Electrochemical Removal of NO x on Ceria-Based Catalyst-Electrodes

This study reports the electrochemical properties for NOx reduction of a ceria-based mixed ionic electronic conducting porous electrode promoted by Pt nanoparticles, as efficient catalyst for NO oxidation, and BaO, as sorbent to store NOx. This catalytic layer was deposited by screen-printing on a dense membrane of gadolinia-doped ceria, an O2− ionic conductor. The targeted Ba and Pt loadings were 150 and 5 μg/cm2, respectively. The NOx selective electrochemical reduction was performed between 400 ◦C and 500 ◦C with and without oxygen in the feed. Variations of the open-circuit voltage with time were found to be a good sensor of the NOx storage process on the ceria-based catalyst-electrode. However, no N2 production was observed in the presence of O2 phase in spite of nitrates formation.


Introduction
Nitrogen oxides (NO x : NO + NO 2 ) are pollutants mainly emitted by thermal engines using fossil fuels.They are at the origin of safety problems.The design of more efficient and stable catalysts to reduce NO x to nitrogen in atmospheres containing an excess of oxygen, such as exhaust gas emitted by diesel and lean-burn gasoline engines, has attracted much attention in the last years.Due to the more and more stringent emission standards, an efficient and clean technology to control the NO x emissions from automotive engines becomes an important demand.In Europe [1], the NO x emission limit is 80 mg/km for diesel passenger cars according to the Euro 6 standard in force since 2014.The two actual technologies on the market are using an additional reducing agent to remove NO x , i.e., diesel fuel post-injections for the NO x Lean Trap Catalyst (LTC) and urea for the Selective Catalytic Reduction (SCR) [2].The former is containing Pt and BaO for the oxidation NO into NO 2 , and the storage of NO 2 as nitrates, respectively.When the LTC surface is saturated by nitrates, a diesel post injection is triggered for few seconds to decompose nitrates and reduce NO x into N 2 .However, these periodic short rich phases provoke a fuel overconsumption.SCR catalysts are based on zeolite materials which are cheap and robust but the control of the urea injection to avoid NH 3 release is tricky.In addition, recent studies on the on-road emissions of NO x from Euro 6 Diesel vehicles have clearly shown that these two technologies are not sufficiently effective to comply the standards.Therefore, alternative solutions have to be developed.One promising solution could be the Selective Electrochemical Reduction (SER) of NO x into N 2 in a Solid Oxide Electrolysis Cell (SOEC).This latter process does not need an additional reducing agent to reduce NO x .The reduction is ensured by a cathodic polarization.Then, SER saves the large reducing agent storage system and avoids the emission of pollutants such as NH 3 (SCR) or VOC (LTC) produced by the reducing agents [3][4][5].It was first proposed by Pancharatnam et al. [6] in 1976 that NO can be electrochemically reduced into N 2 on a Pt cathode interfaced on an oxygen ionic conductor.The reaction mechanism was investigated by Gür and Huggins [7] in 1979.However, diesel exhausts contain a large quantity of oxygen.The competitive electrochemical reduction of O 2 into O 2− is much faster than the SER of NO x on a Pt electrode [8].To improve the selectivity of the NO x electrochemical conversion, some researchers have recently shown that the addition of a NO x sorbent (K, Ba) can promote SER of NO x into N 2 [9][10][11][12].Two approaches are proposed in the literature.The former deals with multilayers configurations with a NO x adsorption layer deposited at the top of a catalyst-electrode layer [9,11].The latter seems to be more promising and refers to the infiltration of a NO x sorbent, such as BaO, into the porosity of the catalyst-electrode [12-14].For instance, remarkable results have been obtained with the infiltration of both BaO particles and a Pt/Al 2 O 3 catalyst into the porosity of symmetrical electrodes of (La 0.85 Sr 0.15 ) 0.99 MnO 3 (LSM) interfaced on a dense GDC membrane [12].At 450 • C, NO x conversion and N 2 selectivity as high as 65% and 86%, respectively, were achieved in an excess of oxygen.
The objective of this study was to infiltrate both BaO and Pt nanoparticles in the porosity of a ceria-based electrode layer to delocalize the SER of NO x into the whole volume of the electrode.The mixed ionic electronic conducting (MIEC) coating was a composite electrode between LSCF (La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3-δ ) and GDC (Gd 0.2 Ce 0.8 O 1.9 ), able to conduct both O 2− and electrons.A dense membrane of GDC was used as the electrolyte due to its high ionic conductivity at low temperature [15,16].The NO x electrocatalytic performances were investigated between 400 • C and 500 • C in NO/O 2 /He atmospheres.

Catalyst-Electrode Characterizations
The Pt-BaO/LSCF-GDC electrode structure and morphology were characterized by SEM (Scanning Electronic Microscopy).The thickness of the LSCF/GDC electrode was around 10 µm (Figure 1a).The morphology of the LSFC-GDC composite layer was shown on Figure 1b.The electrode is made of micrometric and agglomerated grains, probably due to the high calcination temperature used during the preparation (1200 • C).Pores size is around 1-3 µm.TEM (Transmission Electronic Microscopy) observations after the extractive replica preparation method of the samples, have revealed the presence of Pt nanoparticles. Figure 2 shows a typical Pt nanoparticle.The mean diameter of Pt nanoparticles was around 10 nm.Unfortunately, BaO particles were not detected neither with SEM, most probably due to small grain sizes, nor with TEM.This suggests that BaO particles are heterogeneously distributed in the electrode.Recently, the group of K. Kammer Hansen [3] has infiltrated BaO in the porosity of a similar cell made of 60 wt % of La 0.85 Sr 0.15 Co x Mn 1-x O 3+δ and 40 wt % of Ce 0.9 Gd 0.1 O 1.95 .They have shown that BaO nanoparticles in the inner layers are only present on the LSM phase.Therefore, BaO in our case, could be preferentially localized on LSCF grains.

NO x Electrocatalytic Conversions without Oxygen in the Feed
The parameter ∆NO x difference between inlet (NO x,in ) and outlet (NO x,out ) concentrations of NO x (NO + NO 2 ), was used to highlight the NO x storage process as nitrates on the catalyst-electrode.Upon open-circuit voltage (OCV), without any polarization and then electrochemical reaction, positive values of ∆NO x indicate that a part of NO x species is stored on the catalyst-electrode.
diameter of Pt nanoparticles was around 10 nm.Unfortunately, BaO particles were not detected neither with SEM, most probably due to small grain sizes, nor with TEM.This suggests that BaO particles are heterogeneously distributed in the electrode.Recently, the group of K. Kammer Hansen [3] has infiltrated BaO in the porosity of a similar cell made of 60 wt % of La0.85Sr0.15CoxMn1-xO3+δand 40 wt % of Ce0.9Gd0.1O1.95.They have shown that BaO nanoparticles in the inner layers are only present on the LSM phase.Therefore, BaO in our case, could be preferentially localized on LSCF grains.

NOx Electrocatalytic Conversions without Oxygen in the Feed
The parameter ΔNOx difference between inlet (NOx,in) and outlet (NOx,out) concentrations of NOx (NO + NO2), was used to highlight the NOx storage process as nitrates on the catalyst-electrode.Upon open-circuit voltage (OCV), without any polarization and then electrochemical reaction, positive values of ΔNOx indicate that a part of NOx species is stored on the catalyst-electrode.
Figure 3 shows variations of OCV (Open-Circuit Voltage) and ΔNOx with time at 500 °C.Initial positive values of ΔNOx confirm that a part of NOx is stored on the catalyst-electrode, most probably on BaO sites as Ba(NO3)2.Similar experiments carried out with a Ba free composite electrode gave negligible values of ΔNOx, then confirming the active role of BaO to store NOx.

NOx Electrocatalytic Conversions without Oxygen in the Feed
The parameter ΔNOx difference between inlet (NOx,in   On the complete composite electrode, traces of oxygen in the feed or oxygens stored in GDC allow the catalytic production of NO2 on Pt nanoparticles.An NO2 production peak at around 40 ppm was detected 4 min after the reactants introduction, corresponding to the beginning of the ΔNOx plateau.The duration of this latter is approximately 3 min and then the ΔNOx value gradually decreases down to zero after 30 min on stream.This slow process corresponds to the saturation of the Ba sites.The most interesting point is the OCV variations with time.The OCV value corresponds to the potential difference between that of the catalyst-electrode and that of the Au counter-electrode.On the complete composite electrode, traces of oxygen in the feed or oxygens stored in GDC allow the catalytic production of NO 2 on Pt nanoparticles.An NO 2 production peak at around 40 ppm was detected 4 min after the reactants introduction, corresponding to the beginning of the ∆NO x plateau.The duration of this latter is approximately 3 min and then the ∆NO x value gradually decreases down to zero after 30 min on stream.This slow process corresponds to the saturation of the Ba sites.The most interesting point is the OCV variations with time.The OCV value corresponds to the potential difference between that of the catalyst-electrode and that of the Au counter-electrode.As reported in the literature [10,17,18], OCV values in a single chamber electrochemical cell refer to the difference in the thermodynamic activity of adsorbed atomic oxygen between the two electrodes.Variations of the potential of the Au counter-electrode prepared from Au paste are negligible in this temperature range [17].Therefore, recorded variations of OCV are linked to modifications of the oxygen coverage on the catalyst-electrode.Before the introduction of the reactants, sample was only exposed to He and the OCV value was −50 mV at 500 • C, suggesting that the oxygen coverage in presence of traces of oxygen is lower on Pt-Ba/LSCF-GDC than on Au.After the introduction of the NO x reactant, OCV values rapidly drop to reach a plateau at −160 mV which exactly corresponds to that of ∆NO x , demonstrating that the catalyst-electrode potential gives a direct evidence of the NO x storage progress, as already reported on a Pt-Ba electrode [10].The NO x storage process strongly decreases the oxygen coverage on the electrode most probably because NO 2 and traces of oxygen are consumed to produce nitrates according to the reaction shown in Equation (1) whereas NO 2 and O 2 can be adsorbed on Au.
After the plateau, OCV gradually increases with time, in good concordance with the ∆NO x decay.
The NO x SER performance was investigated at 450 • C and 550 • C. Different negative polarizations were applied between the catalytic electrode and the counter electrode (Figures 4 and 5) between −5 V and −7 V. Please note that the ohmic drop was not subtracted from these values.Without any oxygen in the feed, nitrogen oxides, mainly composed of NO, can be electrochemically reduced into N 2 with a 100% selectivity.At 500 • C, NO x conversions are around 20% and only slightly vary with the negative potential (Figures 4b and 5a) whereas produced negative currents strongly increase (from −2.4 mA to −4.2 mA).This indicates that the generated current is not only produced by the NO x SER and the N 2 production.One can assume that, under these operating conditions, GDC can be concomitantly electrochemically reduced.On the opposite, at lower temperature, i.e., 450 • C, the NO x conversion linearly increases with the cathodic potential from 10.5% upon −5 V up to 15.5% upon −7 V.At 450 • C (Figure 4a), current intensities are lower and proportional to the NO x conversion, suggesting that there is no GDC reduction.Figure 5 shows that the NO concentration decrease upon cathodic polarizations is gradual and slow.At 450 °C (Figure 5a), the NO concentration does not reach a plateau after more than 40 min polarization.Figure 6 displays variations of the NO concentration and the current with time at 450 °C upon −5 V.The decay of the NO concentration with time is slow, confirming the low kinetic rate Figure 5 shows that the NO concentration decrease upon cathodic polarizations is gradual and slow.At 450 • C (Figure 5a), the NO concentration does not reach a plateau after more than 40 min polarization.Figure 6 displays variations of the NO concentration and the current with time at 450 • C upon −5 V.The decay of the NO concentration with time is slow, confirming the low kinetic rate of the process.The current variation shows a negative peak after 3 min on stream which corresponds to the ∆NO x value plateau (Figure 3).This demonstrates that the generated current is dependent on the NO x storage process and then on the NO 2 production peak.

NO x Electrocatalytic Conversion in the Presence of 1% O 2
Electrocatalytic performances were also investigated in the presence of 1% O 2 .At 500 • C, the inlet NO and NO 2 concentrations were about 650 ppm and 40 ppm, respectively, due to the production of NO 2 in the stainless steel pipes.The presence of O 2 concentration increases the steady stable value of NO 2 concentration, as NO is oxidized into NO 2 onto the catalyst-electrode.The NO conversion into NO 2 is around 9% at 500 • C. Values of ∆NO x are positive right after the reactive mixture introduction (Figure 7).Therefore, as expected, the NO x storage process is taking place on the catalyst-electrode.Let us note that without any BaO in the electrode, values of ∆NO x is closed to zero meaning that no NO x storage is taking place.In presence of BaO, contrary to the experiments without oxygen, ∆NO x variations do not exhibit a plateau but a sharp peak after 3 min on stream followed by a gradual decay down to −10 ppm.Features of the OCV variations are symmetrically inverted with an initial drop to negative values (−45 mV) followed by a progressive increase up to around 0 mV.The observed slight time shift between the ∆NO x positive peak and the OCV negative one is only due to the slow response time of the NO x analyzer.Therefore, these experiments in presence of oxygen confirm that the OCV value is a good sensor of the NO x storage process.Negative values of OCV indicate that the NO x storage process is running while a null value points that the catalyst-electrode surface is saturated by nitrates compounds.
Figure 8 gives the variations of NO and NO 2 concentrations upon different voltages from ±3 V to ±1 V at 400 • C. At OCV, the NO conversion into NO 2 is 13%.Whatever the polarization, no production of nitrogen was detected in spite of the NO x storage process (Figure 7) which proves the presence of Ba.Cathodic and anodic polarizations only induce the electrochemical reduction of NO 2 into NO and the reverse reaction, respectively.Conversion of NO into NO 2 can reach 15% upon +3 V.The most important point is the ability of the electrode to electrochemically reduce NO 2 into NO for low potentials.At 400 • C, upon −1 V, the NO 2 conversion into NO can reach 30%.However, this conversion is not enhanced by higher cathodic potentials, suggesting that the electrochemical reduction of oxygen becomes predominant.These results are in contradiction with those obtained by Shao et al. [12] with a LSM electrode infiltrated with both BaO and Pt/Al 2 O 3 .This study has evidenced high conversions of NO x into N 2 in excess of oxygen upon cathodic potentials.Several causes can explain the different results obtained in this study, such as a quite heterogeneous distribution of the BaO particles in the electrode or a chemical reactivity between BaO and LSCF/GDC materials as proposed by the group of Kammer Hansen [3].The localization of BaO nanoparticles is quite important since the production of N 2 will depend on the efficiency of the selective electrochemical reduction of Ba(NO 3 ) 2 .
Ba(NO 3 ) 2 + 10 e − → BaO + N 2 + 5 O 2− (2) Catalysts 2017, 7, 61 7 of 10 Figure 8 gives the variations of NO and NO2 concentrations upon different voltages from ±3 V to ±1 V at 400 °C.At OCV, the NO conversion into NO2 is 13%.Whatever the polarization, no production of nitrogen was detected in spite of the NOx storage process (Figure 7) which proves the presence of Ba.Cathodic and anodic polarizations only induce the electrochemical reduction of NO2 into NO and the reverse reaction, respectively.Conversion of NO into NO2 can reach 15% upon +3 V.The most important point is the ability of the electrode to electrochemically reduce NO2 into NO for low potentials.At 400 °C, upon −1 V, the NO2 conversion into NO can reach 30%.However, this conversion is not enhanced by higher cathodic potentials, suggesting that the electrochemical reduction of oxygen becomes predominant.These results are in contradiction with those obtained by Shao et al. [12] with a LSM electrode infiltrated with both BaO and Pt/Al2O3.This study has evidenced high conversions of NOx into N2 in excess of oxygen upon cathodic potentials.Several causes can explain the different results obtained in this study, such as a quite heterogeneous distribution of the BaO particles in the electrode or a chemical reactivity between BaO and LSCF/GDC materials as proposed by the group of Kammer Hansen [3].The localization of BaO nanoparticles is quite important since the production of N2 will depend on the efficiency of the selective electrochemical reduction of Ba(NO3)2.(2) Figure 8 gives the variations of NO and NO2 concentrations upon different voltages from ±3 V to ±1 V at 400 °C.At OCV, the NO conversion into NO2 is 13%.Whatever the polarization, no production of nitrogen was detected in spite of the NOx storage process (Figure 7) which proves the presence of Ba.Cathodic and anodic polarizations only induce the electrochemical reduction of NO2 into NO and the reverse reaction, respectively.Conversion of NO into NO2 can reach 15% upon +3 V.The most important point is the ability of the electrode to electrochemically reduce NO2 into NO for low potentials.At 400 °C, upon −1 V, the NO2 conversion into NO can reach 30%.However, this conversion is not enhanced by higher cathodic potentials, suggesting that the electrochemical reduction of oxygen becomes predominant.These results are in contradiction with those obtained by Shao et al. [12] with a LSM electrode infiltrated with both BaO and Pt/Al2O3.This study has evidenced high conversions of NOx into N2 in excess of oxygen upon cathodic potentials.Several causes can explain the different results obtained in this study, such as a quite heterogeneous distribution of the BaO particles in the electrode or a chemical reactivity between BaO and LSCF/GDC materials as proposed by the group of Kammer Hansen [3].The localization of BaO nanoparticles is quite important since the production of N2 will depend on the efficiency of the selective electrochemical reduction of Ba(NO3)2.Ba(NO 3 ) 2 , located near triple phase boundaries (gas/O 2− /e − ) in the vicinity of the layer closed to the CGO electrolyte will be much more easily electrochemically reducible than nitrate nanoparticles only present on LSFC grains at the surface of the electrode.During this latter case, the electrochemical reduction of Ba(NO 3 ) 2 will be in strong competition with the oxygen electrochemical reduction.Therefore, if we consider that Ba(NO 3 ) 2 is preferentially formed on LSCF nanoparticles as suggested by the group of K. Kammer Hansen [3], it explains the low kinetic of the reduction process (Figures 5 and 6) observed without oxygen in the feed as nitrates have to diffuse near TPBs (Triple Phase Boundary) and the non-production of N 2 in presence of oxygen as oxygen becomes predominantly electrochemically reduced.Additional characterizations are clearly needed to elucidate the origin of the non-selective electrochemical reduction of NO x of the ceria-based catalyst electrode.

Electrode Preparation
The LSCF (70 wt %)/GDC(30 wt %) composite electrode was prepared by screen-printing (semin-automatic AUREL C890 machine, Modigliana (FC), Italy) [19] on a GDC dense disk (diameter 17 mm, thickness 1 mm) followed by a calcination step at 1200 • C. A gold film was deposited on the opposite side of the GDC pellet in order to act as counter electrode by using a gold paste (Metalor, Oullins, France) annealed at 500 • C for 2 h.Generated micrometric gold particles were found to be inactive for NO x activation as verified through blank experiments under our experimental conditions.Nanoparticles of Pt and BaO were impregnated in the porosity of the LSCF/GDC electrode.An aqueous solution containing both precursors of Pt(NH 3 ) 4 (NO 3 ) 2 and Ba(NO 3 ) 2 has been prepared.The concentrations of the Pt and Ba solutions were 0.32 × 10 −3 mol/L and 1.36 × 10 −2 mol/L, respectively.In addition, PVP (polyvinylpyrrolidone) as the surfactant was also added to the solution (10 wt %) to ensure a suitable viscosity.A controlled volume of 40 µL of the Pt/BaO solution has been infiltrated four times successively into the porosity of the LSCF/GDC electrode by using a micropipette.Between each infiltration, the sample was dried for 1 h at 80 • C. The targeted loadings of the Pt and Ba into the porous electrode were 5 µg/cm 2 and 150 µg/cm 2 , respectively.The sintering temperature of infiltrated electrodes was set at 600 • C for 1 h in air.The Pt nanoparticles in the LSCF/GDC film were characterized by TEM (JEOL 2010 LaB6, JEOL, Peabody, MA, USA) using an extractive replica technique (Figure 2) as described in [20].A series of composite electrodes without BaO have been also prepared with the same loading of Pt to compare the NO x storage capacity.

Measurements of the Electrocatalytic Performances
The Pt-BaO/LSCF-GDC/GDC/Au samples were placed in a single chamber quartz reactor described elsewhere [21].All the current collectors were made of gold.A gold mesh was placed on the top of the Pt-BaO/LSCF-GDC porous working electrode as current collector.The quartz reactor was installed in a tubular furnace.A potentiostat-galvanostat (Voltalab 80, Radiometer Analytical, Hach Company, Loveland, CO, USA) was used to polarize samples between the catalyst-electrode and the Au counter-electrode.
The reactive mixture gases were composed of NO (4000 ppm in He, LINDE France S.A., Saint-Priest, France, 99.95% purity) and O 2 (LINDE France S.A. 99.95% purity).Helium (LINDE, 99.95% purity) was used as the carrier gas.The gas compositions were adjusted with mass flow controllers (Brooks) with an accuracy of 1%.The testing temperature range was varied from 400 • C to 500 • C. Concentrations of NO, N 2 O and NO 2 were monitored by an online analyzer (EMERSON NGA 2000, Emerson Process Management, Bron, France) while N 2 and O 2 contents were determined with a micro-chromatograph (R3000 SRA Instruments, Marcy l'Etoile, France).The NO x conversion experiments were investigated in presence of 680 ppm NO x (670 ppm NO and 10 ppm NO 2 ) without O 2 and for a total flow rate of 42 mL/min.Based on above experiments, the NO x conversion experiments were also studied in presence of 1% O 2 .

Figure 2 .
Figure 2. TEM ((Transmission Electronic Microscopy) image of a Pt nanoparticle infiltrated in the catalyst-electrode.

Figure 3 .
Figure 3. Variations of open-circuit voltage (OCV) and ΔNOx as a function of time at 500 °C.Reactive mixture: NO/NO2: 680 ppm/10 ppm in He.

Figure 2 .
Figure 2. TEM ((Transmission Electronic Microscopy) image of a Pt nanoparticle infiltrated in the catalyst-electrode.

Figure 3 10 Figure 1 .
Figure 3 shows variations of OCV (Open-Circuit Voltage) and ∆NO x with time at 500 • C. Initial positive values of ∆NO x confirm that a part of NO x is stored on the catalyst-electrode, most probably on BaO sites as Ba(NO 3 ) 2 .Similar experiments carried out with a Ba free composite electrode gave negligible values of ∆NO x , then confirming the active role of BaO to store NO x .

Figure 2 .
Figure 2. TEM ((Transmission Electronic Microscopy) image of a Pt nanoparticle infiltrated in the catalyst-electrode.
) and outlet (NOx,out) concentrations of NOx (NO + NO2), was used to highlight the NOx storage process as nitrates on the catalyst-electrode.Upon open-circuit voltage (OCV), without any polarization and then electrochemical reaction, positive values of ΔNOx indicate that a part of NOx species is stored on the catalyst-electrode.

Figure 3
shows variations of OCV (Open-Circuit Voltage) and ΔNOx with time at 500 °C.Initial positive values of ΔNOx confirm that a part of NOx is stored on the catalyst-electrode, most probably on BaO sites as Ba(NO3)2.Similar experiments carried out with a Ba free composite electrode gave negligible values of ΔNOx, then confirming the active role of BaO to store NOx.

Figure 3 .
Figure 3. Variations of open-circuit voltage (OCV) and ΔNOx as a function of time at 500 °C.Reactive mixture: NO/NO2: 680 ppm/10 ppm in He.

Figure 3 .
Figure 3. Variations of open-circuit voltage (OCV) and ∆NO x as a function of time at 500 • C. Reactive mixture: NO/NO 2 : 680 ppm/10 ppm in He.

Figure 4 .
Figure 4. Variations of NO x conversion and currents as a function of applied potentials (a) at 450 • C and (b) at 500 • C. Reactive mixture: NO/NO 2 : 680 ppm/10 ppm in He.

Figure 4 .
Figure 4. Variations of NOx conversion and currents as a function of applied potentials (a) at 450 °C and (b) at 500 °C.Reactive mixture: NO/NO2: 680 ppm/10 ppm in He.

Figure 5 .
Figure 5. Variations of NO and N2 concentrations as a function of time at (a) 450 °C and (b) 500 °C.Reactive mixture: NO/NO2: 680 ppm/10 ppm in He.

Figure 5 .
Figure 5. Variations of NO and N 2 concentrations as a function of time at (a) 450 • C and (b) 500 • C. Reactive mixture: NO/NO 2 : 680 ppm/10 ppm in He.

Figure 6 .
Figure 6.Variations of NO concentration and the current as a function of time at 450 • C. Reactive mixture: NO/NO 2 : 680 ppm/10 ppm in He.