Electroanalytical Overview: The Electroanalytical Sensing of Hydrazine

In this overview, we explore the electroanalytical sensing of the important chemical reagent hydrazine, highlighting the plethora of electrochemical sensing strategies utilised from the first reports in 1951 to the present day. It is observed that a large proportion of the work developing electrochemical sensors for hydrazine focus on the use of metallic nanoparticles and some other surface modifications, although we note that the advantages of such strategies is often not reported. The use of nanoparticle-modified electrodes to this end is explored thoroughly, indicating that they allow the same electrochemical response as that of a macroelectrode made of the same material, with clear cost advantages. It is recommended that significant studies exploring the surface coverage/number of nanoparticles are performed to optimise electroanalytical devices and ensure that thin-layer effects are not producing false observations through electrocatalysis. Development of these sensor platforms has begun to transition away from classical macroelectrodes, toward more mass producible supporting electrodes such as screen-printed and inkjet-printed electrodes. We suggest significant advances in this area are still to be found. The vast majority of developed electroanalytical sensors for hydrazine are tested in aqueous based environments, such as tap, river and industrial effluent waters. There is significant scope for development of hydrazine sensors for gaseous environments and biologically relevant samples such as blood, serum and urine; aiming to produce sensors for accurate occupational exposure monitoring. Finally, we suggest that the levels of publications with independent validation of hydrazine concentrations with other well-establish laboratory-based measurements is lacking. We believe that improving in these areas will lead to the development of significant commercial products for the electroanalytical detection of hydrazine.


Introduction: hydrazine
Hydrazine is an important chemical reagent, with the chemical formula N 2 H 4 , widely used in agricultural chemicals, air bags, pharmaceutical intermediates, photography chemicals, textile dyes, fuel for rockets, spacecraft and fuel cells and in boiler feed water systems where it acts a scavenger to remove traces of oxygen and reduces corrosion of metal pipes and fittings. 1, 2 The US National Institute for Occupational Safety and Health (NIOSH) report that the recommended exposure limit (REL) in air is 0.03 ppm (2 hrs) and the permissible exposure limit 8 hr total weight average (TWA) is 1 ppm, while the "Immediately Dangerous to Life or Health" (IDLH) is 50 ppm. 3 The U.S. Environmental Protection Agency has classified hydrazine as an irritant and Group B2, probable human carcinogen. 4 The Threshold limit value is reported to be no higher than 10 ppb. 5,6 Due to hydrazine's industrial significance and toxicological effects, its sensing is widely explored. Methods for the detection of hydrazine sensing include: Solid-phase spectrophotometry, 2 gas chromatography-mass spectrometry, 7 high-performance liquid chromatography-tandem mass spectrometry, 8 and matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. 9 Many sensors have been reported for the sensing of hydrazine, such as fluorescence, 10,11 surface-enhanced Raman spectroscopy, 12 chemiluminescence 13 and electrochemistry to name just a few. In this overview we focus specifically on providing a thorough overview of the endeavours dedicated to the electroanalytical sensing of hydrazine.
The electrochemical oxidation of hydrazine was studied as early as 1951 via polarography and oxide-coated platinum rotating disc electrodes. 14,15 This reported that the reaction in aqueous media yields nitrogen as a principal component and that the reaction proceeded more rapidly on an oxide-coated surface over that of a bare (clean) platinum surface. Bard explored the effect of electrode pre-treatment on the chronopotentiometric oxidation of hydrazine at a platinum electrode in aqueous solutions agreeing with Karp and Meites 15 that a freshly prepared layer of platinum oxide plays an active role. 16 Michlmayr and Sawyer explored the electrochemical oxidation of hydrazine, 1,1-dimethylhydrazine and 1,2-dimethylhydrazine in DMSO with chronopotentiometry, controlled-potential coulometry, and cyclic voltammetry at platinum electrodes, reporting that the overall reaction is a one-electron oxidation. 17 Wang and Cao 18 reported the first use of glassy carbon for the electrochemical oxidation of hydrazine, exploring the mechanism in both aqueous and non-aqueous media. Carbon based materials have the advantages of being economical, stable, having chemical inertness, give rise to relatively wide potential windows and low background current making them suitable for different types of electroanalysis, in-particular the sensing of hydrazine.
Compton et al. have recently provided a thorough overview of the electrochemical oxidation of hydrazine demonstrating that the phenomenological Butler-Volmer theory is not appropriate for interpreting the electrochemical process of hydrazine but rather reveals a strong potential dependence on the anodic transfer coefficient, consistent with the symmetric Marcus-Hush theory. 19,20 Hydrazine in aqueous media has pK a1 and pK a2 values of 8.0 and -1.0 at 298K respectively, which corresponds to the following chemical equilibria: 19 Figure 1A helpfully shows the speciation of hydrazine in aqueous media as a function of pH, which is useful for electroanalysts to know the species they are measuring within different pH aqueous media.
The electrochemical oxidation of hydrazine at a glassy carbon electrode in aqueous media is shown in Figure 1B along with fitting of the voltammetric profiles with DIGISIM, a commercial electrochemical simulation software. An excellent fit between theory and experiment is evident ( Figure 1B) and the electrochemical mechanism is described as: 19 19,20 has elegantly shown that hydrazine is only electroactive in its unprotonated form, N 2 H 4 whereas the protonated species N 2 H 5 + is electro-inactive.
Overall, the electrochemical oxidation of hydrazine occurs via 4-electron process with the second step (see mechanism above) rate-determining step (rds). The standard rate constant of the rds step is 4.5 × 10 -5 cm s -1 ; the second-order rate constant for the protonation of hydrazine is 1.1 × 10 5 M -1 s -1 . The approximate estimation of the associated standard heterogeneous rate constant, k 0 , for the rate-determining step has been reported to be 4.5 (± 1.0) × 10 -5 cm s -1 . 19, 20 As can be seen from Table 1, a diverse range of metal and metal oxide in the form of nano particles, spheres, rods, cubes and more, are used for the electroanalytical sensing of hydrazine.
The most common is the use of nanoparticles to modify an underlying supporting electrode, such as carbon electrodes, which allows electrical "wiring" of the nanoparticle. From an electroanalytical point of view, the electrochemical response at a bare glassy carbon is reported to exhibit slow electrode kinetics with a voltammetric signal that occurs at high overpotentials with a relatively small electroanalytical signal. When the electrode is modified with the chosen nanoparticles, the electrochemical oxidation of hydrazine signal occurs at lower overpotentials compared to that seen at the bare electrode and with a larger electroanalytical signal. 24,25 The reason for the improvements is based around arguments on the nanoparticles increased surface area, improved electrode kinetics, changed electronic structure and adsorption behaviour. 26 Using such an approach provides advantages over that of say carbon based materials but economics need to be considered. Generally, such as response is routinely reported to be "electrocatalytic".
Annalakshmi et al. 27 reported the sensing of hydrazine using trimetallic NiFeCo nanospheres supported upon a glassy carbon electrode, where the nanospheres were synthesized through a one-pot, facile hydrothermal methodology; see Figure 2. The nanosphere modified electrode was reported to be an excellent electrocatalyst, where the bare glassy carbon electrode was reported to exhibit a high oxidation overpotential (+0.83 V), and no noticeable peak was observed compared to the nanosphere decorated glassy carbon electrode which exhibited a voltammetric signal at +0.56 V. The authors state that the modified electrode gives rise to an improved response, which might arise from its low impedance behaviour, nanospheres-like architecture and synergic effect among the metallic nanoparticles. 27 The electroanalytical sensor was able to detect hydrazine over a large concentration range (0.020 -3080 μM) with a very low LOD of 6.4 nM. The practical applicability of the sensor was successfully validated in spiked river, lake, tap, and sewage water with satisfactory recoveries. 27 Generally researchers overlook exploring a coverage study, that is, the effect of increasing the number of nanoparticles on the electrode surface against the electrochemical/electroanalytical response.
One must not forget false electrocatalysis can result for thick porous layers of nanoparticles where the electrochemical response changes from that, at low monolayer coverage, of semiinfinite diffusion to that of thin layer diffusion for large coverages/porous layers. 28 If we consider that the case of nanoparticle modified electrodes have been extensively developed to sense hydrazine, let's again ask why would you adopt this strategy? The answer lies in not only the physical and chemical properties of nanoparticles differing greatly from that of the bulk material, but also the geometric configuration of the constructed electrochemical platform. If we consider an isolated nanoparticle for a simple electrode reaction (e.g.: + -), a diffusion-limited current (at suitably slow scan rates) is given by: equations is due to a 'shielding effect', produced as a result of the underling supporting electrode that reduces the limiting current. 29 The above cases are for diffusion-limited cases, in essence when the voltammetric experiment is conducted at suitably slow scan-rates. When the scan rate of the voltammetric experiment is increased, the voltammetry deviates from that of the expected steady-state to that of a peak-shaped response. 30 However, in reality it is very rare for electroanalysts to conduct electroanalytical measurements on a single sphere. In most, if not all cases, electroanalysts will decorate a chosen electrode surface with nanoparticles; for example, see Table 1 for the various endeavours using a diverse range of nanoparticle compositions.
The voltammetry at nanoparticle-modified electrodes has been elegantly reported by Compton et al. 30 , where the mass transport is different compared to that of a bulk electrode. In electroanalytical experiments, the modification of an electrode surface with nanoparticles results in a random array, that is, an assembly of nanoparticles randomly distributed over the supporting electrode surface. In this case, the voltammetry is different to that observed for a single hemi-sphere or sphere (see above) due to the difference in mass transport. Figure 3 shows simulations for a single reversible electron transfer process at a spherical nanoparticle array on an inert surface. Figure 3A shows that four distinct categories arise, dependent upon the applied experimental voltage scan rates, which is represented in a dimensionless form: , where T is the temperature and R is the universal gas constant. As the voltammetric scan rate changes, so does the diffusion layer. 28,30 In the case of category 1, the diffusion layers are small, corresponding to fast voltammetric scan rates, and the nanoparticles are diffusionally independent. Mass transport in this regime is linear and the cyclic voltammetric profile is the classic peak shape and the peak current (I p ) is governed by the Randles-Ševčík equation, i.e. , while the chronoamperometric ∝ response is governed by the Cottrell equation, i.e.
. In the case of category 2 (see ∝ 1/ figure 3A), the voltammetric scan rate is slowed and the diffusion layers become larger where diffusional independence is still observed. In this category, diffusion is hemispherical (or convergent) and the resultant cyclic voltammetric responses are steady-state shaped (rather than peak shaped). In category 3, the scan rate is further slowed, and we see that the diffusional layers are no longer independent, but rather overlap. In this case, the cyclic voltammetric response is peak shaped but the absolute current is smaller than theoretically expected. Last, category 4, the diffusion layers heavily overlap, and mass transport is linear over the entire nanoparticle array. The observed voltammetric response is the same as that observed in category 1. Note that the reason for the nanoparticle size dependence of the peak potential is that this reflects a switchover between rate-determining electrode kinetics, which control the current before the peak potential, to rate determining mass transport (diffusion) after the peak potential.
Next, if the voltammetric scan rate is fixed, what is the effect of changing the surface coverage?
As shown in figure 3B, linear sweep voltammograms are shown for a fixed scan rate where the y axis is presented in terms of the dimensionless current per particle: , N is the number of spherical particles present on the surface. The x-axis is presented in terms of a dimensionless potential, . In figure 3B, the effect of the surface = / ( -0 ) coverage, upon the voltammetric response is shown from 0.1 -10 -4 , where and A is the area of the supporting electrode. Figure 3B elegantly shows that at low nanoparticle coverage ( = 10 -4 ), the diffusion zones at each nanoparticle are diffusionally independent and Θ a steady-state response is observed, falling within category 1 or 2. As the nanoparticle coverage is increased, the distance between each nanoparticle decreases and passes through category 3 finally arriving at category 4; the change in the voltammetric wave shape is easily visualised ( Figure 3). An important observation is the dimensionless current decreases since the region of the solution available to each nanoparticle is reduced. 30 The above demonstrates that the voltammetric shape of a nanoparticle array will depend upon the scan rate and surface coverage of the nanoparticles. 30 One interesting aspect most pertinent to electroanalysts is that when in category 4, where heavy diffusional overlap occurs at a nanoparticle random array, the current response (the electroanalytical response) is similar to that obtained if one had used a complete electrode (e.g. a film or a solid electrode) of the same material. This unique property is extremely useful in electroanalysis as in the category 4 region, a nanoparticle array yields a similar amount of electrolytic depletion to a macroelectrode of the same total area. Consequently, minimal amounts of expensive catalyst, such as platinum or gold, can be used to offer a maximal electrochemical response over that of a solid electrode comprised of platinum or gold with significant cost savings. This critical issue is generally overlooked/not reported when nanoparticle modified electrodes are utilised, particularly to the sensing of hydrazine.
To this end, Batchelor-McAuley and co-workers 31 have explored the random distribution of palladium nanoparticles supported on a boron-doped diamond (BDD) electrode with that of a palladium plated BDD microelectrode array comprising 362 palladium 25 µm diameter microdiscs. In comparison of the two electrodes, the palladium nanoparticle decorated BDD electrode exhibits low micro-molar detection and a highly linear response toward hydrazine (see Table 1). The authors suggest that it is likely that the palladium nanoparticles are also acting as an array of microelectrodes with AFM images of the modified electrode revealing nanoparticles which are close together which effectively makes them act as single, larger particles. 31 One important aspect, is that both the array and the nanoparticle assemble provide lower detection limits and highly linear responses compared to that of palladium macroelectrodes. The beneficial cost implications of using palladium nano-or micro-particles in sensors compared to a palladium macroelectrode are evident and gives a substantial reason to pursue nanoparticle-decorated electrodes.
Rather than directly modify an electrode surface, as new 2D nanomaterials have appeared with reported beneficial properties, such as large surface areas and improved conductivity, these have utilised to support various nanoparticle compositions. 26  gave rise to the highest electroanalytical sensitivity and lowest limit of detection. 32 This is attributed by the authors to be due to a higher electrochemical active surface area-to-volume ratio as well as to quantum confinement, for which the valence band centre of the PdNPs was shifted ca. +0.2 eV vs Pd bulk leading to faster charge transfer. 32 Using the 3.7 nm diameter PdNPs -rGO, a very low limit of detection of about 7 nM (at a rotation speed of 6000 rpm) was shown to be possible with a wide linear range of 0.04 -200µM. The sensor was shown to be highly selective to hydrazine without interference from uric acid, glucose, ammonia, caffeine, methylamine, ethylenediamine, hydroxylamine, n-butylamine, adenosine, cytosine, guanine, thymine, and l-arginine. The PdNPs -rGO based hydrazine sensor was shown to successfully determine hydrazine in spiked wastewater samples. Another approach has utilised reduced graphene oxide which is modified with cobalt oxide nanocubes@gold (rGO-Co 3 O 4 @Au) nanocomposites, 33 fabricated using a one-pot hydrothermal synthesis ( Figure 5A).
Using amperometry, the detection of hydrazine was shown to be viable over the range of 10 -620 μM with a LOD of 0.44 μM. The effect of interferents were explored (NO 3 − , SO 4 2− , Cl − , Ag + , Na + , K + , ethanol, 4-nitrophenol, ascorbic acid and glucose) which showed no detrimental effect upon the sensor. The sensing of hydrazine in spiked sea, lake and river water was shown to be viable. Following a similar approach, a gold tetra phenyl porphyrin (AuTPP) modified reduced graphene oxide nanocomposite film modified glassy carbon electrode (GCE) was fabricated and explored towards hydrazine sensing (see figure 5B). Initially graphene oxide (GO) was prepared from graphite by a modified Hummer's method, which was then mixed with the Au-TPP and drop cast upon a GCE surface and electrochemically reduced to prepare the final AuTPP -rGO/GCE sensor. Using amperometry, a linear 20 nM to 198 μM was shown to be possible with a very low LOD of 3 nM. The sensor was found to selectively detect hydrazine in the presence of 500 fold excess concentrations of a range of interfering ions and was shown to be viable for hydrazine sensing in spiked ground, rain and river water samples. 34 Zhang et al. utilised Au nanoparticles N-doped porous carbon anchored upon reduced graphene oxide nanosheets, supported upon a GCE. It was fabricated via a confinement synthetic process in the frame structure of zeolitic imidazolate framework-67 (ZIF-67). 35 The authors explore the role of the N-doped porous carbon and found that its incorporation provided a framework to immobilize Au nanoparticles. This prevents shifting and agglomeration, improving the wettability of rGO and therefore, avoiding irreversible restacking due to π−π interactions of rGO layers, which can reduce the performances of rGO-based support. 35 Figure 6 shows a schematic illustration of the sensing mechanism (note the GCE is absent) which was explored toward sensing hydrazine in both aqueous and gaseous environments. In the aqueous solutions a linear detection from 0.05 to 1.00 μM was shown to be viable with an LOD of 9.6 nM, with the authors validating the sensor towards the sensing of hydrazine in spiked drinking, river and  figure 7B, which proceeds via an EC cat mechanism with a catalytic rate constant (k cat ) found to be 8.1 (± 0.1)x10 4 M -1 s −1 , confirming the AI 3+ hematein complex has a high electrocatalytic activity. The sensor was shown to successfully determine hydrazine in spiked drinking and river water with good recoveries (96 and 90% respectively). Screen-printing and inkjet printing appear to be useful fabrication approaches for producing next generation disposable sensors that have scales of economy, allowing low-cost hydrazine sensors to be realised; future work should be directed to this endeavour.

Conclusions and outlook
We have overviewed the electroanalytical sensors that have been reported for the sensing of