Effect of mass transfer on the oxygen reduction reaction catalyzed by platinum dendrimer encapsulated nanoparticles

Abstract

Here we report on the effect of the mass transfer rate (k_t) on the oxygen reduction reaction (ORR) catalyzed by Pt dendrimer-encapsulated nanoparticles (DENs) comprised of 147 and 55 atoms (Pt₁₄₇ and Pt₅₅). The experiments were carried out using a dual-electrode microelectrochemical device, which enables the study of the ORR under high k_t conditions with simultaneous detection of H₂O₂. At low k_t (0.02 to 0.12 cm s^-1) the effective number of electrons involved in ORR, n_eff, is 3.7 for Pt₁₄₇ and 3.4 for Pt₅₅. As k_t is increased, the mass-transfer-limited current for the ORR becomes significantly lower than the value predicted by the Levich equation for a 4-electron process regardless of catalyst size. However, the percentage of H₂O₂ detected remains constant, such that n_eff barely changes over the entire k_t range explored (0.02 cm s^-1). This suggests that mass transfer does not affect n_eff, which has implications for the mechanism of the ORR on Pt nanoparticles. Interestingly, there is a significant difference in n_eff for the two sizes of Pt DENs (n_eff = 3.7 and 3.5 for Pt₁₄₇ and Pt₅₅, respectively) that cannot be assigned to mass transfer effects and that we therefore attribute to a particle size effect.

Here we report on the effect of mass transfer rate (k_t) on the oxygen reduction reaction (ORR) catalyzed by Pt dendrimer-encapsulated nanoparticles (DENs) comprised of 147 and 55 atoms (Pt₁₄₇ and Pt₅₅, respectively). The experiments were carried out in the dual-electrode microelectrochemical device shown in Fig. 1 (1). This configuration enables the study of the ORR under high k_t conditions with simultaneous detection of H₂O₂. This is important because the methods currently used to study ORR at high k_t, such as ultramicroelectrodes (UMEs) (2), nanoelectrodes (3), and microjet electrodes (4) do not incorporate H₂O₂ detection. Accordingly, ORR data is generally interpreted based on steady-state, mass-transfer-limited currents alone. Using the microelectrochemical device, two principal effects are evident as k_t is increased. First, the ORR current becomes significantly smaller than predicted by the Levich equation (5). Second, the effective number of electrons involved in the ORR, n_eff, remains constant over the entire k_t range, for both Pt₁₄₇ and Pt₅₅. We draw three conclusions from these observations: (i) The decrease in ORR current at high k_t is caused by kinetic effects; (ii) increasing k_t does not affect the mechanism of the ORR catalyzed by Pt DENs; and (iii) there is a difference in n_eff between Pt₁₄₇ and Pt₅₅ that cannot be attributed to mass-transfer effects.

Fig. 1.

(A) Schematic representation of the hybrid PDMS/quartz microelectrochemical device used for ORR studies. Solution is introduced to the microchannel using a syringe pump connected to the reservoir on the left. A combined reference/counter electrode is placed in the other reservoir so that it is in close proximity to PPC microband electrodes. Note that this drawing is not to scale. In the actual device, the length of the microband electrodes is typically 40 μm with a gap of 15 μm between them. (B) Optical micrograph of the PPC microbands and the microchannel after platinization of the collector electrode and electrochemical deposition of Pt DENs onto the generator electrode. The outlet reservoir is visible to the right of the electrodes.

The electrochemical ORR has been the subject of extensive research and has been reviewed on several occasions (6–10). The generally accepted mechanism for the ORR at a metallic surface is a multielectron reaction with intermediates as shown in Scheme 1 (2, 7, 9, 11). Specifically, at the electrode surface, O₂ can be reduced directly to H₂O or H₂O₂ in 4- or 2-electron processes, respectively. Subsequently, electrogenerated H₂O₂ can disproportionate into H₂O and O₂, or by addition of two electrons, form H₂O. According to Scheme 1, the yield of H₂O₂, quantified by n_eff, is determined by the relative rates of steps c, d, and e, as well as the rate of mass transfer of H₂O₂ away from the electrode surface (2).

Scheme 1.

The development of O₂/H₂ polymer electrolyte membrane fuel cells (PE-MFCs) as sustainable energy sources (12, 13) has led to increased interest in developing efficient, non-Pt catalysts for the ORR (14–18). In spite of this, Pt is still considered to be the most effective catalyst for the ORR, even though it is not optimal for this application due to its high overpotential and its low Earth abundance. In the absence of viable alternatives, interest has been focused on reducing the amount of Pt required for fuel cell applications. There are two ways to do this: dilute it with a second metal or decrease the size or total number of Pt monometallic catalyst particles. However, a decrease in Pt surface coverage (or, correspondingly, an increase in k_t) has been associated with slower ORR kinetics (19, 20) and an increase in the amount of H₂O₂ produced (21–23). The presence of H₂O₂ is a major concern for fuel cell applications due to its corrosive nature and because it lowers the maximum attainable voltage (24, 25).

A number of studies have been reported that focus on the role of mass transfer in ORR efficiency. The analytical tools used for this purpose include the rotating disk and rotating ring-disk electrodes (RDE and RRDE) (21, 26–28), UMEs (2, 29), nanoelectrodes (3), microjet electrodes (4), and microfluidic thin-layer cells (22, 23). In most of these studies, k_t is increased by decreasing the surface coverage of the catalyst (21–23, 27, 28). For example, Behm and coworkers studied the ORR using two-dimensional arrays of 100-nm Pt islands immobilized on glassy carbon (GC) electrodes (22, 23). The key finding was that the amount of electrogenerated H₂O₂ increased with decreasing catalyst coverage. A “desorption + readsorption + reaction” mechanism was proposed to explain this effect (26, 30). Under this assumption, more H₂O₂ is obtained at high k_t (low catalyst loading or high flow rates), due to lower probability for H₂O₂ readsorption and further reaction on different Pt sites (Scheme 1) (26, 30). Kucernak and coworkers studied the effect of high k_t conditions (up to 2 cm s^-1) on the ORR catalyzed by single, submicron Pt particles affixed to the ends of nanoelectrodes (3). It was found that the steady-state mass-transfer-limited current for the ORR decreased with increasing k_t. This was attributed to a shift in n_eff from 4 at k_t = 0.01 cm s^-1 to 3.5 at k_t = 2 cm s^-1 (3). However, because a collector electrode was not present, it was not possible to correlate these results to the presence of H₂O₂.

Here we study the effect of k_t on the ORR using Pt DENs having diameters of 1–2 nm and low (< 0.3 nm) size polydispersity. In the microelectrochemical device shown in Fig. 1, Pt₁₄₇ or Pt₅₅ DENs are electrochemically immobilized onto a pyrolyzed photoresist carbon (PPC) generator electrode (1). This results in robust attachment of approximately 1 monolayer of Pt DENs onto the PPC surface (1). Importantly, a collector electrode is present downstream for amperometric detection of H₂O₂. By controlling the flow rate through the microchannel, it is possible to vary k_t between 0.02 and 0.12 cm s^-1. Indeed, such high mass transfer conditions are not achievable using the RRDE or UMEs with diameters larger than 2 μm. At low k_t (0.02 cm s^-1), which is similar to that achieved at the upper end of k_t values accessible with the RRDE, n_eff is 3.7 for Pt₁₄₇ and 3.4 for Pt₅₅, in agreement with our previous findings (31). As k_t increases, the mass-transfer-limited current for the ORR becomes significantly lower than the value predicted by the Levich equation for a 4-electron process for both Pt₁₄₇ and Pt₅₅. However, the percentage of H₂O₂ detected remains constant, such that n_eff does not change over the entire k_t range employed. Interestingly, there is a significant difference in n_eff for the two sizes of Pt DENs (average n_eff = 3.7 and 3.5 for Pt₁₄₇ and Pt₅₅, respectively) that cannot be assigned to mass transfer effects.

The results reported here are significant for several reasons. First, we show that simultaneous measurement of the H₂O₂ current is vital for correct interpretation of ORR data. Second, while others have previously shown that the ORR catalyzed by large Pt nanoparticles is affected by changes in k_t (22, 23) we show here that this does not apply for smaller Pt materials. Finally, we show that the size of very small Pt nanoparticles (diameters < 2 nm) has an effect on the mechanism of the ORR, and hence its product distribution. These observations are relevant to the development of efficient Pt electrocatalysts for the ORR.

Results and Discussion

Evaluation of ORR Activity for Pt₁₄₇ and Pt₅₅ at Low Mass Transfer Rates.

The ORR activity of Pt₁₄₇ and Pt₅₅ DENs was evaluated using the poly(dimethylsiloxane) (PDMS)/quartz hybrid microelectrochemical device shown in Fig. 1 and discussed in detail elsewhere (1). Following platinization of the collector (Fig. 1B) and immobilization of Pt DENs on the generator (1), the ORR measurements were carried out in air-saturated 0.10 M HClO₄ electrolyte solution. To evaluate the ORR kinetics at low k_t, solution was flowed through the device at 0.05 μL min^-1 (0.04 cm s^-1), corresponding to a k_t value of 0.02 cm s^-1. This is on the same order as that achieved using an RRDE.

We previously used the RDE to evaluate the ORR characteristics of Pt DENs having different sizes (31). The results showed that the steady-state limiting current for Pt₁₄₇ was higher than for Pt₅₅, an effect that was tentatively assigned to H₂O₂ becoming a significant product of the ORR for the smaller catalytic particles (31). We pointed out that if this were correct, it was a rare case of a very slight change in particle size resulting in a shift in product distribution. Fig. 2 shows generation-collection linear scan voltammograms (LSVs) for the ORR obtained using the microelectrochemical device for Pt₁₄₇ (blue lines) and Pt₅₅ (red lines) DENs at k_t = 0.02 cm s^-1. The potential of the generator electrode was swept from +0.80 to -0.10 V (vs. Ag/AgCl) at 50 mV s^-1. The potential of the collector electrode was held at +1.10 V, where we observed reliable and stable mass-transfer-limited currents for H₂O₂ oxidation (1). The onset of catalytic current is at ∼+0.60 V, and a steady-state, mass-transfer-limited current is apparent over the range 0.00 to -0.10 V. There is a slight decrease in the ORR current at potentials more negative than 0 V due to hydrogen adsorption onto the catalyst (32). In addition, the adsorption of hydrogen is accompanied by an increase in hydrogen peroxide formation (manifested as an increase in the collector electrode current), an effect previously observed by others (32). To minimize artifacts arising from hydrogen adsorption, generator-collector current pairs were measured at the potential where the ORR current reached its maximum value.

Fig. 2.

Generation (upper curves) and collection (lower curves) LSVs obtained using Pt₁₄₇ (blue lines) and Pt₅₅ (red lines) DENs in air-saturated, aqueous 0.10 M HClO₄. The potential of the generator electrode was scanned from +0.80 to -0.10 V (vs. Ag/AgCl) while holding the potential of the platinized collector at +1.10 V. The flow rate was 0.05 μL min^-1 corresponding to k_t = 0.02 cm s^-1, and η was 40.0%. The scan rate was 50 mV s^-1.

Importantly, the ORR limiting current for Pt₁₄₇ DENs (55 nA) is noticeably higher than for Pt₅₅ DENs (37 nA), although the LSVs have the same general shape. The downstream collector electrode makes it possible to determine the amount of H₂O₂ produced during ORR, at the same time the mass-transfer-limited currents are measured at the generator. The LSVs in Fig. 2 show that the amount of H₂O₂ oxidized at the collector electrode is higher for Pt₅₅ DENs. As we hypothesized in our earlier publication, this indicates that the peroxide pathways play a more important role in the ORR catalyzed by smaller nanoparticles (31).

The value of n_eff was determined using the ratio of the steady-state mass-transfer-limited current for the ORR (i_gen) and the oxidation current for H₂O₂ (i_col); that is, n_eff = 4 - [(2i_col)/(ηi_gen)], where η represents the collection efficiency of the device (details on how η is calculated are provided in Materials and Methods) (33). In Fig. 2, i_col has been background subtracted against the current value obtained by holding the collector electrode at a potential (1.10 V), where the ORR does not proceed. The measured η at 0.05 μL min^-1 was 40.0%. Note that the collector current does not exhibit a clear, steady-state plateau (vide supra), and, as such, i_col was measured at the potential where i_gen attained its maximum value (0 V). The calculated values of n_eff are 3.7 ± 0.1 and 3.4 ± 0.1 (average of three independent measurements), for Pt₁₄₇ and Pt₅₅, respectively. Using these values for n_eff, i_gen can be estimated using the Levich equation (Eq. 1) (5), as follows:

[1]

Here, F is the Faraday constant, c_O2 and D_O2 are the bulk concentration and diffusion coefficient, respectively, of O₂ [c_O2 = 0.20 mM, (34) D_O2 = 1.67 × 10^-5 cm² s^-1] (35), V_f is the volume flow rate (0.05 μL min^-1), x_e and w are the length (40 μm) and the width (100 μm) of the electrode, respectively, d is the width of the microchannel (100 μm), and h is the height of the microchannel (20 μm). Using the previously determined n_eff values, i_gen is calculated to be 55 nA for Pt₁₄₇ and 50 nA for Pt₅₅. For Pt₁₄₇, the calculated value is in excellent agreement with the measured i_gen of 55 nA. However, for Pt₅₅, the calculated value is substantially higher than the experimentally determined i_gen (37 nA). Although we do not understand this result, such inconsistencies are common in the ORR RRDE literature (36, 37). It is possible that the relatively high scan rate employed (50 mV s^-1) also contributes to this effect.

ORR Measurements at High Mass Transfer Rates.

To learn more about the effect of the mass transfer rate of O₂ on n_eff, we acquired ORR generation collection LSVs while flowing air-saturated 0.10 M HClO₄ through the microchannel at flow rates ranging from 0.05 to 10 μL min^-1 (0.04 to 8.3 cm s^-1). This corresponds to k_t values from 0.02 to 0.12 cm s^-1. This range is considerably higher than that accessible with the RRDE. Indeed, to achieve the upper limit of k_t = 0.12 cm s^-1, a RRDE would have to be rotated at 150,000 rpm, which is well above the typical maximum rotation rate of 7,000 rpm. Fig. 3A shows a plot of i_gen and simultaneously measured i_col values obtained for Pt₁₄₇ (blue dots) and Pt₅₅ (red dots) as a function of flow. Values for i_gen and i_col were obtained from both low-to-high and high-to-low flow rates, and were found to be independent of this parameter. This confirms that the Pt coverage is constant throughout the experiments (that is, that the Pt DENs remain attached to the electrode surface). In Fig. 3A, i_col was normalized with respect to η for each flow rate. Recall that i_gen - i_col current pairs were measured at the potential where i_gen reached its maximum value and i_col values were background subtracted (vide supra). Fig. 3A shows that, for both Pt₁₄₇ and Pt₅₅ DENs, as k_t increases, i_gen increases and begins to level off at flow rates higher than 2 μL min^-1. The same pattern is observed for i_col. Furthermore, across the entire flow-rate range, the ORR limiting current for Pt₁₄₇ remains higher than that for Pt₅₅.

Fig. 3.

(A) Plot of i_gen (ORR mass-transfer-limited current) and simultaneously acquired i_col (H₂O₂ oxidation current) for Pt₁₄₇ (blue circles) and Pt₅₅ (red circles) measured while flowing air-saturated, aqueous 0.10 M HClO₄ through the microchannel at flow rates ranging from 0.05 to 10 μL min^-1 (k_t from 0.02 to 0.12 cm s^-1). Note that i_col was normalized with respect to η (η values for 0.05, 0.075, 0.1, 0.25, 0.5, 1, 2.5, 5, and 10 μL min^-1 flow rates were 40.0, 37.2, 36.3, 34.7, 34.2, 34.0, 33.5, 33.3 and 32.7%, respectively). (B) Plot of i_gen from A vs. and k_t, for Pt₁₄₇ (blue circles) and Pt₅₅ (red circles). The dashed black lines were calculated using the Levich equation (Eq. 1) for a 4-electron (upper line) and a 2-electron (lower line) process.

It is interesting to compare the experimental values of i_gen with the theoretical values predicted by the Levich equation (Eq. 1) for 4- and 2-electron ORR processes. In Fig. 3B, the i_gen values from Fig. 3A are plotted vs. and k_t. The current for 4- and 2-electron ORR processes are plotted on the same graph (dashed lines). For Pt₁₄₇, Fig. 3B indicates that as k_t increases, i_gen appears to drift from the 4-electron to the 2-electron ORR pathway (Scheme 1). Interestingly, at V_f higher than 1 μL min^-1, the measured i_gen for Pt₅₅ is smaller than the value predicted by Eq. 1 for a 2-electron process, an observation that will be discussed later. Importantly, for both Pt₁₄₇ and Pt₅₅ DENs, the plot of i_gen vs. is not linear. A similar trend in the ORR limiting current was observed during Pt-catalyzed ORR studies using UMEs (2) and nanoelectrodes (3), and for Cu using hydrodynamic electrodes (4). These results were interpreted in terms of O₂ reduction not following a simple 4- or 2-electron transfer process, but instead undergoing a transition from 4 electrons at the lowest k_t values towards 2 electrons as the mass transfer rate increased. It is important to note that in these earlier reports a collector electrode was not used, and therefore the amount of H₂O₂ generator could not be measured directly and then correlated to the ORR current. Because we (1) and others (36, 37) have noted that the mass-transfer-limited current for the ORR is often unreliable for the determination of n_eff, an accurate calculation requires the ratio of i_col to i_gen. Here, the concomitant measurement of H₂O₂ oxidation current allows the calculation of n_eff at each k_t value. Fig. 4 shows a plot of n_eff vs. V_f and k_t, calculated for data presented in Fig. 3A. At all but the lowest flow rates, the value of n_eff (that is, the percentage of H₂O₂ collected during ORR) is nearly constant. Consequently, although the magnitude of i_gen suggests a shift in the mechanism of ORR with the change in mass transfer rate, n_eff remains constant over the entire k_t range. The average n_eff is 3.7 and 3.5 for Pt₁₄₇ and Pt₅₅, respectively, similar to the values calculated at k_t = 0.02 cm s^-1 (3.7 for Pt₁₄₇ and 3.4 for Pt₅₅). We attribute the shift in i_gen observed here, and previously by others (2–4) to kinetic effects: As k_t increases, some O₂ molecules flow past the generator electrode faster than they can be reduced. This results in a decrease in i_gen from its theoretical mass-transfer-limited value. However, the mechanism of ORR remains unaffected. Although we do not currently have a definitive explanation for the difference in n_eff between Pt DENs of different sizes, it is possible that the decrease in particle size leads to a different O₂ binding mechanism onto the Pt surface, which ultimately affects the ORR pathway. For small Pt NPs, molecular O₂ can bind to corner and edge sites in the so-called single-bond binding geometry, in which one oxygen atom is against the metal surface and the second is exposed to the solvent. Such single-bond binding can lead to the formation of OOH groups, and ultimately, H₂O₂. Experimental and theoretical work is currently underway to fully understand this effect.

Fig. 4.

The value of n_eff calculated from the experimental results in Fig. 3A for Pt₁₄₇ (blue circles) and Pt₅₅ (red circles) vs. and k_t. The black lines represent the average values of n_eff across the entire mass transfer range shown.

Conclusions

Using a dual-electrode microelectrochemical device we investigated the effect of mass transfer on the ORR catalyzed by Pt₁₄₇ and Pt₅₅ DENs. At low values of k_t (0.02 cm s^-1), n_eff was 3.7 for Pt₁₄₇ and 3.4 for Pt₅₅, in agreement with previous findings obtained using a RDE (31). As k_t was increased from 0.02 to 0.12 cm s^-1, the measured ORR mass-transfer-limited current for both Pt₁₄₇ and Pt₅₅ was smaller than the value predicted by the Levich equation, similar to results obtained by others (2–4). Previously, this was interpreted as a transition of the ORR from four electrons at the lowest k_t values towards two electrons as the mass transfer rate increased. Here, we observed that the percentage of H₂O₂ produced was unaffected by k_t, and n_eff was nearly constant across the entire k_t range. Hence, the concomitant measurement of H₂O₂ allows us to conclude that the decrease in the expected mass-transfer-limited current for ORR at high k_t is in fact due to kinetic effects, rather than a change in the ORR mechanism.

Importantly, we also observed an unambiguous nanoparticle size effect on the mechanism of ORR, with an average n_eff higher for Pt₁₄₇ (3.7) than for Pt₅₅ (3.5). It is certainly surprising that a change in particle size of just 100 atoms is able to shift the reaction mechanism. Accordingly, we are investigating this effect in more detail, as it may have an important role to play in ORR electrocatalysis specifically and electrocatalysis generally. The results of these studies will be reported in due course.

Materials and Methods

Chemicals.

PDMS channels were prepared using a Sylgard 184 elastomer kit obtained from K. R. Anderson, Inc. Quartz microscope slides (25 mm × 75 mm, 1 mm thick) were purchased from Technical Glass Products. Positive-tone photoresist (AZ 1518) and developer (AZ 400K) were purchased from Capitol Scientific, Inc. Sixth-generation poly(amidoamine) dendrimers terminated with hydroxyl groups (G6-OH) were purchased as a 5% (wt/wt) solution in methanol (Dendritech, Inc.). Prior to use, the methanol was removed under vacuum and the dendrimer was reconstituted in water at a concentration of 100 μM. K₂PtCl₄ (99.99%), NaBH₄ (99.99%), and LiClO₄ (99.99%) were purchased from Sigma-Aldrich, Inc.; HClO₄ (Ultrex II) was from J. T. Baker and ferrocenemethanol (FcMeOH, 97%) was from Acros Organics. These reagents were used without further purification. Deionized water having a resistivity of 18.2 MΩ cm was used for all experiments (Milli-Q gradient system, Millipore).

Preparation of Microelectrochemical Devices.

The procedure for fabricating PPC microband electrodes is described in detail elsewhere (1). Briefly, PPC microbands, 40 μm wide and separated by a 15 μm gap, were fabricated by slow pyrolysis of patterned AZ 1518 photoresist at 1,000 °C under a forming gas of 5% H₂ and 95% N₂ (1). The thickness of the resulting PPC features, measured using profilometry, was 300 nm. PDMS microfluidic channels, 6 mm long, 100 μm wide and 20 μm high, were prepared using a previously described replica micromolding method (1, 38). Two reservoirs (1.0 mm diameter) were punched at the ends of each channel to accommodate introduction of solution. The PDMS layer and quartz substrate were exposed to an air plasma (60 W, model PDC-32G, Harrick Scientific) for 45 s, joined together under an optical microscope, and then placed over a hot plate at 80 °C for 5 min. A syringe pump (Pump 11, Pico Plus Elite, Harvard Apparatus), connected to the inlet (left side) reservoir using Teflon tubing, was used to push solution through the microelectrochemical device.

Preparation of Pt DENs.

Pt DENs comprised of 147 and 55 atoms were prepared according to our previously published procedure (31, 39). A 0.10 mM aqueous solution of G6-OH was mixed with sufficient aqueous 0.10 M K₂PtCl₄ such that the final Pt²⁺-to-dendrimer ratios were 147∶1 or 55∶1. After stirring for three days (40), a 10-fold molar excess of aqueous 1.00 M NaBH₄ was added to reduce Pt²⁺. This solution was kept capped for 24 h to maximize reduction of Pt²⁺ (40). Note that we have previously shown that reduction of the Pt^2+/G6-OH precursor by does not yield fully reduced Pt DENs (40, 41). However, electrochemical cycling does result in complete reduction (vide infra) (40, 41). The average diameters and size distributions of the Pt DENs were 1.7 nm and 1.4 nm for Pt₁₄₇ and Pt₅₅, respectively (31). After synthesis, the quality of the Pt DENs was verified by UV-vis absorption spectroscopy (31, 42). Before using for electrochemical experiments, Pt DENs were dialyzed for 24 h in 4.0 L of Millipore water (Milli-Q Gradient PF-06073) to remove salts and other impurities. After dialysis, but prior to immobilization onto PPC electrodes, the DEN solution was mixed with sufficient LiClO₄ to yield a 0.10 M solution.

Electrochemistry.

The microelectrochemical device was configured in a three-electrode arrangement, comprising the two PPC microbands as working electrodes. A Ag/AgCl reference electrode, which also acted as a counter electrode was positioned in the outlet (right-hand side) reservoir (Scheme 1). All potentials are referenced to Ag/AgCl (3.4 M KCl, model 66-EE009 “no-leak” Ag/AgCl, Dionex). LSV with two working electrodes was carried out using a computer-based bipotentiostat (Model CHI760B potentiostat, CH Instruments). All electrochemical experiments were performed at 22 ± 1 °C.

Prior to the immobilization of Pt DENs on the generator, Pt was electrodeposited onto the PPC collector electrode to increase its sensitivity for amperometric H₂O₂ detection (Fig. 1B) (1). After platinization, LSVs were acquired at the PPC generator electrode in the ORR region of interest (+0.80 to -0.10 V vs. Ag/AgCl) using aqueous 0.10 M HClO₄, to check that the generator electrode had not been contaminated with Pt during the platinization step. Pt DENs were then immobilized on the PPC generator using a previously reported procedure (1, 31, 39). Briefly, a 10.0 μM Pt DEN solution containing 0.10 M LiClO₄ was flowed through the microchannel at 0.05 μL min^-1 while the electrode potential was swept three times between +0.10 and +1.00 V (vs. Ag/AgCl) at 10 mV s^-1. We previously showed that this results in robust attachment of approximately 1 monolayer of Pt DENs onto both GC and PPC surfaces (1, 31). Before acquiring ORR generation collection LSVs, the DEN-modified electrodes were scanned 20 times at 100 mV s^-1 between +1.25 V and -0.20 V (vs. Ag/AgCl) in air-saturated 0.10 M HClO₄ solution. As we have shown previously, these conditioning scans (43) are commonplace and required to achieve stable and reproducible LSVs (39).

Upon completion of the ORR measurements, η was determined for each microelectrochemical device using the following method. Generation collection LSVs were recorded while pumping an aqueous solution containing 0.10 mM FcMeOH and 0.10 M KNO₃ through the microchannel at flow rates ranging from 0.05 to 10 μL min^-1 (0.04 to 8.3 cm s^-1). At each flow rate, the values of η were calculated as the ratio of the steady-state mass-transfer-limited currents at the collector and generator electrodes, for the electrolysis of FcMeOH. The calculated values of η ranged from 32.7–40.0%.