 3.28 × 10⁻³ M CO₂ (corresponding to the addition of 1.5 mL of its saturated aqueous solution). It is indistinguishable from that obtained in the absence of CO₂ shown in grey. Identical cyclic voltammograms were obtained upon introducing 0.5, 1.0 and 2.0 mL of the saturated CO₂ solution to the supporting electrolyte (results not shown). This indicates that CO₂ is electrochemically inactive in the inspected potential region and its introduction does not affect the accessible potential window of the supporting electrolyte. Importantly, this finding simplifies further analysis of cyclic voltammograms obtained in the reaction mixtures of NH₂NH₂ and CO₂ and allows the equilibrium NH₂NH₂ concentration [NH₂NH₂] to be reliably determined.

The black curve in Fig. 4A displays a characteristic cyclic voltammogram obtained upon introducing 1.5 mL of water saturated by CO₂ to the NH₂NH₂ solution and homogenizing such obtained reaction mixture for 5 minutes. The Faradaic response of NH₂NH₂ obtained upon introducing CO₂ ($i_{FA}^{CO2}$) is lower compared to that before ($i_{FA}^0$) indicating that the reaction between the two species takes place. At this point we note that the baseline of the cyclic voltammetric response of the pure supporting electrolyte remained unchanged upon introducing CO₂ (Fig. 4D), which rules out the adsorption of CO₂ at the electrode/electrolyte interface. Therefore, observed changes of the hydrazine response cannot be caused by blockage of the electrode by adsorbed CO₂. Importantly, no new features appear suggesting that the reaction product (denoted as P further in the text) is electrochemically inactive in the inspected potential region and does not interfere with the response of NH₂NH₂. Most likely, the oxidation potential of P is beyond the electrode potential region inspected in this work.

According to Eq. (12) the Faradaic peak current is proportional to the bulk concentration of electroactive species and the area of electrochemically active electrode surface. The decrease of the NH₂NH₂ concentration is due to its consumption by the reaction with CO₂ as well as due to the dilution of reactants by their mixing. The active area may be reduced by the electrode deactivation caused by intermediates generated in the NH₂NH₂ oxidation process. Its contribution to the observed decrease of the Faradaic response was quantified by control experiments in which an argon gas was introduced to the system instead of CO₂.

Fig. 4C shows characteristic cyclic voltammograms obtained before (grey curve) and after (black curve) introducing 1.5 mL of water saturated by Ar to the solution of NH₂NH₂. Fig. 5 A summarizes values of $i_{FA}^{CO2}/i_{FA}^0$ and $i_{FA}^{Ar}/i_{FA}^0$ as a function of introduced gas volume. Values of $i_{FA}^{CO2}/i_{FA}^0$ substantially decrease with added volume, reaching the value below 0.20 for 2.0 mL, while values of $i_{FA}^{Ar}/i_{FA}^0$ are significantly higher, being in the interval between 0.74 and 0.80 for all values of added volume. For Ar, observed changes account for the dilution of the NH₂NH₂ solution and the deactivation of the electrode surface. It is worth reminding that the electrode deactivation is specific to the NH₂NH₂ oxidation process as no decrease of the Faradaic response was registered upon repeated electrode potential cycling in the solution of Ru(acac)₃ (data not shown).

Fig. 5.

(A) Experimentally determined $i_{FA}^{CO2}/i_{FA}^0$ (full circles) and $i_{FA}^{Ar}/i_{FA}^0$ (empty circles) values as a function of introduced gas volume. Shown data (average values and standard deviations) were obtained in two independent experimental campaigns. Theoretical values of $i_{FA}^{CO2}/i_{FA}^0$ obtained by finite-element-method numerical simulations assuming 1/1 reaction stoichiometry employing kinetic (coloured dotted lines) and thermodynamic (black dashed line, ∞) model. Numbers denote values of $k_f^{bulk}$ (1, 5, 7 and 10) in 10⁹ M⁻¹ s⁻¹ employed in the kinetic model. (B) Values of K obtained from experimentally obtained data shown in (A) for different V_CO2 values assuming 1/1 (circles) and 1/2 (squares) reaction stoichiometry.

4.2. Analysis of experimentally obtained results

Assuming the volume additivity the dilution contribution c _{Ar, NH2NH2}/c _{0, NH2NH2} may be expressed as $V_{NH2NH2}/\left(V_{NH2NH2}+V_{Ar}\right)$. The electrode deactivation contribution is given by the ratio of electrochemically active surface area values after and before the introduction of Ar, A_Ar/A ₀. Following Eq. (12), this ratio may be expressed as

$$A_{Ar}/A_0=i_{FA}^{Ar}/i_{FA}^0\times \left(V_{NH2NH2}+V_{Ar}\right)/V_{NH2NH2}$$ (13)

Values determined for all V_Ar values average to 〈A_Ar/A ₀〉 = 0.835 ± 0.016. The equilibrium NH₂NH₂ concentration [NH₂NH₂] in reaction mixtures was determined from the ratio of Faradaic response obtained after and before introducing CO₂, $i_{FA}^{CO2}/i_{FA}^0$ taking the effect of the electrode deactivation into account

$$\left[N{H}_{2}N{H}_2\right]=c_{0,\mathit{NH}2\mathit{NH}2}\times i_{\mathit{FA}}^{\mathit{CO}2}/i_{\mathit{FA}}^0\times {〈A_{\mathit{Ar}}/A_0〉}^{-1}$$ (14)

The equilibrium amount of substance n _NH2NH2 was determined as $\left[\mathrm{N}{\mathrm{H}}_2\mathrm{N}{\mathrm{H}}_2\right]\left(V_{NH2NH2}+V_{CO2}\right)$ and further employed to obtain equilibrium concentration values of CO₂, [CO₂] and the reaction product P, [P] as follows.

CO₂ dissolved in an aqueous environment forms a diprotic carbonic acid and NH₂NH₂ is a dibasic alkali. Therefore, the reaction between these two reactants may proceed following different stoichiometric ratios expressed by a generalized chemical reaction

$${\nu }_{CO2}\mathrm{C}{\mathrm{O}}_2+{\nu }_{NH2NH2}\mathrm{N}{\mathrm{H}}_2\mathrm{N}{\mathrm{H}}_2\rightleftharpoons \mathrm{P}$$ (15)

Stoichiometric ratios ν _CO2/ν _NH2NH2 of 1/1, 1/2 and 2/1 are considered as possible in this work. For the three scenarios the reaction extent ξ was calculated from the NH₂NH₂ consumption

$$\xi =\left(c_{0,NH2NH2}{V}_{NH2NH2}-n_{NH2NH2}\right)/{\nu }_{NH2NH2}$$ (16)

Such obtained ξ values were employed to determine equilibrium concentration values of CO₂ and P

$$\left[C{O}_2\right]=\left(c_{0,\text{CO}2}{V}_{CO2}-{\nu }_{CO2}\xi \right)/\left(V_{NH2NH2}+V_{CO2}\right)$$ (17)

$$\left[P\right]=\xi /(V_{\mathit{NH}2\mathit{NH}2}+V_{\mathit{CO}2})$$ (18)

where c _{0, CO2} is the concentration of CO₂ in its saturated aqueous solution (see Experimental section for more details). Equilibrium concentration values (referenced to c ⁰ = 1 M) were utilized to obtain the equilibrium constant K as

$$K=\left[P\right]{\left(c^0\right)}^{{\nu }_{CO2}+{\nu }_{NH2NH2}-1}/{\left[C{O}_2\right]}^{{\nu }_{CO2}}{\left[N{H}_{2}N{H}_2\right]}^{{\nu }_{NH2NH2}}$$ (19)

The calculation performed for ${\nu }_{CO2}/{\nu }_{NH2NH2}=2/1$ results in [CO₂] < 0 M ruling out this stoichiometric scenario. For ${\nu }_{CO2}/{\nu }_{NH2NH2}=1/2$ one obtains [CO₂] > 0 M and resulting values of K range from 6.0 × 10⁴ to 9.4 × 10⁵ (Fig. 5B, squares). A pronounced dependence of K values on V _CO2 expressed as the ratio of their extreme values (15.4) suggests that the reaction between CO₂ and NH₂NH₂ does not follow this stoichiometric scenario. For the 1/1 case obtained values of K range from 7.6 × 10² to 3.0 × 10³ (Fig. 5B, circles) showing significantly less pronounced trend (the ratio of extreme values is 3.9). Therefore, out of all proposed three scenarios the 1/1 stoichiometry is the most realistic one. The K value obtained for V_gas= 0.5 mL is most likely influenced by high scattering of $i_{FA}^{Ar}/i_{FA}^0$ values (Fig. 5A) and was thus left out from further analysis. Values obtained for higher V_gas values average to 〈K〉 = (2.8 ± 0.6) × 10³ (dashed line in Fig. 5B). We emphasize that depicted values of $i_{FA}^{Ar}/i_{FA}^0$, $i_{FA}^{CO2}/i_{FA}^0$ and K were obtained as an average of data acquired in two independent experimental campaigns performed on different days corroborating the validity of presented findings.

Lee et al. [53] have recently investigated the composition of reaction products in mixtures of CO₂ and NH₂NH₂ prepared by introducing gaseous CO₂ to concentrated (32 to 64 wt.%) aqueous solutions of NH₂NH₂. The chemical composition of reaction products was probed by nuclear magnetic resonance spectroscopy complemented by theoretical calculations. The authors arrived at the conclusion that the reaction stoichiometry changes depending on the ratio of initial amounts of substance $\chi =n_{0,CO2}/n_{0,NH2NH2}$. For 0 < χ < 0.30 hydrazinium mono-carbonate corresponding to the ${\nu }_{CO2}/{\nu }_{NH2NH2}=1/1$ stoichiometry was obtained as the main reaction product while for 0.30 < χ < 0.64 dihydrazinium hydrazinedicarbamate (${\nu }_{CO2}/{\nu }_{NH2NH2}=2/3$) was found as the major product. Motivated by these results we further calculated values of K from our own experimental data (which were obtained at χ values ranging from 0.3 to 1.2) considering the 2/3 stoichiometry. The values of K thus obtained range from 2.8 × 10¹⁰ to 4.5 × 10¹¹ (individual values not plotted). Their ratio amounts to 16.6 and is much higher than that determined for the 1/1 stoichiometry (3.9). This indicates that in diluted solutions of NH₂NH₂ as employed in our work the reaction between CO₂ and NH₂NH₂ follows the 1/1 reaction stoichiometry (i.e. forming hydrazine mono-carbamate) even at χ values higher than 0.3. The formation of hydrazine mono-carbamate in the reaction between hydrazine and carbon dioxide was also observed by Pollet et al. [69] who used this adduct as a precursor to form polymer gels. The average of K values obtained for the 1/1 reaction stoichiometry 〈K〉 = (2.8 ± 0.6) × 10³ is thus considered as the equilibrium constant value of the chemical reaction between NH₂NH₂ and CO₂ determined in our work. This value is three orders of magnitude higher than that reported for the reaction between CO₂ and ammonia [70] and one order of magnitude higher than that between CO₂ and monoethanolamine [71]. Because of that, we consider hydrazine as a suitable sorbent for the CO₂ capture.

4.3. Finite element method numerical simulations

The electrochemical oxidation of NH₂NH₂ coupled to its bulk chemical reaction with CO₂ was further modelled theoretically employing the finite element method numerical simulations in the environment of commercial software COMSOL Multiphysics®. [64] The developed computational model relies on formulating and solving a boundary value problem (a time-dependent continuity equation complemented by a set of boundary conditions) for concentration values of system constituents. Simulations were performed in a 2D rectangular computational domain (see Fig. 2 for details). The oxidation of NH₂NH₂ is described by non-zero flux Neumann boundary conditions at the boundary representing the electrode employing Butler-Volmer kinetic terms. Two individual computational approaches were developed and employed in this work, differing in the description of the chemical reaction between NH₂NH₂ and CO₂ in the electrolyte bulk. Both models consider 1/1 stoichiometry (reaction scheme is depicted in Fig. 6 ). The two approaches are further referred to as “kinetic“ and “thermodynamic“ model (for the formulation of respective boundary value problems see Calculation section).

In the kinetic model, four continuity equations are solved for individual system constituents – NH₂NH₂, CO₂, P and P´, the latter cumulatively representing products of the hydrazine oxidation (N₂ + 4H⁺). The bulk chemical equilibrium between NH₂NH₂ and CO₂ is taken into account by formulating reaction rate (source/sink) terms in respective continuity equations. The rate constant of the forward reaction $k_f^{bulk}$ is estimated assuming that the reaction rate is limited by the diffusion of reactants

$$k_f^{bulk}=8RT/3\eta$$ (20)

where η = 9.6 × 10⁻⁴ Pa s is the viscosity of the supporting electrolyte (aqueous 0.1 M K₂SO₄) [72]. At 298 K the $k_f^{bulk}$ value amounts to 7 × 10⁹ M⁻¹ s⁻¹. As the Eqn 20) is an approximation we let the value of $k_f^{bulk}$ to vary by ±40 % i.e. to attain values between 5 and 10 × 10⁹ M⁻¹ s⁻¹. To account for slower reaction kinetics we further consider the value of 1 × 10⁹ M⁻¹ s⁻¹. Values of $k_b^{bulk}$ were calculated from the relationship

$$K/c^0=k_f^{bulk}/k_b^{bulk}$$ (21)

employing experimentally obtained equilibrium constant value 〈K〉 = 2.8 × 10³ and $k_f^{bulk}$ values determined by Eq. (20).

In the thermodynamic model the number of solved continuity equations is reduced to three by summing up relationships formulated for CO₂ and P and evaluating the relative distribution of these two system constituents. The chemical equilibrium is taken into account by coupling such obtained distribution with the continuity equation expressed for NH₂NH₂, assuming the validity of the experimental 〈K〉 value. Reaction rate (source/sink) terms are completely omitted. The kinetic and the thermodynamic model are described in more detail in the Calculation Section.

In both models, initial values of concentration employed in simulations reconstruct the initial composition of experimentally prepared reaction mixtures (see Experimental Section for details). Both models were further complemented by time-dependent functions enabling the concentration profiles to be calculated at linearly varied values of the electrode potential E(t). The value of the electric current i flowing through the electrode/electrolyte interface was obtained by integrating the current density

$$j=Fn{D}_{NH2NH2}{\left|\partial \left[\mathrm{N}{\mathrm{H}}_2\mathrm{N}{\mathrm{H}}_2\right]/\partial x\right|}_{x=0}$$ (22)

over the boundary representing the electrode (see Fig. 2)

The solid black curve in Fig. 6 shows a simulated cyclic voltammogram of the system initially containing 3.85 × 10⁻³ M NH₂NH₂ in the entire computational domain i.e. in the absence of CO₂. Identical response was obtained by the kinetic and the thermodynamic approach. Importantly, the obtained Faradaic peak current value (2.0 × 10⁻³ A) is in a perfect agreement with the result of the closed-form solution (Eq. (12), 2.1 × 10⁻³ A). This verifies the correctness of computational models developed and employed in this work. The shape of the theoretical cyclic voltammogram corresponds to the experimentally obtained one (Fig. 4A and 4C, grey curve). Nonetheless, we note that the value of experimental (Fig. 4) and theoretical (Eq. (12) and Fig. 6) Faradaic peak current differs by one order of magnitude due to above-discussed low electroactivity of NH₂NH₂ molecules. Coloured dotted curves in Fig. 6 display the simulated cyclic voltammograms of the system with additionally introduced 3.28 × 10⁻³ M CO₂ at zero time, employing the kinetic model and $k_f^{bulk}$ values varied as discussed above. This CO₂ concentration corresponds to the addition of 1.5 mL of its saturated aqueous solution in experiments. The effect of the electrode deactivation was taken into account by multiplying theoretically obtained i(E) profiles by the experimentally determined value of 〈A_Ar/A ₀〉. Observed decrease of the Faradaic response reflects partial consumption of NH₂NH₂ due to its chemical reaction with CO₂. Theoretical values of $i_{FA}^{CO2}/i_{FA}^0$ obtained by the kinetic model at varied values of $k_f^{bulk}$ and V _CO2 are plotted as dotted curves in Fig. 5A. Results obtained for $k_f^{bulk}$ ranging between 5 and 10 × 10⁹ M⁻¹ s⁻¹ are in a very good agreement with experimentally obtained data-points (full black circles) while results calculated for 1 × 10⁹ M⁻¹ s⁻¹ are clearly off the experimental trend. This confirms that the rate of the forward reaction is limited by the diffusion of reactants. Figs. 6 and 5A further depict results of simulations employing the thermodynamic model (black dashed curves). Such obtained dependence of $i_{FA}^{CO2}/i_{FA}^0$on V _CO2 is in a perfect agreement with experimental data and data obtained employing the kinetic model considering the highest value of $k_f^{bulk}$. The agreement between experimentally obtained and theoretically predicted data corroborates that the reaction between NH₂NH₂ and CO₂ has 1/1 stoichiometry and further indicates that its rate is limited by the diffusion of these two reactions partners.

4.4. Econometric analysis and applicability of the presented platform

Results obtained in this work demonstrate that the newly designed and manufactured integrated 3DP platform based on the PA12 cell and activated PLA-CNT composite electrodes is capable of a comprehensive electrochemical analysis of an interfacial charge transfer reaction (oxidation of NH₂NH₂) coupled to a bulk chemical process (synthetic reaction between NH₂NH₂ and CO₂). Electrochemical experiments were complemented by finite-element-method numerical simulations employing two computational approaches utilizing the kinetic and the thermodynamic approach to formulate the bulk chemical reaction between NH₂NH₂ and CO₂. Obtained results indicate that the hydrazine oxidation is irreversible under given conditions, the bulk reaction between NH₂NH₂ and CO₂ has 1/1 reaction stoichiometry and the corresponding equilibrium constant amounts to 〈K〉 = (2.8 ± 0.6) × 10³.

The presented cyclic voltammetric analysis performed with activated 3DP PLA-CNT electrodes was additionally repeated with conventional platinum electrodes, providing very similar characteristics (results not shown). While platinum electrodes may be readily made by segmenting wires purchased from commercial sources, the manufacture of PLA-CNT electrodes requires the possession of the CAD software and the 3D printer. Currently, the unit price of the PLA-CNT composite is roughly 750 times lower than that of high purity platinum. The cell prototype devised and manufactured in this work consumed material (polyamide 12) worth of 1.8 Euro. This offers significant reduction of operational expenses in the electrode as well as cell manufacturing. Moreover, the combination of CAD modelling with bi-material 3D printing offers integrated electrochemical designs to be easily manufactured with multitude of sizes and shapes. Importantly, the presented cell design combines advanced functionalities that include separate WE, RE and CE compartments, inlet/outlet system to deoxygenate the electrolyte and an interface that enables air-tight introduction of reactants during the experiment. An outstanding performance of the platform was additionally verified by investigating characteristics of the electron transfer between activated 3DP PLA-CNT electrodes and aqueous electrolytes containing Ru(acac)₃ electroactive probe. Analysis of obtained data confirms that these electrodes have near-unity relative area of the electrochemically active surface and show almost no kinetic hindrance to the interfacial charge transfer. Electrochemical cells derived from the presented design are currently being applied to investigate electrochemical properties of biologically active compounds at polarized liquid-liquid interfaces [34,73,74]. In response to current healthcare crisis due to Covid 19, microchannels based on architectures developed in this work are being integrated within lab-on-a-chip microfluidic devices dedicated to the analysis of human viruses [75,76]. The presented design with integrated PLA-CNT electrodes will be further utilized as an electroanalytical platform to detect and quantify environmentally problematic nitrite [77] and nitrogen monoxide [78] in aqueous solutions. Additionally, an electrochemical reactor dedicated for the conversion of aqueous nitrite to ammonia as added-value chemical [79] is currently being developed. Computational models devised to simulate interfacial electron transfer processes and coupled bulk chemical reactions will be generalized by formulating electrodiffusion equations i.e. taking charge electroneutrality as well as ion migration in time-dependent electric fields into account [80,81].

5. Conclusions

In this work, we have devised and manufactured an integrated electrochemical platform employing computer assisted design combined with bi-material fused deposition modelling 3D printing. Polyamide PA12 filament was utilized to manufacture the electrochemical cell while polylactic acid-carbon nanotube composite was employed as the electrode material. The cell design involves separate compartments for reference and counter electrode connected to the central working electrode compartment by micro-holes. In-built gas inlet/outlet system enables the deoxygenation of the electrolyte and a specially designed interface allows reactants to be introduced to the cell in the course of the experiment. The functionality of the manufactured platform was verified by the inspection of the electron transfer between anodically activated electrodes and Ru(acac)₃ electroactive probe dissolved in the aqueous electrolyte by cyclic voltammetry. Obtained characteristics confirm near-unity relative area of electrochemically active electrode surface and virtually no kinetic hindrance for the electron transfer.

The manufactured platform was further applied to investigate the electrochemical oxidation of aqueous hydrazine coupled to its bulk reaction with carbon dioxide. Cyclic voltammograms of hydrazine solutions obtained in the absence of carbon dioxide confirm that hydrazine undergoes a totally irreversible electrooxidation similarly to the reported works employing platinum electrodes. Concentration-resolved experiments confirmed a linear proportionality between the Faradaic response of hydrazine and its bulk concentration. This allowed cyclic voltammetry to be used as a tool to determine the concentration of unreacted hydrazine in its reaction mixtures with carbon dioxide. Results obtained at systematically varied concentration ratios allowed the value of the equilibrium constant to be determined employing reaction models considering diverse stoichiometric scenarios. The analysis of obtained results leads to the conclusion that under experimental conditions as used in this work the hydrazine reacts with carbon dioxide with the stoichiometric ratio of 1/1 and the corresponding K value amounts to (2.8 ± 0.6) × 10³.

The finite element method numerical simulations were employed to reconstruct experimentally observed Faradaic currents of the hydrazine oxidation as a function of the amount of introduced carbon dioxide to the system. Two particular computational models were developed, differing in the definition of the homogeneous reaction between hydrazine and carbon dioxide (kinetic vs. thermodynamic model). The best agreement between computed Faradaic response and corresponding experimental data was obtained when the homogeneous reaction between carbon dioxide and hydrazine was considered as very fast i.e. limited by the diffusion of reactants. Therefore, the rate of the electrochemical hydrazine oxidation is limited by its diffusion towards the electrode as well as by kinetics of the electron transfer but not by the homogeneous reaction between hydrazine and carbon dioxide.

Importantly, our work is the first reported application of 3D printed electrodes and cells in the electrochemical analysis of an interfacial electron transfer process coupled to a bulk chemical reaction. We are convinced that the integrated 3D printed platform and computational approaches developed and presented in this work will find use in exploring systems in physical and environmental electrochemistry as well as in electroanalytical applications.

CRediT authorship contribution statement

João Giorgini Escobar: Conceptualization, Methodology, Validation, Formal analysis, Investigation. Eva Vaněčková: Conceptualization, Methodology, Validation, Formal analysis, Investigation. Štěpánka Nováková Lachmanová: Funding acquisition, Writing - review & editing. Federico Vivaldi: Conceptualization, Methodology, Writing - review & editing. Jan Heyda: Conceptualization, Methodology, Writing - review & editing. Jiří Kubišta: Conceptualization, Methodology, Writing - review & editing. Violetta Shestivska: Investigation, Writing - review & editing. Patrik Španěl: Conceptualization, Methodology, Writing - review & editing. Karolina Schwarzová-Pecková: Funding acquisition, Writing - review & editing. Jiří Rathouský: Funding acquisition, Writing - review & editing. Táňa Sebechlebská: Conceptualization, Methodology, Validation, Formal analysis, Investigation, Resources, Writing - original draft, Visualization, Supervision. Viliam Kolivoška: Conceptualization, Methodology, Validation, Formal analysis, Investigation, Resources, Writing - review & editing, Visualization, Supervision, Project administration, Funding acquisition.

Declaration of Competing Interest

The authors declare no conflict of interests.