Effect of Electronic Conductivities of Iridium Oxide/Doped SnO 2 Oxygen-Evolving Catalysts on the Polarization Properties in Proton Exchange Membrane Water Electrolysis

: We have developed IrO x /M-SnO 2 (M = Nb, Ta, and Sb) anode catalysts, IrO x nanoparticles uniformly dispersed on M-SnO 2 supports with fused-aggregate structures, which make it possible to evolve oxygen efﬁciently, even with a reduced amount of noble metal (Ir) in proton exchange membrane water electrolysis. Polarization properties of IrO x /M-SnO 2 catalysts for the oxygen evolution reaction (OER) were examined at 80 ◦ C in both 0.1 M HClO 4 solution (half cell) and a single cell with a Naﬁon ® membrane (thickness = 50 µ m). While all catalysts exhibited similar OER activities in the half cell, the cell potential ( E cell ) of the single cell was found to decrease with the increasing apparent conductivities ( σ app, catalyst ) of these catalysts: an E cell of 1.61 V (voltage efﬁciency of 92%) at 1 A cm − 2 was achieved in a single cell by the use of an IrO x /Sb-SnO 2 anode (highest σ app, catalyst ) with a low Ir-metal loading of 0.11 mg cm − 2 and Pt supported on graphitized carbon black (Pt/GCB) as the cathode with 0.35 mg cm − 2 of Pt loading. In addition to the reduction of the ohmic loss in the anode catalyst layer, the increased electronic conductivity contributed to decreasing the OER overpotential due to the effective utilization of the IrO x nanocatalysts on the M-SnO 2 supports, which is an essential factor in improving the performance with low noble metal loadings.


Introduction
Proton exchange membrane water electrolysis (PEMWE) is an attractive method to produce high purity hydrogen with high energy conversion efficiency, even at high current densities, together with easy maintenance, start-up, and shut-down [1][2][3][4]. Such superlative characteristics make PEMWE suitable for leveling of the large fluctuations of renewable energy sources when used in combination with stationary fuel cells. Conventional PEMWE cells, however, are costly because large amounts of noble metals are used as the electrocatalysts, e.g., (Ir + Pt) black at the oxygen-evolving anode [≥2 mg(Ir + Pt) cm −2 ] and Pt black at the hydrogen-evolving cathode [≥2 mg(Pt) cm −2 ] to maintain high conversion efficiencies with long lifetimes [2,[5][6][7].
Iridium-based anodes have been employed so far, in spite of the high cost and limited availability of Ir, because they have exhibited relatively high activities and high stabilities for the oxygen evolution reaction (OER) [8][9][10]. It is essential to develop new anode catalysts that utilize Ir more effectively, working toward much higher mass activity (MA, current per mass of noble metal) for the OER, as well as high durability, while clarifying the reaction mechanisms [11,12]. In order to increase the MA, iridium or iridium oxide (IrO x ) nanoparticles have been mixed or dispersed on various supports such as metal carbides [13][14][15] and oxides [16][17][18][19][20][21][22]. Considering the stability at the high oxygen-evolving potentials in strong acidic media and the need for high electronic conductivity, doped tin oxides have been reported as promising candidates as support materials [23,24]. Indeed, thin films and bulk powders of SnO 2 doped with Sb, Nb, Ta, In, and F have exhibited electronic conductivities ≥0.1 S cm −1 , which are sufficiently high for consideration as catalyst supports [25,26]. It has been reported that the cell potentials (E cell ) of PEMWE single cells with IrO x supported on SnO 2 anodes reached values ≤1.65 V (≥ 90% voltage efficiency) at 1 A cm −2 with moderate Ir-metal loadings of 0.75 to 1 mg(Ir) cm −2 [17,18,27,28]. However, the polarization performances of such catalysts are still not sufficient in the catalyst layers of single cells for the further reduction of the Ir-loading down to 1/10 of those in conventional cells, i.e., target values of ≤0.2 mg(Ir) cm −2 . One of the reasons for this is the large contact resistance between SnO 2 particles, even though the bulk electronic conductivity of the doped SnO 2 itself is high.
Recently, Kakinuma et al. synthesized several M-doped SnO 2 (M = Nb, Ta, and Sb) materials with fused-aggregate network structures as corrosion-resistant cathode catalyst supports for polymer electrolyte fuel cells [29][30][31]. Unique advantages of these supports are their enhanced electronic conductivity and high gas diffusion rate. Onto such M-SnO 2 supports, we succeeded in dispersing IrO x nanoparticles as novel anode catalysts for PEMWE. It was found that an IrO x /Ta-SnO 2 catalyst exhibited an apparent MA of 15 A mg(Ir) −1 for the OER in 0.1 M HClO 4 solution at 1.5 V (iR-free) vs. RHE and 80 • C, which suggests the possibility of reducing the loading of Ir in an anode catalyst to a level as low as 0.1 mg(Ir) cm −2 at a voltage efficiency of 90% (E cell = 1.65 V) operated at 1 A cm −2 , i.e., the anode potential of 1.5 V, cathode potential of −0.05 V, and the ohmic loss of the PEM of 0.10 V [32].
In the present research, we examined the polarization properties of a series of IrO x /M-SnO 2 (M = Nb, Ta, and Sb) catalysts for the OER at 80 • C in both 0.1 M HClO 4 solution (half cell) and a single cell with a Nafion ® membrane (thickness = 50 µm). We, for the first time, found that the E cell of the single cell decreased with the increasing apparent conductivities (σ app, catalyst ) of these catalysts, whereas they exhibited similar OER activities in the half cell test. The highest performance, E cell of 1.61 V (voltage efficiency = 92%) at 1 A cm −2 was obtained in a single cell with total noble metal loading of 0.46 mg(Ir + Pt) cm −2 , in which the IrO x /Sb-SnO 2 anode catalyst (highest σ app, catalyst ) contribute greatly. Figure 1 shows a transmission electron microscopic (TEM) image of IrO x particles with particle size distribution, which were dispersed on Sb-SnO 2 with fused-aggregate network structures (IrO x /Sb-SnO 2 ). TEM images and particle size distribution histograms for IrO x /Nb-SnO 2 and IrO x /Ta-SnO 2 , cited from our previous work [32], are also shown for comparison in Figure S1 in the Supplementary Materials. IrO x nanoparticles of 1 to 3 nm in diameter were found to be dispersed uniformly on the oxide supports. The average sizes and the standard deviations of the IrO x nanoparticles were 2.0 ± 0.3, 2.2 ± 0.3, and 2.0 ± 0.4 nm for the IrO x /Nb-SnO 2 , IrO x /Ta-SnO 2 , and IrO x /Sb-SnO 2 catalysts, respectively. For a conventional catalyst employed as a reference (mixture of commercial IrO 2 and Pt black, 1:1 mass ratio), scanning electron microscopic (SEM) and TEM images of IrO 2 and Pt particles are shown in Figure S2. The average particle size of commercial IrO 2 was 25 nm.

Physical Properties of IrO x /M-SnO 2 Catalysts
exhibited the highest σapp, support among the supports examined, i.e., three orders of magnitude higher than that of Nb-SnO2. The σapp, support values of all doped-SnO2 increased by ca. two orders of magnitude by dispersing IrOx on their surface. In particular, the σapp, catalyst of the IrOx/Sb-SnO2 catalyst was the highest value of 8.1 × 10 −1 S cm −1 . As reported previously for Pt/Nb-SnO2 [33] and IrOx/M-SnO2 (M = Nb and Ta) [32], such an increase in the conductivity for IrOx/Sb-SnO2 is ascribed to the shrinkage of the depletion layer of the SnO2 support particles [33]. Thus, we successfully synthesized IrOx/M-SnO2 catalysts with similar microstructures but with a range of different of σapp, catalyst values. Since the thermodynamically stable species at OER potentials in acidic media is IrO2 (rutile) [34], the surface and/or interior of the IrOx particles on M-SnO2 can be converted to IrO2 during steadystate OER operation. Here, we compare the values of specific surface area of IrO2 (SIrO2) for commercial IrO2 powder and IrOx/M-SnO2. Assuming spherical particles of commercial IrO2 powder with 25 nm diameter based on SEM and TEM images in Figure S2, the value of SIrO2 was calculated to be 21 m 2 g(IrO2) −1 or 24 m 2 g(Ir) −1 . In contrast, as shown in the TEM images of Figure 1 and Figure S1, the average size of the IrOx particles dispersed on the M-SnO2 support was ca. 2 nm for the asprepared catalyst, in which the Ir 4+ (IrO2) percentage was ca. 20% (Table 1), as analyzed by XPS. Then, we calculated the surface coverage of IrO2 on the particles by estimating the ratio of surface atoms to the total number of atoms (Nsurface/Ntotal), assuming that the face-centered cubic (fcc) Ir particles have an ideal cubo-octahedral shape. The calculation method [35,36] is shown in Appendix S2. The value of Nsurface/Ntotal for a 2.0 nm particle was calculated to be 52%. For the case of IrOx/Nb-SnO2 as an example, the value of 16% Ir 4+ can be rationally explained if 31% of the surface atoms (=16/52) were oxidized to IrO2 in the as-prepared catalyst. If all of the surface atoms were oxidized to IrO2 with an Ir metal core during the OER, the initial particle size of 2 nm would be nearly unchanged. Thus, the value of SIrO2 (on an Ir metal core) was 133 m 2 g(Ir) −1 . On the other hand, if all Ir atoms in the particle were oxidized to IrO2 during the OER, the particle size could increase from 2.0 nm to 2.6 nm while maintaining a constant Ntotal, resulting in a value of 102 m 2 g(Ir) −1 . Thus, we are able to estimate the increase in SIrO2 of the IrOx/Nb-SnO2 by a factor of 4.3 (=102/24) to 5.5 (=133/24), compared with that of the commercial IrO2. Values similarly calculated for IrOx/Ta-SnO2 and IrOx/Sb-SnO2 are summarized in Table S1. We also characterized these catalysts by BET surface area (Brunauer-Emmett-Teller adsorption method) of the M-SnO 2 supports (S SnO2 ), the iridium loadings, the percentage of Ir 4+ (IrO 2 ) in IrO x evaluated by X-ray photoelectron spectroscopy (XPS), the amounts of M-SnO 2 supports, the apparent electrical conductivities of the M-SnO 2 supports (σ app, support ) and IrO x -dispersed catalysts (σ app, catalyst ) (see Materials and Methods, Figure S3 and Appendix S1). These results are summarized in Table 1. While Sb-SnO 2 exhibited a somewhat larger S SnO2 value, similar amounts of iridium metal were loaded with similar percentages of Ir 4+ on all three catalysts. Marked differences among synthesized IrO x /M-SnO 2 catalysts are seen between σ app, support and σ app, catalyst values. The Sb-SnO 2 support exhibited the highest σ app, support among the supports examined, i.e., three orders of magnitude higher than that of Nb-SnO 2 . The σ app, support values of all doped-SnO 2 increased by ca. two orders of magnitude by dispersing IrO x on their surface. In particular, the σ app, catalyst of the IrO x /Sb-SnO 2 catalyst was the highest value of 8.1 × 10 −1 S cm −1 . As reported previously for Pt/Nb-SnO 2 [33] and IrO x /M-SnO 2 (M = Nb and Ta) [32], such an increase in the conductivity for IrO x /Sb-SnO 2 is ascribed to the shrinkage of the depletion layer of the SnO 2 support particles [33]. Thus, we successfully synthesized IrO x /M-SnO 2 catalysts with similar microstructures but with a range of different of σ app, catalyst values.
Since the thermodynamically stable species at OER potentials in acidic media is IrO 2 (rutile) [34], the surface and/or interior of the IrO x particles on M-SnO 2 can be converted to IrO 2 during steady-state OER operation. Here, we compare the values of specific surface area of IrO 2 (S IrO2 ) for commercial IrO 2 powder and IrO x /M-SnO 2 . Assuming spherical particles of commercial IrO 2 powder with 25 nm diameter based on SEM and TEM images in Figure S2, the value of S IrO2 was calculated to be 21 m 2 g(IrO 2 ) −1 or 24 m 2 g(Ir) −1 . In contrast, as shown in the TEM images of Figure 1 and Figure S1, the average size of the IrO x particles dispersed on the M-SnO 2 support was ca. 2 nm for the as-prepared catalyst, in which the Ir 4+ (IrO 2 ) percentage was ca. 20% (Table 1), as analyzed by XPS. Then, we calculated the surface coverage of IrO 2 on the particles by estimating the ratio of surface atoms to the total number of atoms (N surface /N total ), assuming that the face-centered cubic (fcc) Ir particles have an ideal cubo-octahedral shape. The calculation method [35,36] is shown in Appendix S2. The value of N surface /N total for a 2.0 nm particle was calculated to be 52%. For the case of IrO x /Nb-SnO 2 as an example, the value of 16% Ir 4+ can be rationally explained if 31% of the surface atoms (=16/52) were oxidized to IrO 2 in the as-prepared catalyst. If all of the surface atoms were oxidized to IrO 2 with an Ir metal core during the OER, the initial particle size of 2 nm would be nearly unchanged. Thus, the value of S IrO2 (on an Ir metal core) was 133 m 2 g(Ir) −1 . On the other hand, if all Ir atoms in the particle were oxidized to IrO 2 during the OER, the particle size could increase from 2.0 nm to 2.6 nm while maintaining a constant N total , resulting in a value of 102 m 2 g(Ir) −1 . Thus, we are able to estimate the increase in S IrO2 of the IrO x /Nb-SnO 2 by a factor of 4.3 (=102/24) to 5.5 (=133/24), compared with that of the commercial IrO 2 . Values similarly calculated for IrO x /Ta-SnO 2 and IrO x /Sb-SnO 2 are summarized in Table S1.  Figure 2a shows the iR-free polarization curves for the OER on IrO x /M-SnO 2 and conventional catalysts in air-saturated 0.1 M HClO 4 solution at 80 • C. The current is shown as the apparent MA, i.e., current value per mass of Ir (or Ir + Pt for the conventional catalyst) loaded on the electrode. In order to remove oxygen gas bubbles effectively from the electrode surface, the flow rate of the electrolyte solution was adjusted at 160 cm s −1 [32]. The polarization curves for IrO x /Nb-SnO 2 and IrO x /Ta-SnO 2 are taken from our previous work [32]. It should be noted that a small error in the iR subtraction (ca. 10 mV at 1 A cm −2 ) has been corrected. These IrO x /M-SnO 2 catalysts showed onset potentials for the OER from 1.35 to 1.40 V, which was similar to that for the conventional catalyst. Clearly, the MAs of the IrO x /M-SnO 2 catalysts were considerably higher than that of the conventional catalyst. The values of apparent MA exceeding 10 A mg(Ir) −1 for IrO x /Nb-SnO 2 , IrO x /Ta-SnO 2 , and IrO x /Sb-SnO 2 at 1.5 V were 28, 36, and 27 times larger, respectively, than that of the conventional one.   Figure 2a shows the iR-free polarization curves for the OER on IrOx/M-SnO2 and conventional catalysts in air-saturated 0.1 M HClO4 solution at 80 °C. The current is shown as the apparent MA, i.e., current value per mass of Ir (or Ir + Pt for the conventional catalyst) loaded on the electrode. In order to remove oxygen gas bubbles effectively from the electrode surface, the flow rate of the electrolyte solution was adjusted at 160 cm s −1 [32]. The polarization curves for IrOx/Nb-SnO2 and IrOx/Ta-SnO2 are taken from our previous work [32]. It should be noted that a small error in the iR subtraction (ca. 10 mV at 1 A cm −2 ) has been corrected. These IrOx/M-SnO2 catalysts showed onset potentials for the OER from 1.35 to 1.40 V, which was similar to that for the conventional catalyst. Clearly, the MAs of the IrOx/M-SnO2 catalysts were considerably higher than that of the conventional catalyst. The values of apparent MA exceeding 10 A mg(Ir) −1 for IrOx/Nb-SnO2, IrOx/Ta-SnO2, and IrOx/Sb-SnO2 at 1.5 V were 28, 36, and 27 times larger, respectively, than that of the conventional one. Because such enhancement factors of the MAs are much larger than those for SIrO2 described above (4.0 to 5.5 times, see Table S1), we examined the Tafel plots for the OER at IrOx/M-SnO2 and conventional catalysts, as shown in Figure 2b. Linear relationships are observed between the logarithm of MA and the iR-free potential (E) at E <1.43 V. The Tafel slope for the conventional catalyst (63 mV) was close to the commonly reported value (60 mV) for IrO2 electrodes in acidic solution Because such enhancement factors of the MAs are much larger than those for S IrO2 described above (4.0 to 5.5 times, see Table S1), we examined the Tafel plots for the OER at IrO x /M-SnO 2 and conventional catalysts, as shown in Figure 2b. Linear relationships are observed between the logarithm of MA and the iR-free potential (E) at E <1.43 V. The Tafel slope for the conventional catalyst (63 mV) was close to the commonly reported value (60 mV) for IrO 2 electrodes in acidic solution [28,38]. In contrast, the values of Tafel slopes for IrO x /M-SnO 2 catalysts ranged from 46 mV (IrO x /Ta-SnO 2 ) to 52 mV (IrO x /Sb-SnO 2 ). The existence of such low Tafel slopes, in comparison with that of bulk IrO 2 , implies that the OER rates on the IrO x /M-SnO 2 catalysts might be promoted by an interaction between the IrO x nanoparticles and the doped SnO 2 supports [28,32,39,40]. It has been also reported that an IrO x shell on an Ir core exhibited higher OER activity than IrO x [41,42]. Hence, the enhanced MAs of IrO x /M-SnO 2 might be ascribed not only to a significant increase in the active surface area, by the use of IrO x nanoparticles, but also their interaction with the oxide supports.

Oxygen Evolution Activities of IrO x /M-SnO 2 Catalysts in a Single Cell
We prepared catalyst-coated membranes (CCMs) with low noble metal loadings by the use of the IrO x /M-SnO 2 catalysts with 0.11 mg(Ir) cm −2 at the anode and a commercial Pt/GCB (Pt supported on graphitized carbon black) with 0.35 ± 0.02 mg(Pt) cm −2 at the cathode. A conventional anode catalyst (IrO 2 + Pt black, described above) with 2.66 mg(Ir + Pt) cm −2 and a Pt black cathode catalyst with 2.01 mg(Pt) cm −2 were employed in a reference CCM. The current-potential (I-E) curves of single cells operated at 80 • C are shown in Figure 3.
Catalysts 2019, 9, x FOR PEER REVIEW 5 of 12 [28,38]. In contrast, the values of Tafel slopes for IrOx/M-SnO2 catalysts ranged from 46 mV (IrOx/Ta-SnO2) to 52 mV (IrOx/Sb-SnO2). The existence of such low Tafel slopes, in comparison with that of bulk IrO2, implies that the OER rates on the IrOx/M-SnO2 catalysts might be promoted by an interaction between the IrOx nanoparticles and the doped SnO2 supports [28,32,39,40]. It has been also reported that an IrOx shell on an Ir core exhibited higher OER activity than IrOx [41,42]. Hence, the enhanced MAs of IrOx/M-SnO2 might be ascribed not only to a significant increase in the active surface area, by the use of IrOx nanoparticles, but also their interaction with the oxide supports.

Oxygen Evolution Activities of IrOx/M-SnO2 Catalysts in a Single Cell
We prepared catalyst-coated membranes (CCMs) with low noble metal loadings by the use of the IrOx/M-SnO2 catalysts with 0.11 mg(Ir) cm −2 at the anode and a commercial Pt/GCB (Pt supported on graphitized carbon black) with 0.35 ± 0.02 mg(Pt) cm −2 at the cathode. A conventional anode catalyst (IrO2 + Pt black, described above) with 2.66 mg(Ir + Pt) cm −2 and a Pt black cathode catalyst with 2.01 mg(Pt) cm −2 were employed in a reference CCM. The current-potential (I-E) curves of single cells operated at 80 °C are shown in Figure 3. The performances of the cells with three kinds of IrOx/M-SnO2 anodes were found to be enhanced in the order: IrOx/Nb-SnO2 < IrOx/Ta-SnO2 << IrOx/Sb-SnO2. For example, as shown in Table  2, the Ecell at 1 A cm −2 decreased from 1.91 V for IrOx/Nb-SnO2 cell to 1.61 V for the IrOx/Sb-SnO2 cell. The latter value was somewhat larger than that of the reference (conventional) cell (1.55 V). It is noteworthy that the initial cathode performance of Pt supported on high-surface-area carbon (Pt/C) was comparable to that of Pt black, even though Pt black has still been predominantly used in practical PEMWEs in order to ensure a long lifetime of the MEA [2]. In order to mitigate the corrosion of the carbon support, we used Pt supported on GCB in place of high-surface-area carbon. In any case, we consider that the increase in the overvoltage of our cell compared with that of the conventional cell can be ascribed predominantly to the anode catalyst with reduced amount of noble metal (<1/20). As shown in Figure S4, the values of MA based on mass of Ir for the IrOx/Sb-SnO2 catalyst at Ecell >1.45 V were considerably larger than that of the conventional cell. Interestingly, the Ecell of 1.61 V for the IrOx/Sb-SnO2 cell corresponds to a voltage efficiency of 92%, which is the highest performance at the significantly low Ir loading of 0.11 mg(Ir) cm −2 at the anode reported so far [28,[43][44][45]. The performances of the cells with three kinds of IrO x /M-SnO 2 anodes were found to be enhanced in the order: IrO x /Nb-SnO 2 < IrO x /Ta-SnO 2 << IrO x /Sb-SnO 2 . For example, as shown in Table 2, the E cell at 1 A cm −2 decreased from 1.91 V for IrO x /Nb-SnO 2 cell to 1.61 V for the IrO x /Sb-SnO 2 cell. The latter value was somewhat larger than that of the reference (conventional) cell (1.55 V). It is noteworthy that the initial cathode performance of Pt supported on high-surface-area carbon (Pt/C) was comparable to that of Pt black, even though Pt black has still been predominantly used in practical PEMWEs in order to ensure a long lifetime of the MEA [2]. In order to mitigate the corrosion of the carbon support, we used Pt supported on GCB in place of high-surface-area carbon. In any case, we consider that the increase in the overvoltage of our cell compared with that of the conventional cell can be ascribed predominantly to the anode catalyst with reduced amount of noble metal (<1/20). As shown in Figure S4, the values of MA based on mass of Ir for the IrO x /Sb-SnO 2 catalyst at E cell >1.45 V were considerably larger than that of the conventional cell. Interestingly, the E cell of 1.61 V for the IrO x /Sb-SnO 2 cell corresponds to a voltage efficiency of 92%, which is the highest performance at the significantly low Ir loading of 0.11 mg(Ir) cm −2 at the anode reported so far [28,[43][44][45]. Next, we discuss the essential parameters necessary to improve the anode performance. Referring to the properties of IrO x /M-SnO 2 catalysts in Table 1, the only marked differences are seen for the values of σ app, catalyst (or σ app, support ). The ohmic resistances of the cells (R ohm, cell , obs ) measured at 1 kHz during the operation are shown in Table 2: the R ohm, cell, obs values ranged from 75 to 258 mΩ cm 2 .
To start, we calculated values of R ohm, cell, calc for comparison with the observed values. First, we estimated R ohm, anode of the anode catalyst layers (CLs) as follows. The thickness of the IrO x /Sb-SnO 2 CL was ca. 10 µm, as observed by scanning ion microscopy (SIM; see Figure S5). Since we prepared all CLs in the same manner, we assumed the identical thickness for the IrO x /Ta-SnO 2 and IrO x /Nb-SnO 2 CLs. Assuming the porosity of the CLs to be 50%, we calculated their R ohm values based on their σ app, catalyst values. The values of R ohm, anode thus calculated for IrO x /Sb-SnO 2 , IrO x /Ta-SnO 2 , and IrO x /Nb-SnO 2 were 3, 68, and 1333 mΩ cm 2 , respectively. Second, for the Nafion ® electrolyte membrane with the thickness of 50 µm, we adopted the R ohm, Nafion to be 50 mΩ cm 2 . The R ohm, cell of the conventional cell in Table 2 was just 75 mΩ cm 2 , which is assumed to include R ohm, anode (IrO 2 + Pt black: the electronic conductivity of IrO 2 powder is very high [37]) and R ohm, cathode (Pt black), together with contact resistances with the gas diffusion layers (Pt/Ti mesh and carbon paper, see Materials and Methods). This value of R ohm, cell agrees precisely with those of polymer electrolyte fuel cells (PEFCs) with Nafion ® membrane of the identical thickness and Pt/C catalysts for the anode and cathode [46][47][48]. Thus, by adding the R ohm, anode of IrO x /M-SnO 2 to 75 mΩ cm 2 stated above, we calculated the R ohm, cell, calc values to be 78, 143, and 1408 mΩ cm 2 , for the cells with IrO x /Sb-SnO 2 , IrO x /Ta-SnO 2 , and IrO x /Nb-SnO 2 , respectively. The former two values are relatively consistent with those of R ohm, cell, obs . However, a large discrepancy is seen between R ohm, cell, obs and R ohm, cell, calc for the cell with IrO x /Nb-SnO 2 . One of the possible reasons is that σ app, catalyst was measured in ambient air (low humidity) at room temperature, while R ohm, cell, obs was measured during operation with the anode in pure water at 80 • C. It has been shown that the electronic conductivities of SnO 2 -based materials increase with humidity [49,50]. Water molecules adsorbed on the SnO 2 surface can act as electron donors, resulting in an increase in the carrier concentration near the surface. Such a tendency was shown to be more marked for SnO 2 samples with lower electronic conductivity [49,50]. Thus, it can be easily understood that the value of R ohm, cell, obs of IrO x /Nb-SnO 2 (in pure water at 80 • C) could be much smaller than that of R ohm, cell, calc . Taking into account such an effect of water on the electronic conductivity of the M-SnO 2 , it is appropriate to employ R ohm, cell, obs as a measure of the apparent resistance of the anode catalyst layer, rather than R ohm, cell, calc based on σ app, catalyst (measured in air).
It is clearly seen in Figure 3 and Table 2 that E cell decreased with decreasing R ohm, cell, obs . However, this is not simply due to the reduction of the ohmic (iR) loss. For example, the reduction of the iR loss at 1 A cm −2 is only ca. 0.08 V by replacing the IrO x /Ta-SnO 2 anode catalyst with IrO x /Sb-SnO 2 , but the reduction of the E cell in such a case was as large as 0.23 V. On the other hand, the OER activities (MA values or Tafel slopes) of the three IrO x /M-SnO 2 catalysts measured in 0.1 M HClO 4 solution in the previous section can be regarded as being at a similar level.
This interesting phenomenon can be reasonably explained as follows. As illustrated in Figure 4, for the measurement of the OER activities in 0.1 M HClO 4 electrolyte solution in the channel flow cell (half cell), we dispersed IrO x /M-SnO 2 CLs uniformly on the Au substrate with the thickness corresponding to a ca. two-monolayer height of M-SnO 2 support particles (ca. 60 nm), intending that all catalyst particles can be in contact with the electrolyte solution. Therefore, it is expected that all of the IrO x nanocatalyst particles are able to function without any influence of the small electronic (ohmic) resistances of such thin CLs. In contrast, for the measurement of single cell (MEA) performance, the thickness of the anode CL was 10 µm (ca. 170 times thicker than that in the half cell). Consequently, electrons generated at the IrO x nanoparticles in the OER must be transported in the CL to the current collector (Pt/Ti), even though protons can be effectively supplied to the IrO x surface through the electrolyte binder (ionomer) network. Hence, the higher the σ app, catalyst value (lower R ohm, cell, obs ) is, the lower the OER overvoltage will be in the single cell, due to an effective utilization of the IrO x nanocatalyst particles on the M-SnO 2 support. thickness of the ionomer (volume ratio of ionomer to support, I/S), primary and secondary pore volumes, etc. While an effect of I/S on the performance of IrO2/TiO2 anode has been reported recently [52], more comprehensive research is necessary to optimize the single cell performance toward the near-ideal value evaluated in the half cell, together with high durability. Durability testing of single cells with IrOx/Sb-SnO2 anode catalyst is in progress in our laboratory.

Preparation and Characterization of IrOx/M-SnO2 Catalysts
The IrOx/M-SnO2 catalysts were prepared in the similar manner described in our previous paper [32]. Briefly, Sn0.96Nb0.04O2−δ, Sn0.975Ta0.025O2−δ, and Sn0.95Sb0.05O2−δ (projected composition, where δ is the mole fraction of oxygen deficiencies) with the fused-aggregate network structure were synthesized by the flame pyrolysis method [29]. The amount of each dopant (Nb, Ta, and Sb) was chosen to provide the highest electronic conductivity in SnO2 [29][30][31]. The surface areas of the doped SnO2 supports were measured by the BET adsorption method (BELSORP-mini, Nippon BEL Co., Osaka, Japan). IrOx nanoparticles were uniformly dispersed on the doped SnO2 supports by the colloidal method. The amounts of iridium (Ir 0 + Ir 4+ ; excluding the amount of oxygen in IrOx) loaded on the supports were quantitatively analyzed by use of an inductively-coupled plasma mass analyzer (ICP-MS; 7500CX, Agilent Technologies Inc., Tokyo, Japan) after dissolving the IrOx completely by the alkaline carbonate-fusion method.
The IrOx/M-SnO2 catalysts were observed by TEM (H-9500, operated at 200 kV, Hitachi High-Technologies Co., Tokyo, Japan). The average diameter and size distributions of the loaded IrOx particles were estimated from ca. 300 particles in more than six TEM images with 150 × 150 nm areas. To estimate the content of Ir 4+ , the electronic states of iridium in the IrOx/M-SnO2 were characterized by XPS (JPS-9010, JEOL Co., Ltd., Tokyo, Japan) with Mg-Kα radiation (see Figure S3). The apparent electrical conductivities of the M-SnO2 supports and IrOx/M-SnO2 catalysts were measured by the same method described in a previous paper [33]. The conventional catalyst (IrO2 and Pt black) was observed by SEM (SU9000, operated at 30 kV, Hitachi High-Technologies Co., Tokyo, Japan) and TEM. As is clear from Figure 4b, other essential factors are the transport rates of protons and oxygen in the ionomer coated on the catalysts, in addition to the O 2 gas diffusion rate in the CL. Similar to the case of CLs in PEFC [51], it is very important to control the microstructure of the CLs, i.e., thickness of the ionomer (volume ratio of ionomer to support, I/S), primary and secondary pore volumes, etc. While an effect of I/S on the performance of IrO 2 /TiO 2 anode has been reported recently [52], more comprehensive research is necessary to optimize the single cell performance toward the near-ideal value evaluated in the half cell, together with high durability. Durability testing of single cells with IrO x /Sb-SnO 2 anode catalyst is in progress in our laboratory.

Preparation and Characterization of IrO x /M-SnO 2 Catalysts
The IrO x /M-SnO 2 catalysts were prepared in the similar manner described in our previous paper [32]. Briefly, Sn 0.96 Nb 0.04 O 2−δ , Sn 0.975 Ta 0.025 O 2−δ , and Sn 0.95 Sb 0.05 O 2−δ (projected composition, where δ is the mole fraction of oxygen deficiencies) with the fused-aggregate network structure were synthesized by the flame pyrolysis method [29]. The amount of each dopant (Nb, Ta, and Sb) was chosen to provide the highest electronic conductivity in SnO 2 [29][30][31]. The surface areas of the doped SnO 2 supports were measured by the BET adsorption method (BELSORP-mini, Nippon BEL Co., Osaka, Japan). IrO x nanoparticles were uniformly dispersed on the doped SnO 2 supports by the colloidal method. The amounts of iridium (Ir 0 + Ir 4+ ; excluding the amount of oxygen in IrO x ) loaded on the supports were quantitatively analyzed by use of an inductively-coupled plasma mass analyzer (ICP-MS; 7500CX, Agilent Technologies Inc., Tokyo, Japan) after dissolving the IrO x completely by the alkaline carbonate-fusion method.
The IrO x /M-SnO 2 catalysts were observed by TEM (H-9500, operated at 200 kV, Hitachi High-Technologies Co., Tokyo, Japan). The average diameter and size distributions of the loaded IrO x particles were estimated from ca. 300 particles in more than six TEM images with 150 × 150 nm areas. To estimate the content of Ir 4+ , the electronic states of iridium in the IrO x /M-SnO 2 were characterized by XPS (JPS-9010, JEOL Co., Ltd., Tokyo, Japan) with Mg-Kα radiation (see Figure S3). The apparent electrical conductivities of the M-SnO 2 supports and IrO x /M-SnO 2 catalysts were measured by the same method described in a previous paper [33]. The conventional catalyst (IrO 2 and Pt black) was observed by SEM (SU9000, operated at 30 kV, Hitachi High-Technologies Co., Tokyo, Japan) and TEM.

Evaluation of OER Activities of Catalysts in Electrolyte Solution
The polarization properties of the IrO x /M-SnO 2 catalysts were examined by a channel flow electrode cell technique [32]. The electrolyte solution used was 0.1 M HClO 4 , which was purified in advance by conventional pre-electrolysis [53]. The working electrode consisted of Nafion ® -coated IrO x /M-SnO 2 particles uniformly dispersed on an Au substrate with a geometric area of 0.04 cm 2 . The amount of the Ir catalyst loaded was maintained constant at 5 µg(Ir) cm −2 . The amounts of Nb-SnO 2 , Ta-SnO 2 , and Sb-SnO 2 supports thus loaded on the Au substrate were 39, 43, and 40 µg cm −2 , respectively, which corresponds to a ca. two-monolayer height of the M-SnO 2 support particles with an average diameter of ca. 30 nm. A mixture of commercial IrO 2 (Tokuriki Honten Co., Ltd., Tokyo, Japan) and Pt black (Ishifuku Metal Industry Co., Ltd., Tokyo, Japan) was used as a reference with 100 µg cm −2 (Ir + Pt; 1:1 mass ratio). All electrode potentials are referred to the reversible hydrogen electrode, RHE.
The OER activities of the IrO x /M-SnO 2 catalysts were evaluated by linear sweep voltammetry (LSV) at a sweep rate of 10 mV s −1 and 80 • C. To minimize the effect of O 2 bubbles, the 0.1 M HClO 4 electrolyte solution was supplied to the flow channel at a constant flow rate of 160 cm s −1 . To subtract iR loss from the polarization curve, the AC impedance of the electrolyte solution was measured by a frequency response analyzer (VersaSTAT 4, Princeton Applied Research, Berwyn, PA, USA) with a modulation amplitude of 10 mV in the frequency range from 10 kHz to 1 Hz.

Evaluation of Single Cell Performances
CCMs were prepared as follows. First, the anode catalyst ink was prepared by mixing the IrO x /M-SnO 2 powder, water, ethanol, and Nafion ® binder solution (DE521, Du Pont Co., Tokyo, Japan) as the ionomer in a ball-mill for 30 min. The cathode catalyst ink was prepared from commercial Pt/GCB (Pt 50 wt%, TEC10EA50E, Tanaka Kikinzoku Kogyo, Tokyo, Japan). The I/S was adjusted to 0.7 (dry basis) in each ink. Then, the catalyst inks were directly sprayed onto the Nafion ® membrane (thickness 50 µm, NRE 212, Du Pont Co., Tokyo, Japan) by the pulse-swirl-spray technique (PSS, Nordson Co., Tokyo, Japan) to prepare the CCM with an active geometric area of 25 cm 2 . The CCMs were hot-pressed at 140 • C and 2.5 MPa for 3 min. The Ir loading amount for the anode CL was 0.11 mg(Ir) cm −2 , and the Pt loading amount for the cathode CL was 0.35 ± 0.02 mg(Pt) cm −2 . As a reference, a conventional anode catalyst (mixture of IrO 2 and Pt black, 1:1 mass ratio) with 2.66 mg(Ir + Pt) cm −2 and a Pt black cathode catalyst with 2.01 mg(Pt) cm −2 were employed. The CCM was sandwiched by two gas diffusion layers (GDLs); a Pt-plated Ti mesh (Bekaert Toko Metal Fiber Co., Ltd., Ibaraki, Japan) for the anode and a carbon fiber paper with microporous layer (25BC, SGL Carbon Group Co., Ltd., Tokyo, Japan) for the cathode. The MEA thus prepared was mounted into a single cell holder (Japan Automobile Research Institute standard cell) with ribbed single serpentine flow channels.
Ultrapure water was circulated at a flow rate of 40 mL min −1 for the anode. Hydrogen gas was purged to the cathode. I-E curves were measured galvanostatically at 80 • C under steady-state conditions. The ohmic resistance of the cell was measured by a digital AC milliohmmeter (Model 3566, Tsuruga Electric, Co., Osaka, Japan) at 1 kHz during the operation.
The thickness of the anode CL was observed after preparation of a cross-sectional sample of the CCM by SIM in a focused ion beam system (FIB, FB-2200, Hitachi High-Technologies Co., Ltd., Tokyo, Japan).

Conclusions
The polarization performances of the IrO x /M-SnO 2 (M= Nb, Ta, Sb) anode catalysts with fused-aggregate network structures were examined for the OER in both a half cell (0.1 M HClO 4 ) and a single cell with a Nafion ® membrane at 80 • C. These catalysts exhibited similar high values of MA for the OER, regardless of the values of σ app, catalyst in the half cell, whereas the E cell decreased with decreasing R ohm, cell, obs, catalyst in the single cell tests. In addition to the reduction of the iR loss, the predominant reduction of the anodic overvoltage is ascribed to the increased effective utilization of IrO x nanocatalyst particles supported on M-SnO 2 with higher σ app, catalyst . Specifically, a single cell exhibited a promising performance E cell = 1.61 V (voltage efficiency of 92%) at 1 A cm −2 and 80 • C with the use of an IrO x /Sb-SnO 2 anode (0.11 mg(Ir) cm −2 ) and Pt/GCB cathode (0.35 mg(Pt) cm −2 ).

Supplementary Materials:
The following are available online at http://www.mdpi.com/2073-4344/9/1/74/s1, Figure S1. TEM images and particle size distribution histograms for IrO x /M-SnO 2 (M = Nb, Ta, and Sb) catalysts; Figure S2. SEM and TEM images and particle size distribution histograms for a conventional catalyst; Figure  S3. XP spectra of IrO x /M-SnO 2 (M = Nb, Ta, and Sb) catalysts; Figure S4. I-E curves of single cells, in which the current is shown as the apparent MA per mass of iridium metal; Figure S5. SIM image of the cross-section at the anode for the CCM with IrO x /Sb-SnO 2 catalyst; Appendix S1. Calculation method for the amount of the M-SnO 2 supports; Appendix S2. Calculation method for specific surface areas of IrO 2 ; Table S1. Diameters of IrO 2 (d IrO2 ) and specific surface areas of IrO 2 (S IrO2 ).