Online Monitoring of Electrochemical Carbon Corrosion in Alkaline Electrolytes by Differential Electrochemical Mass Spectrometry

Abstract Carbon corrosion at high anodic potentials is a major source of instability, especially in acidic electrolytes and impairs the long‐term functionality of electrodes. In‐depth investigation of carbon corrosion in alkaline environment by means of differential electrochemical mass spectrometry (DEMS) is prevented by the conversion of CO2 into CO3 2−. We report the adaptation of a DEMS system for online CO2 detection as the product of carbon corrosion in alkaline electrolytes. A new cell design allows for in situ acidification of the electrolyte to release initially dissolved CO3 2− as CO2 in front of the DEMS membrane and its subsequent detection by mass spectrometry. DEMS studies of a carbon‐supported nickel boride (NixB/C) catalyst and Vulcan XC 72 at high anodic potentials suggest protection of carbon in the presence of highly active oxygen evolution electrocatalysts. Most importantly, carbon corrosion is decreased in alkaline solution.

Abstract: Carbon corrosion at high anodic potentials is am ajor source of instability,e specially in acidic electrolytes and impairs the long-term functionality of electrodes.In-depth investigation of carbon corrosion in alkaline environment by means of differential electrochemical mass spectrometry (DEMS) is prevented by the conversion of CO 2 into CO 3 2À .
We report the adaptation of aD EMS system for online CO 2 detection as the product of carbon corrosion in alkaline electrolytes.Anew cell design allows for in situ acidification of the electrolyte to release initially dissolved CO 3 2À as CO 2 in front of the DEMS membrane and its subsequent detection by mass spectrometry.D EMS studies of ac arbon-supported nickel boride (Ni x B/C) catalyst and Vulcan XC 72 at high anodic potentials suggest protection of carbon in the presence of highly active oxygen evolution electrocatalysts.M ost importantly,carbon corrosion is decreased in alkaline solution.
Carbon materials in their various allotropic forms,a sb ulk materials (e.g.g raphite and glassy carbon), powders (e.g. carbon nanotubes and graphene), carbon fibers,c arbon foils and pastes,a mong others,a re extensively employed in electrochemical technologies.I ne lectrocatalysis,c arbon is used as support for dispersion of precious-metal catalyst nanoparticles to enhance their utilization, and as aconductive matrix to boost charge transfer of inherently low-conductivity catalysts. [1] Recent developments of so-called heteroatomdoped carbon catalysts or catalyst supports,f or example, nitrogen-, boron-, or phosphorus-doped carbon, has revealed interesting new applications of such carbon-based materials as noble-metal-free catalysts for the oxygen reduction reaction (ORR), [2] the oxygen evolution reaction (OER), [3] and the CO 2 reduction reaction (CO 2 RR), [4] to name but afew.A core concern of using glassy carbon electrodes [5] and carbon as an electrode material, catalyst, or catalyst support in electrochemical systems in general relates to its susceptibility to corrode under oxidizing conditions [6,7] through dissolution, gasification, or exfoliation under formation of corrosion products that affect the carbon properties.I nt he past three decades,s tudies on carbon corrosion predominantly focused on acidic electrolytes, [8][9][10] mainly because of the broad research interest in proton exchange membrane fuel cells (PEMFCs) and electrolyzers.C arbon corrosion was intensively studied using various analytical techniques,i ncluding Raman spectroscopy, [11] FT-IR spectroscopy, [12] X-ray diffractometry, [11] X-ray photoelectron spectroscopy, [13] and identical location transmission electron microscopy. [14] Carbon becomes thermodynamically unstable at potentials higher than its equilibrium potential of 0.207 Vv ersus reversible hydrogen electrode (RHE). [15] Thec onsequences of carbon corrosion typically include ad ecrease of the electrochemically active surface area (ECSA) as well as the conductivity. Electrochemical oxidation of carbon leads to the formation of both soluble and insoluble organic and inorganic products in the electrolyte.T ypical products of carbon electrooxidation include CO and CO 2 , [6,16,17] [Eqs. (1) and (2)).
CO formation is thermodynamically hindered due to its high standard potential, while CO oxidation to CO 2 is favored [Eq. (3)] with astandard potential of E 0 = À0.103 V SHE .
In contrast to soluble inorganic and insoluble organic products of carbon oxidation, such as graphite oxides and surface oxygen functional groups (C = O, C À O À Ca nd O À C = O), [18] soluble organic products,s uch as mellitic and humic acids,a re formed at very low concentrations and therefore considered insignificant. [19] Studies show that carbon materials with ah igh degree of graphitization, such as carbon nanotubes and graphene,e xhibit comparatively superior corrosion resistance as compared to amorphous carbon. [9,10,20] Suppressing carbon corrosion in electrochemical applications is therefore of crucial importance.C arbon oxidation, as well as the underlying corrosion mechanisms has been widely investigated in acidic electrolytes. [10,17,21] In contrast, studies of carbon corrosion in alkaline electrolytes has scarcely been reported, except for af ew early reports dating back to the 1980s. [19,22] In most OER measurements catalyzed by carbon or carbon-supported catalysts,t he current measured during potentiostatic polarization is often exclusively ascribed to O 2 evolution with carbon oxidation being presumed or shown to be negligible based on Faradaic efficiencym easurements. However,a tt he anodic conditions of O 2 evolution on highly active OER catalysts,O 2 evolution and carbon oxidation are expected to proceed concurrently [23] as depicted in Scheme 1a.T he OER occurs from apurely thermodynamic point of view at potentials higher than 1.23 Vversus RHE that are far above the thermodynamic equilibrium potential of 0.207 Vv ersus RHE [17] of carbon oxidation. This implies that the Faradaic current measured during potentiostatic O 2 evolution is supposedly asum of the OER (i OER )and carbon oxidation (i C,Ox ). To understand the corrosion of carbon and its implication on catalyst stability and long-term system performance,i ti si mportant to decouple the current measured during the OER into the contributions i OER and i C,Ox .
Differential electrochemical mass spectrometry (DEMS) is ap owerful technique that can be used to probe carbon corrosion and its mechanisms by direct detection of gaseous and volatile corrosion products dissolved in the electrolyte. [24] DEMS has been used to study carbon corrosion in acidic electrolytes by direct detection of CO 2 as ac orrosion marker. [8][9][10] On the other hand, direct detection of CO 2 as am arker for carbon corrosion in alkaline electrolytes is challenging because of CO 2 dissolution in high pH electrolytes under formation of carbonate according to Equation (4). [25] Consequently,the corrosion of carbon-based catalysts and catalyst supports at alkaline conditions has hardly been addressed despite the broad scope of applications under these conditions.R ecently,Y ie tal. investigated the electrochemical stability of glassy carbon under anodic conditions in acidic and alkaline electrolyte by means of spectroscopic methods. [13] They proposed aradical decomposition mechanism for glassy carbon at high anodic polarization in alkaline media. Edges of small graphitic domains are oxidized until they become hydrophilic and dissolve in the electrolyte. [13] To achieve online detection of electrochemical carbon corrosion in alkaline electrolytes,wedesigned aDEMS cell [26] ( Figure S1 in the Supporting Information) which includes an additional channel allowing for acidification of the electrolyte without changing the electrochemical conditions at the working electrode.A ccording to the Bjerrum plot for CO 2 , acidifying the electrolyte below ap Ho f4s hifts the CO 2 / CO 3 2À equilibrium completely towards CO 2 . [27] Thus,a cidifying the electrolyte will cause release of CO 2 initially dissolved as carbonate and its subsequent detection by mass spectrometry.Aschematic representation of the proposed processes is depicted in Scheme 1b.F or further information about the DEMS measurements see Supporting Information.
Thep roof of concept was carried out by measuring the total current density (j F ,F igure 1f irst row) response during potential step polarization in 0.1m KOH(pH 12.9) of agraphite electrode,juxtaposed with the corresponding subsequently measured mass spectrometric ion currents of O 2 (i 32 )and CO 2 (i 44 ), without (Figure 1s econd and third row) and with (Figure 1f ourth row) acidification (0.15 m H 2 SO 4 ,p H0.7) of the electrolyte in front of the membrane of the DEMS system indicates that the total measured j F is essentially the same for the two independent measurements.The independence of the electrochemical response is due to the cell design, in which acid injection and electrochemistry are spatially separated avoiding changes of the environment in front of the electrode. Evidently,the j F represents asum of O 2 formation and carbon oxidation (Figure 1f irst row). Without introduction of the acid, only the O 2 ion current was detectable in the mass spectrometer.T he fact that CO 2 could not be directly detected during anodic polarization of the graphite electrode in 0.1m KOHu nderlines the presence of the reaction in Equation (4). Note that the O 2 MS signal was always recorded without acidification in order to avoid signal changes caused by the injected acid. Thel ack of studies on carbon corrosion in alkaline media leads to the assumption that carbon corrosion in alkaline environments occurs similarly and at comparable rates as in acidic electrolytes.A ccording to Nernst equation, the equilibrium potential of areaction shifts with the pH when either protons or hydroxide ions are involved in the reaction. Thus,f or both the electrochemical Scheme 1. Schematic of carbon oxidation during OER in alkaline electrolytes. The current is the sum of i OER and i C,Ox and it is supposed that ahighly active OER catalyst is protecting the carbon support against corrosion (a). Concept of CO 2 detection as amarker for electrochemical carbon corrosion in alkaline electrolytes, for example during the OER. The formed CO 2 is converted into CO 3 2À which is again liberated as CO 2 by injecting an acid and in turn collected through aT eflon membrane at the inlet of the MS (b). carbon corrosion and the OER, the equilibrium potential is pH dependent. However, the kinetics of the reactions and their dependence on electrolyte pH might differ substantially. Chronopotentiometric (CP) measurements of ag raphite electrode at an applied current density of 5.5 mA cm À2 employing electrolytes with ap Ho f1and 13 reveal ac lear dependence of both the obtained potential and the measured CO 2 ion currents on the electrolyte pH value (Figure 2a and Figure S4). Note that acid was injected for CO 2 release during all measurements.A se xpected, both acidic and alkaline environments afforded similar potentials versus RHE at the applied current density (Figure 2a), however, the measured ion charge for CO 2 (Q 44 )o bserved by integrating the whole ion current for CO 2 which was produced during the measurement ( Figure S4) varied substantially (Figure 2b). The Q 44 decreased from pH 1to1 3.
Comparing electrochemical carbon corrosion in electrolytes of pH 1a nd 13, CO 2 was barely detectable at high pH values,w ith al ow CO 2 ion charge of 18.5 AE 6.6 pC,w hile it increased substantially to 104.9 AE 1.8 pC in acidic electrolyte. This situation can be explained by the fact that during CP measurements involving competing reactions (here carbon oxidation and OER), the measured potential is closest to that of the predominant reaction. Having this in mind, the results point towards ac hange in the relative contributions of the OER and carbon oxidation with increasing pH value. Reconsidering Scheme 1a,i na na cidic medium the red curve would be shifted to more cathodic potential with respect to the curve of the OER and vice versa when the reaction environment is alkaline.S ince the thermodynamics for both reactions in relation to the equilibrium potentials are similarly influenced by the pH value and should be invariant when referenced versus the RHE, the difference in CO 2 detection and hence carbon oxidation, is related to ap Hdependent change in the reaction kinetics.O bviously,t he OER is kinetically favored in alkaline pH while carbon oxidation proceeds at higher rates in acidic environments.The results indicate aclear difference between carbon oxidation in alkaline and acidic conditions,indicating that adirect extrapolation of carbon corrosion from acidic conditions to alkaline environment might be misleading.
Clearly,d eposition of an OER electrocatalyst on the carbonaceous electrode surface leads to an enhancement of the OER kinetics by decreasing the overpotential for the OER. Thus,afurther shift in the relative contributions of the OER and carbon oxidation in favor of the OER is expected in alkaline electrolytes.I tc an therefore be supposed that when as ufficient coverage of an OER active catalyst is homogeneously dispersed on carbon, the tendency for carbon to undergo oxidation will be suppressed kinetically provided that the density of OER active sites is not limiting. Tw omodel systems,Vulcan XC 72 carbon (denoted as Vulcan) and nickel boride (Ni x B), aw ell-established active catalyst for the OER, [28] supported on Vulcan XC 72 carbon (denoted as Ni x B/C-10 for am ixture with 10 wt %N i x B) were employed to support the aforementioned presumption.
Chronopotentiometric measurements at various current densities reveal agradual increase of the recorded ion current of CO 2 ,w hich was normalized by the mass of Vulcan on the electrode,w ith increasing applied current density when Vulcan is used as catalyst (Figure 3b lack curves,f or nonnormalized ion currents see Figure S5). Adding 10 wt %Ni x B  to Vulcan leads to the disappearance of the mass signal of CO 2 regardless of the applied current (Figure 3blue curves), hence carbon oxidation is presumably suppressed by the enhanced kinetics of the OER.
Additionally,the potential afforded to drive the reactions at the necessary rates to fulfill the applied currents increased substantially with increasing current density in the case of pure Vulcan reaching values higher than 3V versus RHE. Thep resence of several plateaus in the potential time transient of the Vulcan sample indicates that various reactions at different potentials have to proceed to provide the applied currents,w hile for Ni x B/C-10 only slight changes in the potential occur. By determining Q 44 from the produced CO 2 -Signal ( Figure S6) from the chronopotentiometric measurements,q uantitative analysis of the CO 2 MS signal (for calibration and carbon loss calculations see Supporting Information) reveals aF aradaic efficiencyf or Vulcan of 75-80 %towards CO 2 ( Figure S7) while no CO 2 formation can be observed for Ni x B/C-10 at any of the applied current densities. It has to be noted that despite reports suggesting that carbon oxidation in alkaline solution leads to dissolved carbonaceous molecules,i no ur study only CO 2 was used as carbon oxidation marker.I na ddition to CO 2 ,O 2 was detected as as econd reaction product ( Figure S8, S9). However,V ulcan only produced as mall amount of O 2 which did not change substantially with the current density.T he O 2 MS signal detected for measurements involving Ni x B/C-10 gradually increases with increasing current density.A tt he current densities that are necessary to achieve ar easonable CO 2 MS signal, the O 2 formation is already so vigorous that the saturation concentration of O 2 in the electrolyte is exceeded and bubble formation hampers ap roper calibration of the system for O 2 .N evertheless,t he quantitative CO 2 data together with the qualitative O 2 data revealed that carbon oxidation is presumably suppressed by the catalytically increased OER kinetics.T hus,s upporting OER catalysts on carbonaceous materials or preparing them based on carbonaceous precursors might lead to ap rotection of the carbon material by the enhanced reaction kinetics provided by the catalyst at alkaline OER conditions.T hese findings are in agreement with the results of Lafforgue et al. [14] on the carbon corrosion in presence of Pt. Different to Ni x B, Pt shows only minor activity for the OER, thus carbon corrosion is enhanced. Additionally,i nP EMFC research it is well known that addition of the active OER catalyst IrO 2 to the ORR catalyst hampers carbon corrosion.
Furthermore,t he occurrence of aC O 2 MS signal if catalytic activity is lost over longer time,b ye ither catalyst deactivation or particle loss,p oints towards an increased carbon corrosion rate further corroborating that the presence of an OER catalyst protects carbon from oxidation even under alkaline OER conditions ( Figure S10).
In conclusion, we successfully developed an ew experimental DEMS-based procedure with au nique cell design that makes it possible to directly detect CO 2 formation as am arker for carbon corrosion in alkaline electrolytes,w hich has hitherto not been possible.I tw as demonstrated that during OER using carbon or carbon supported catalysts, OER and carbon oxidation proceed concurrently,h owever, carbon oxidation was considerably suppressed upon enhancing the OER kinetics using ah ighly active OER catalyst. Therefore,t his study does not only present an ew methodology for detecting carbon corrosion in alkaline electrolytes but also provides insight in the fate of carbon during electrocatalytic OER on carbon or carbon supported catalysts.T he results are therefore not only valuable for fundamental understanding but are also of practical importance for monitoring carbon corrosion in technical applications.