Reaction Kinetics of Gas–Solid Exchange Using Gas Phase Isotopic

Research Article pubs.acs.org/acscatalysis

Reaction Kinetics of Gas−Solid Exchange Using Gas Phase Isotopic Oxygen Exchange Yi-Lin Huang, Christopher Pellegrinelli, and Eric D. Wachsman* University of Maryland Energy Research Center, University of Maryland, College Park, Maryland 20742, United States S Supporting Information *

ABSTRACT: The oxygen transport kinetics of heterogeneous gas−solid exchange has been investigated on the basis of a two-step reaction mechanism, linking surface catalysis to solid-state self-diffusion, via gas phase isotopic oxygen exchange on the mixed ionic electronic conductor (MIEC) La0.6Sr0.4Co0.2Fe0.8O3−x (LSCF) and electronic conductor (La0.8Sr0.2)0.95MnO3±x (LSM). The catalytic activity of LSCF is higher than that of LSM toward the elementary step of oxygen dissociation, likely caused by a higher vacancy concentration. The apparent activation energy for surface exchange of LSCF is lower than values obtained from bulk characterization techniques. The diffusion coefficient (D) for LSM at different temperatures shows a huge deviation from literature values, and an alternate exchange mechanism has been proposed. The fast transport pathway is attributed to the substoichiometry of LSM in the near surface region. These results have significant implications for the improvement of the oxygen reduction reaction for the design of higher-performance materials and the importance and limitations of isotope exchange experimental design. KEYWORDS: isotope exchange, heterogeneous catalysis, oxygen transport, gas−solid reaction, oxygen reduction reaction

M

On the other hand, SSIKTA is a well-established technique for studying the kinetics of heterogeneous catalysis. SSIKTA is well suited for explaining reaction mechanisms and extracting kinetic properties such as the identification of intermediate species, concentrations of intermediates, available surface sites, and determination of possible reaction mechanisms. The data need to be delicately analyzed to obtain accurate results. Previous SSIKTA experiments have focused mostly on the reactions occurring at the surface of the material, but not the mass transport exchange into the catalyst. For applications such as SOFCs, membranes, solar fuels, and electrocatalysis, we are particularly interested in reactions in which gaseous oxygen is being incorporated into the catalyst. On the basis of the data analysi of in situ SSIKTA, we developed a steady-state kinetic model to comprehensively deconvolute surface reaction steps. In this work, oxygen transport mechanisms in two perovskites, La0.6Sr0.4Co0.2Fe0.8O3−x (LSCF) and (La0.8Sr0.2)0.95MnO3±x (LSM), were studied. LSM is a pure electronic conductor with negligible ionic conductivity, and LSCF is a mixed ionic electronic conductor (MIEC). In this ABO3 structure, La and Sr occupy the A sites and the B sites are filled with the transition metals Fe and Co, or Mn. Changing the oxygen partial pressure and temperature will change the concentrations of defects (oxygen vacancies) in the structure and therefore the catalytic properties of the materials. As the concentration of oxygen vacancies increases, the number of available surface sites for

ultivalent metal oxides are well-known for their high catalytic activity for electrochemical reactions with tunable defect concentrations and controllable ionic and electronic conductivities. These oxides not only catalyze the oxygen reduction reaction (ORR) but also conduct through surface diffusion or bulk diffusion. Understanding the fundamentals of the ORR mechanism and accurately obtaining kinetic rates are necessary to further improve material performance and broad en technology applications, such as low-temperature solid oxide fuel cell (SOFC) cathodes,1 electrocatalysts,2 oxygen membranes,3 metal−air batteries,4 and solar fuels.5 Most catalytic studies focus only on either the gaseous products or the process of diffusion into the porous catalysts. Few of them concentrate on the self-exchange process for the mass transfer of reactants to the solid catalyst. Therefore, it is important to use such techniques to gain a fundamental understanding of the ORR mechanism. There have been a variety of techniques aimed at gaining an understanding of the mechanisms and kinetics governing the ORR.6 Isotope exchange7 is a powerful tool for observing the kinetics of heterogeneous catalytic reactions; however, most isotope tracer experiments are limited in their scope. Typically, these experiments take one of two approaches, either ex situ isotope exchange depth profiling (IEDP) with secondary ion mass spectrometry (SIMS)8 or steady-state isotopic−kinetic transient analysis (SSIKTA)7b,d,f,9 where the focus is on the catalysis of gaseous reactants to gaseous products on the surface. Although IEDP can extract kinetic parameters based on isotopic distribution in the solid, it is limited by ex situ processing and lacks the ability to distinguish the surface reaction steps involved. © XXXX American Chemical Society

Received: May 24, 2016 Revised: July 28, 2016

6025

DOI: 10.1021/acscatal.6b01462 ACS Catal. 2016, 6, 6025−6032

Research Article

ACS Catalysis

the system maintains a fixed temperature and a fixed oxygen partial pressure. Therefore, the system is at a quasi-steady state, meaning that the total available surface sites, total concentration of oxygen vacancies, and reaction rates are all constant.7d,e,9 A one-dimensional potential energy diagram for 18O2 molecules interacting with a clean oxide surface is shown in Figure 2.11

dissociation and incorporation also increases. To investigate the mass transport across the gas−solid interface and further separate the contributions from each variable, isothermal isotope exchange (IIE)10 is performed at different oxygen partial pressures and temperatures.

■

EXPERIMENTAL SECTION La0.6Sr0.4Co0.2Fe0.8O3−x (LSCF) (Praxair) and (La0.8Sr0.2)0.95MnO3±x (LSM) (Fuel Cell Materials) powders are used. BET measurements show that the surface areas of the LSCF and LSM powders are 6.5 and 5.6 m2/g, respectively. The particle sizes from light scattering of LSCF and LSM are around 300 and 700 nm, respectively. Therefore, LSCF and LSM samples are weighed out so the surface areas are normalized to 0.1 m2 to allow for direct comparison. Then, the powder is placed in the center of a quartz tube plug-flow reactor. The flow rate is fixed at 20 SCCM using mass flow controllers. 16O and 18O are assumed to be identical, and any isotopic differences can be ignored. Particles are assumed to be uniform in size and spherical in shape. Before each IIE experiment, the sample powder is first pretreated at 800 °C in an 16O2 environment (AIRGAS, 99.999%) for 30 min to ensure that the powder surface is clean and converted completely to normal isotope oxygen (16O). After the pretreatment, the sample is brought to the temperature of interest and equilibrated in 16O2. In a separate line, an equal PO2 of isotope 18O2 (Sigma-Aldrich; 95 atom % 18O) and a small amount of argon (Airgas; 0.1%), used as an inert tracer, are flowed. Then, the gas entering the system is quickly switched from 16O2 to 18 O2 using a pneumatic valve actuator. The evolution of gas phase oxygen isotopologues is monitored using a quadrupole mass spectrometer (QMS). Isotope Exchange Theory. To identify different oxygen reaction paths,8,9 gaseous 18O2 is used as a tracer, as shown in Figure 1. The catalyst surface can be seen as a catalyst bed,

Figure 2. One-dimensional potential energy diagram for 18O2 interacting with a clean solid surface. Oxygen molecules are dissociatively adsorbed on the surface. The 18Os has two different pathways for reaction: (1) desorption, back to the gas phase, and (2) incorporation into the solid phase. The incorporation rate is dependent on 18Os overcoming the energy barrier between the surface and the bulk. The diffusion of 18Olat through the lattice can be modeled as a periodic potential well and quantified by the self-diffusion coefficient (D).

k1 describes dissociative adsorption, the rate of which is dependent on the temperature and energy barrier. To be incorporated into the catalyst lattice, the 18O atom on the surface needs to overcome the energy barrier between two-dimensional surface bonding and three-dimensional lattice bonding. k2 describes the ability of the 18O tracer atom to be incorporated into the lattice structure. After the 18O atom is incorporated, a periodic potential well, between oxygen lattice sites, describes the conduction of 18O through the lattice. The tracer diffusion coefficient (D) is used to quantify the rate of this self-diffusion. A model based on a two-step reaction mechanism across the heterogeneous gas−solid interface was built to simulate the IIE results. Details of IIE theory can be found in the Supporting Information. The coupled reactions can be shown in two elementary steps: (1) dissociative adsorption and (2) incorporation, as highlighted in Figure 1. The reaction equation for dissociative adsorption can be written as k1

O2 + 2* ↔ 2O * k −1

(1)

where the asterisk denotes an available active site and O* is an oxygen atom occupying an active site, with forward and reverse rate constants k1 and k−1, respectively. The fractions of each oxygen isotopologue are products of the surface catalytic reactions and can be expressed as a function of 16ΘO and 18ΘO, the surface coverage fractions of 16O and 18O, respectively.

Figure 1. Schematic of the two-step reaction: (1) dissociative adsorption and (2) incorporation. Gas phase products of isotope oxygen are listed. An 18O atom (yellow) was used as a tracer to help identify different oxygen reaction paths and distinguish it from the background 16 O (red).

16 32

P O2 + 34P O2 + 36P O2 =

2

ΘO + 216ΘO18ΘO + 16ΘO K1Θ 2 *

2

(2)

where K1 is the equilibrium constant for gas−surface exchange and Θ* is the fraction of available active surface sites. The reaction equation for incorporation can be written as

where 18O2 molecules are adsorbed and dissociated on the surface into single oxygen atoms (18O). The 18O atoms have the possibility of being incorporated into the lattice (solid phase) or recombining with another oxygen atom on the surface and desorb back as a gas phase molecule. The concentrations of 16 O2, 16O18O, and 18O2, corresponding to ions at m/z 32, 34, and 36, respectively, provide information about the dissociation of 18O2 on the cathode surface. Throughout an IIE experiment,

k2

O + VO ← → OO + * * k −2

(3)

where k2 and k−2 are the rate constants for the forward and reverse reactions, respectively. The surface coverage of oxygen atoms ΘO can be expressed as a function of the oxygen 6026

DOI: 10.1021/acscatal.6b01462 ACS Catal. 2016, 6, 6025−6032

Research Article

ACS Catalysis concentration in the lattice, [OO], and oxygen vacancy concentration [VO] in the near surface region. ΘO =

[OO]Θ * K 2[VO]

(4)

where K2 is the equilibrium constant for surface−solid exchange. Both dissociation and incorporation must occur to determine surface exchange kinetics using IIE. Because of the nature of the experiment, if there is no O2 dissociation, incorporation cannot be observed. If incorporation is the rate-limiting step, Rex with units of inverse seconds can be directly linked to the fundamental kinetic rate k2, the forward reaction rate for incorporation, through the concentration of oxygen vacancies, [VO].

R ex = k 2[VO]

(5)

where Rex represents the rate of oxygen exchange due to the entropic driving force under steady-state conditions.7e Rex can be converted to the surface exchange coefficient kex, widely used in solid-state kinetic literature, based on the relationship between catalysis in two and three dimensions: R ex =

Scat 3k kex = ex Vcat a

(6)

where Scat and Vcat are the surface area and volume of the catalyst particles, respectively, and a is the radius of the sphere powder. The stoichiometry of the material can be linked directly to equilibrium constants K1 and K2. K 2K11/2PO21/2 =

[OO] [VO]

(7)

A mathematical solution was developed to extract kinetic parameters based on numerical calculations using MATLAB.12 A detailed validation on the fitting is provided in the Supporting Information.

■

RESULTS AND DISCUSSION Figure 3a−e shows IIE of LSCF from 350 to 450 °C at a PO2 of 0.025 atm. Symbols are experimental data, and lines are fitting results. At 350 °C, no observable exchange occurs and all inlet 18 O2 molecules pass through the LSCF powder without any surface reactions. At this temperature, LSCF is limited by dissociation. At 375 °C, IIE of LSCF shows dissociation of a sizable portion of inlet 18O2, suggesting that LSCF has a good catalytic activity toward the dissociation of oxygen above this temperature. Via removal of the contribution of the nonexchange fraction in eqs S3−S11, the 18O flux can be expressed as a single first-order exchange. All curves of 18O exchange flux at different temperatures follow first-order reactions, suggesting that oxygen diffusion in LSCF does not limit the reaction above 375 °C, but the gas−solid surface exchange process does. Higher-temperature IIE results of LSCF at a PO2 of 0.025 atm from 500 to 800 °C are shown in Figure 4. However, it is a concern that the curves have almost identical shapes. Insensitivity to changes in temperature for IIE above 500 °C suggests that the limiting factor at these temperatures is the amount of oxygen flowing into the system, as this would have a first-order dependence.10a,b,13 Mass transfer-limited reactions can be easily misidentified as surface exchange reactions. Figure 5 shows the Arrhenius plot of kex with respect to reciprocal temperature. In the higher-temperature region

Figure 3. IIE of LSCF at a PO2 of 0.025 atm at intermediate temperatures without being limited by gas diffusion: gas phase oxygen fractions at (a) 350, (b) 375, (c) 400, (d) 425, and (e) 450 °C. The initial concentration of 18O2 (t = 0) indicates whether the dissociation limits the reactions. Symbols are experimental data points, and lines are fitting results. 6027

DOI: 10.1021/acscatal.6b01462 ACS Catal. 2016, 6, 6025−6032

Research Article

ACS Catalysis

Figure 5. Arrhenius plot of the surface exchange coefficient (kex) showing an apparent activation energy (EA) from gas phase isotopic oxygen exchange of 42.3 kJ/mol from 375 to 450 °C at a PO2 of 0.025 atm. Included are literature values from other characterization techniques: ECR on bulk14 (PO2 = 0.02−0.01 atm), IEDP-SIMS,15 and ECR on thin films16 (PO2 = 0.01−0.06 atm).

to the mass transfer limitations, rather than intrinsic material properties of LSCF.13 It is common for kinetic experiments on catalysts to be performed in excess reactant regions, where the amount of reactant will not be able to limit the level of conversion. For isotope exchange experiments, the amount of reactant gas used is typically limited because of the cost. Our results suggest that a number of published isotope exchange results, including earlier work by our group,10 are also potentially mass transfer-limited, especially for those measured at high temperatures where surface exchange is so fast that the rate of transfer of oxygen gas to the surface is the rate-determining step. In this case, the apparent surface exchange coefficients represent limitations of the experimental setup and not the surface exchange of the catalyst. Therefore, it is paramount that experimental setup considerations be made. At low temperatures, the apparent activation energy is ∼42.3 kJ/mol, of which surface exchange is the main contribution. Literature values of kex from IEDP-SIMS at a PO2 of 0.025 atm15 show an activation energy of ∼180 kJ/mol. The chemical surface exchange coefficient (kchem) obtained from electrical conductivity relaxation (ECR) on the LSCF bulk (sample size on the order of millimeters) has an activation energy of ∼367 kJ/mol under a switch of PO2 from 0.01 to 0.02 atm and from 0.02 to 0.03 atm.14,15 Although the apparent activation energy from gas phase isotopic oxygen exchange is approximately one-quarter of the value from IEDP-SIMS, k values obtained from LSCF thin film (∼500 nm) measurements by ECR under a switch of PO2 from 0.01 to 1 atm,16 which is less likely to be limited by D, are much closer to our results with an activation energy of 34 kJ/mol. The possible reason is that the small size of powders used in gas phase isotopic oxygen exchange makes it a surface sensitive technique. Therefore, the apparent activation energy from IIE may be closer to the real value for the surface exchange process. A contour plot of the error mapping of the root-mean-square error (rmse) on the D−k plane is presented in Figure S4. From the error mapping in the k−D plane, we can see the fitting is insensitive to changes in D, suggesting that LSCF is limited by surface exchange. Benson et al.15 reported the activation

Figure 4. IIE of LSCF at different temperatures: gas phase oxygen fractions at (a) 500, (b) 600, (c) 700, and (d) 800 °C. Symbols are experimental data points, and lines are fitting results.

(500−800 °C), the surface exchange coefficient (kex) shows only a slight increase as the temperature increases with an apparent activation energy of ∼8 kJ/mol. The observed low activation energy at high temperatures demonstrates that the amount of oxygen flowing through the LSCF powder per unit time limits the number of reactions that can be detected. The low apparent activation energy of k at high temperatures is due 6028

DOI: 10.1021/acscatal.6b01462 ACS Catal. 2016, 6, 6025−6032

Research Article

ACS Catalysis energy of D is 89 kJ/mol using IEDP-SIMS. Bouwmeester et al.17 reported the activation energy for D of LSCF is ∼180 kJ/mol, which is on the same order of magnitude as k from IEDP-SIMS. Thus, the accuracy of measurements can vary depending on sample geometry and the boundary conditions of the characterization technique. Another possible reason for the variation in the apparent activation energies between different characterization techniques is the experimental temperature range. The surface exchange process needs to take into account both the contribution of surface oxygen vacancies and the ability of the material to dissociate oxygen, as shown in eq S21. At higher temperatures, the increase in the number of surface oxygen vacancies for LSCF may shift the apparent activation energy. A change in the activation energy for the surface exchange coefficient has been observed for other ionic conductors, such as Gadolinia-doped Ceria (GDC).18 Thus, gas phase isotopic oxygen exchange on GDC could provide a fundamental understanding of the change in mechanism at different temperatures. To further understand properties, such as active site to surface area ratio and oxygen partial pressure dependence, IIE of LSCF was conducted at different oxygen partial pressures (PO2 = 0.015 and PO2 = 0.02 atm) in the temperature range from 375 to 450 °C, as shown Figure 6. By changing the PO2 of

Figure 7. Log kex vs log PO2 for LSCF in the temperature range of 375−450 °C. The temperatures and linear fitting results in slopes are listed in the figure.

The PO2 dependence of kex at different temperatures is summarized in the double-log plot in Figure 7. The plot of log(PO2) versus log(kex) shows a linear relationship at all temperatures with slopes between 0.87 and 1. This linear relationship between PO2 and k could provide the information about the intermediate oxygen species that participate in the reaction. A slope between 0.87 and 1 indicates the main intermediate oxygen species that is participating in the surface exchange on LSCF from 375 to 450 °C is molecular oxygen, which is consistent with literature results.19 Surface exchange processes for LSM are different from those for LSCF because of its lack of oxygen vacancies. The super stoichiometry of LSM offers a limited number of oxygen vacancies even at high temperatures, resulting in a slow heteroexchange rate (gas−solid). Generally, homoexchange (gas−gas exchange) dominates on LSM, while heteroexchange occurs only at temperatures significantly above 700 °C.7f,11b,20 Gas phase isotopic oxygen exchange experiments on LSM were conducted from 750 to 850 °C. Figure 8 shows IIE of LSM at a PO2 of 0.025 atm at (a) 750, (b) 800, and (c) 850 °C. Experimental data are shown as points, while fitting, using a single-exchange mechanism, is shown as a dashed black line. Extracted kex values for LSM are 6.9 × 10−8, 6.0 × 10−8, and 4.7 × 10−8 cm/s at 750, 800, and 850 °C, respectively, resulting in a negative activation energy. The surface exchange process in LSM is co-limited by several factors and cannot be explained by a single-exchange model. One possible explanation is the presence of two parallel exchange mechanisms on the LSM surface. Ivanov et al.21 and De Souza et al.22 reported that the diffusion along grain boundaries may dominate the diffusion process for LSM in the intermediate temperature range. Therefore, it is reasonable to assume that oxygen atoms on the surface can be transported to the bulk by either of these exchange mechanisms. Here, we consider that LSM provides two parallel pathways for oxygen to be incorporated into the lattice. One pathway is relatively slow, and the diffusion process likely occurs via a vacancy exchange mechanism. The other pathway is fast and possibly occurs through grain boundaries. The boundary condition can be written as

Figure 6. IIE of LSCF at different PO2 values at intermediate temperatures: gas phase oxygen fractions at a PO2 of 0.015 atm at (a) 375, (b) 425, and (c) 450 °C and at a PO2 of 0.02 atm at (d) 375, (e) 425, and (f) 450 °C. Symbols are experimental data points, and lines are fitting results.

the gas flowing into the reactor, we observed that the exchange rate increases as the oxygen partial pressure increases. The number of available active sites for oxygen to adsorb and/or dissociate onto the powder surface is also directly related to the oxygen partial pressure through the Langmuir adsorption11b isotherm.

J18O = Sf Ak f (18ΘO − 18ΘO,lat) + SsAks(18ΘO − 18ΘO,lat) 6029

(8)

DOI: 10.1021/acscatal.6b01462 ACS Catal. 2016, 6, 6025−6032

Research Article

ACS Catalysis

Figure 9. IIE of LSM at different PO2 values at 800 °C: gas phase oxygen fractions at PO2 values of (a) 0.01 and (b) 0.05 atm. Symbols are experimental data points, and lines are fitting results with a singleexchange (black dashed lines) or a two-parallel exchange (green lines) model.

Table 1. Fitting Parameters for LSM with a Two-Parallel Exchange Mechanism at a PO2 of 0.025 atm at Different Temperatures T (°C)

SsA (m−2)

750 800 850

1.3 × 10 1.6 × 1018 2.1 × 1018 18

ks (cm/s) −8

2.9 × 10 2.8 × 10−8 2.5 × 10−8

SfA (m−2)

kf (cm/s)

2.2 × 1018 2.0 × 1018 1.9 × 1018

3.4 × 10−7 1.8 × 10−7 1.3 × 10−7

temperature dependence. It is possible that more accurate fits using the two-parallel exchange model occur solely because of the increased degrees of freedom. De Souza et al.23 reported D* and k* values of LSM at 800 °C from IEDP-SIMS of 5 × 10−15 cm2/s and 6 × 10−9 cm/s, respectively. Notice that IEDP-SIMS is a relatively D sensitive technique. From these values, the calculated characteristic length24 of LSM is ∼8 × 10−7 cm at 800 °C. In our experiments, the sample thickness is equal to the radius of the powder and is on the order of 10−5 cm. For LSM at 800 °C, the value of L (=a/Lc) is around 10−100, suggesting that the diffusion process limits the overall reaction. Even though IIE of LSM is an extremely surface sensitive technique with a small sample thickness, the low D value may still limit the exchange process. Because exchange on LSM is possibly diffusion-limited, eq S32 can be applied to estimate D at different temperatures, regardless of difficulties in accurately fitting k values or the potential mass transfer limitation at high temperatures. The temperature dependence of D for LSM is shown in Figure 10. Compared to D values for LSM from the literature,23 the D value at 750 °C extracted from IIE has a huge deviation. The apparent activation energy of D from isotope exchange is ∼100 kJ/mol, much smaller than the bulk value from the literature, ∼270 kJ/mol.

Figure 8. (a) IIE of LSM at a PO2 of 0.025 atm at (a) 750, (b) 800, and (c) 850 °C with single-exchange (black dashed lines) and a twoparallel exchange (green lines) model fits.

where kf and ks represent the fast and slow surface exchange rates on LSM, respectively, and SfA and SsA represent the available surface sites for fast and slow exchange, respectively. Curves generated from the two-parallel exchange model (green lines) more closely match the experimental data. Figure 9 shows IIE of LSM at different PO2 values at 800 °C. Again, the two-parallel exchange model (green lines) fits the experimental data better than the single-exchange model (black dashed lines). kf has a positive relationship with increases in PO2. In contrast, ks appears to be insensitive to changes in PO2. Because the oxygen stoichiometry of LSM is relatively insensitive to PO2, the slow exchange pathway can be attributed to bulk diffusion via vacancies and the fast exchange process related to grain boundary diffusion.22 Best-fit SfA, SsA, kf, and ks values are listed in Table 1. Both kf and ks show a negative temperature dependence. Though the two-parallel exchange model fits the experimental data of isotope exchange well, it fails to explain the inverse 6030

DOI: 10.1021/acscatal.6b01462 ACS Catal. 2016, 6, 6025−6032

Research Article

ACS Catalysis

vacancies that could contribute to the exchange process. On the other hand, the bulk of LSM still maintains a relatively low concentration of vacancies even at elevated temperatures, limiting diffusion to the near surface. The nonuniformity in vacancy concentration can cause the appearance of a nonconstant diffusion coefficient (D) as a function of depth and prevent a constant D model from being sufficient. To extract accurate kex and D values, the actual properties of the material, in this case LSM, need to be reflected in the boundary conditions. In summary, the oxygen exchange process for LSM appears to be more complicated than exchange for LSCF. This complexity may arise because of the oxygen stoichiometry of LSM at higher temperatures and the effect of oxygen vacancies on the various exchange and conduction mechanisms. Our isotope exchange results show that the high concentration of oxygen vacancies in a region near the LSM surface is a faster pathway for oxygen transport. This concept can extend beyond the surface of LSM particles to grain boundaries,26 which also have concentrations of vacancies higher than that of the bulk material.

Figure 10. Arrhenius plot of the diffusion coefficient (D) for LSM with a comparison to literature values.22 Red dots are experimental results from gas phase isotopic oxygen exchange, and black dots are D values for bulk LSM.

■

Although the lower activation energy of D may come from the kex−D mixed limited regime for the kinetic measurement, higher D values were clearly observed in the probing depth for gas phase isotopic oxygen exchange. To determine the probing depth, the penetration depths at different temperatures were calculated on the basis of diffusion lengths, l = (Dt)1/2, where D is the diffusion coefficient acquired from IIE and t is the total exchange time for each isotope exchange experiment. At 750 °C, the probing depth is ∼60 nm below the surface, and ∼110 nm at 850 °C, as shown in Figure 11. The difference between the acquired D values (average D values at the probing depth) from gas phase isotopic oxygen exchange and the bulk value may suggest that LSM has a high concentration of oxygen vacancies near the surface, a disproportionately large value compared to bulk vacancy concentrations. The high concentration of surface vacancies may account for the fast surface diffusion for LSM seen in the literature.25 Therefore, one possible reason for difficulty in fitting kex from IIE on LSM with eq S27 might be the distribution of oxygen vacancies in LSM. At lower temperatures, the majority of vacancies appear in a region near the surface, and the thickness of this region is small. As the temperature increases, the depth of this region increases, providing higher concentrations of

CONCLUSIONS

Isotope exchange is a powerful technique for probing the kinetics of heterogeneous catalysis. The overall ORR can be visualized and quantified by tracing the movement of labeled oxygen. We have demonstrated IIE of LSCF and LSM at different temperatures and oxygen partial pressures, allowing the kinetic rates to be obtained. Solid-state diffusion and surface heterogeneous catalysis have been linked by the heteroexchange rate coefficient (Rex) and the surface exchange coefficient (kex). The difference between Rex and kex is the contribution of surface vacancies. The isotope exchange profile of LSCF at intermediate and high temperatures can be explained by a simple first-order reaction with different rate-limiting mechanisms. At intermediate temperatures, surface exchange is the rate-limiting step. At high temperatures, because of the limitation of the isotope exchange experimental setup, there is a 100% conversion rate of the 18O flowing through the system. Therefore, this may limit the observation of actual kinetic rates. The experimental design of plug-flow isotope exchange experiments is very important because the observed results depend on reactor design. LSM is a pure electrical conductor below 750 °C with minimal surface vacancies. LSM shows a level of incorporation

Figure 11. Temperature dependence of the probing depth for gas phase isotopic oxygen exchange in LSM sphere powder at (a) 750, (b) 800, and (c) 850 °C. A nonuniform D value as a function of depth leads to the isotopic distribution of 16O (red) and 18O (yellow). 6031

DOI: 10.1021/acscatal.6b01462 ACS Catal. 2016, 6, 6025−6032

Research Article

ACS Catalysis

(8) (a) Kilner, J. A.; De Souza, R. A.; Fullarton, I. C. Solid State Ionics 1996, 86−88 (Part 2), 703−709. (b) Dhallu, M.; Kilner, J. A. J. Fuel Cell Sci. Technol. 2005, 2 (1), 29. (9) Mirodatos, C. Catal. Today 1991, 9 (1−2), 83−95. (10) (a) Kan, C. C.; Kan, H. H.; Van Assche, F. M.; Armstrong, E. N.; Wachsman, E. D. J. Electrochem. Soc. 2008, 155 (10), B985. (b) Kan, C. C.; Wachsman, E. D. Solid State Ionics 2010, 181 (5−7), 338−347. (c) Armstrong, E. N.; Duncan, K. L.; Oh, D. J.; Weaver, J. F.; Wachsman, E. D. J. Electrochem. Soc. 2011, 158 (5), B492. (11) (a) Christmann, K. Introduction to surface physical chemistry; Steinkopff Verlag and Springr-Verlag: Darmstadt, Germany, 1991. (b) Chorkendorff, I.; Niemantsverdriet, J. W. Concepts of modern catalysis and kinetics; Wiley-VCH: Weinheim, Germany, 2003. (12) Huang, Y. L.; Pellegrinelli, C.; Lee, K. T.; Perel, A.; Wachsman, E. D. J. Electrochem. Soc. 2015, 162 (9), F965−F970. (13) Fogler, H. S. Elements of chemical reaction engineering, 3rd ed.; Prentice Hall PTR: Upper Saddle River, NJ, 1999. (14) Wang, S.; van der Heide, P. A. W.; Chavez, C.; Jacobson, A. J.; Adler, S. B. Solid State Ionics 2003, 156, 201−208. (15) Benson, S. J.; Waller, D.; Kilner, J. A. J. Electrochem. Soc. 1999, 146 (4), 1305−1309. (16) Zomorrodian, A.; Salamati, H.; Lu, Z.; Chen, X.; Wu, N.; Ignatiev, A. Int. J. Hydrogen Energy 2010, 35 (22), 12443−12448. (17) Bouwmeester, H. J. M.; Den Otter, M. W.; Boukamp, B. A. J. Solid State Electrochem. 2004, 8 (9), 599−605. (18) Ruiz-Trejo, E.; Sirman, J. D.; Baikov, Y. M.; Kilner, J. A. Solid State Ionics 1998, 113−115 (1−2), 565−569. (19) (a) Adler, S. B.; Chen, X. Y.; Wilson, J. R. J. Catal. 2007, 245 (1), 91−109. (b) Fleig, J.; Merkle, R.; Maier, J. Phys. Chem. Chem. Phys. 2007, 9 (21), 2713−2723. (c) Chueh, W. C.; Haile, S. M. Annu. Rev. Chem. Biomol. Eng. 2012, 3, 313−341. (20) (a) Klier, K.; Novakova, J.; Jiru, P. J. Catal. 1963, 2 (6), 479− 484. (b) Boreskov, G. K. Kinet. Catal. 1977, 18 (5), 910−918. (c) Boreskov, G. K.; Eley, D. D.; Pines, H.; Weisz, P. B. Advances in Catalysis; Academic Press: New York, 1964; Vol. 15. (21) Ivanov, D. V.; Sadovskaya, E. M.; Pinaeva, L. G.; Isupova, L. A. J. Catal. 2009, 267 (1), 5−13. (22) De Souza, R. A.; Kilner, J. A.; Walker, J. F. Mater. Lett. 2000, 43 (1−2), 43−52. (23) De Souza, R. A.; Kilner, J. A. Solid State Ionics 1998, 106 (3−4), 175−187. (24) Carter, S.; Selcuk, A.; Chater, R. J.; Kajda, J.; Kilner, J. A.; Steele, B. C. H. Solid State Ionics 1992, 53−56, 597−605. (25) (a) Harrison, W. A. Phys. Rev. B: Condens. Matter Mater. Phys. 2010, 81 (4), 045433. (b) Choi, Y.; Lin, M. C.; Liu, M. Angew. Chem. 2007, 119 (38), 7352−7357. (c) Backhaus-Ricoult, M.; Adib, K.; St. Clair, T.; Luerssen, B.; Gregoratti, L.; Barinov, A. Solid State Ionics 2008, 179 (21−26), 891−895. (26) (a) Saranya, A. M.; Pla, D.; Morata, A.; Cavallaro, A.; CanalesVázquez, J.; Kilner, J. A.; Burriel, M.; Tarancón, A. Adv. Energy Mater. 2015, 5 (11), 1500377−1500382. (b) Navickas, E.; Huber, T. M.; Chen, Y.; Hetaba, W.; Holzlechner, G.; Rupp, G.; Stoger-Pollach, M.; Friedbacher, G.; Hutter, H.; Yildiz, B.; Fleig, J. Phys. Chem. Chem. Phys. 2015, 17 (12), 7659−7669. (c) Huber, T. M.; Navickas, E.; Friedbacher, G.; Hutter, H.; Fleig, J. ChemElectroChem 2015, 2 (10), 1487−1494. (d) Farlenkov, A. S.; Ananyev, M. V.; Eremin, V. A.; Porotnikova, N. M.; Kurumchin, E. K. Fuel Cells 2015, 15 (1), 131− 139.

and a dissociation rate relatively lower than those of LSCF. LSM has an exchange flux with a possible two-parallel exchange pathway or a different vacancy concentration in the near surface region than in the bulk. Our isotope exchange results suggest the second one is more likely. Studies of isotope exchange can greatly contribute to the elucidation of the interactions between molecular oxygen and the solid surface. It can help characterize the dissociation of oxygen molecules at the oxide surface as well as the conduction of oxygen ions within the oxide. The kinetic data based on IIE results can provide insight about what is actually happening during the ORR. The results allow us to further investigate oxygen exchange mechanisms that occur in the ORR as well as to design and engineer new materials and/or structures that can enhance these surface reactions.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acscatal.6b01462. Isotope exchange theory, validation of the fitting, and isothermal isotope exchange of LSM (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS The authors acknowledge the support of the U.S. Department of Energy, NETL, via Contract DEFE0009084. REFERENCES

(1) (a) Wachsman, E. D.; Lee, K. T. Science 2011, 334 (6058), 935− 939. (b) Wachsman, E. D.; Marlowe, C. A.; Lee, K. T. Energy Environ. Sci. 2012, 5 (2), 5498. (2) (a) Zhang, J.; Sasaki, K.; Sutter, E.; Adzic, R. R. Science 2007, 315 (5809), 220−222. (b) Cheng, F.; Shen, J.; Peng, B.; Pan, Y.; Tao, Z.; Chen, J. Nat. Chem. 2011, 3 (1), 79−84. (c) Gong, K.; Du, F.; Xia, Z.; Durstock, M.; Dai, L. Science 2009, 323 (5915), 760−764. (3) (a) Olivier, L.; Haag, S.; Mirodatos, C.; van Veen, A. C. Catal. Today 2009, 142 (1), 34−41. (b) Liang, F.; Partovi, K.; Jiang, H.; Luo, H.; Caro, J. J. Mater. Chem. A 2013, 1 (3), 746−751. (c) Cao, Z.; Jiang, H.; Luo, H.; Baumann, S.; Meulenberg, W. A.; Assmann, J.; Mleczko, L.; Liu, Y.; Caro, J. Angew. Chem., Int. Ed. 2013, 52 (51), 13794− 13797. (d) Kniep, J.; Lin, Y. S. Ind. Eng. Chem. Res. 2011, 50 (13), 7941−7948. (4) Girishkumar, G.; McCloskey, B.; Luntz, A. C.; Swanson, S.; Wilcke, W. J. Phys. Chem. Lett. 2010, 1 (14), 2193−2203. (5) (a) Scheffe, J. R.; Weibel, D.; Steinfeld, A. Energy Fuels 2013, 27 (8), 4250−4257. (b) Yang, C.-K.; Yamazaki, Y.; Aydin, A.; Haile, S. M. J. Mater. Chem. A 2014, 2 (33), 13612. (6) Maier, J. Solid State Ionics 1998, 112, 197−228. (7) (a) Bouwmeester, H. J.; Song, C.; Zhu, J.; Yi, J.; van Sint Annaland, M.; Boukamp, B. A. Phys. Chem. Chem. Phys. 2009, 11 (42), 9640−9643. (b) den Otter, M. W.; Boukamp, B. A.; Bouwmeester, H. J. M. Solid State Ionics 2001, 139, 89−94. (c) Lacombe, S.; Z, H.; Mirodatos, C. J. Catal. 1995, 11, 9640−9643. (d) Shannon, S. L.; Goodwin, J. G. Chem. Rev. 1995, 95 (3), 677−695. (e) Boreskov, G. K. Heterogeneous catalysis; Nova Science Publishers: Hauppauge, NY, 2003. (f) Hargreaves, J. S. J.; Jackson, S. D.; Webb, G. Isotopes in heterogeneous catalysis; Imperial College Press: London, 2006. (g) Huang, Y. L.; Pellegrinelli, C.; Wachsman, E. D. J. Electrochem. Soc. 2016, 163 (3), F171−F182. 6032

DOI: 10.1021/acscatal.6b01462 ACS Catal. 2016, 6, 6025−6032