Beneficial Lattice Strain in Heterogeneously Doped Ceria - The

Article pubs.acs.org/JPCC

Beneficial Lattice Strain in Heterogeneously Doped Ceria Weida Shen,† Jun Jiang,‡ and Joshua L. Hertz*,†,‡ †

Department of Mechanical Engineering, University of Delaware, 126 Spencer Laboratory, Newark, Delaware 19716, United States Department of Materials Science and Engineering, University of Delaware, 201 DuPont Hall, Newark, Delaware 19716, United States

‡

ABSTRACT: Oxygen ion conduction in heterogeneously doped films composed of alternating layers of pure Y2O3 and pure CeO2 was reported recently, with conduction coming predominantly from vacancies trapped in interfacial space charge regions in CeO2. Here, we expand this concept to films composed of CeO2 heterogeneously doped with Y2O3, Gd2O3, or La2O3 in order to study the effects of heterodopant identity on the oxygen ion conductivity. For all samples, the thickness of the entire structure and that of the individual dopant layers were kept constant. The dopant oxides used in this work adopt the cubic bixbyite crystal structure with pseudo-fluorite lattice parameters that are, relative to CeO2, smaller (Y2O3), larger (La2O3), and nearly equal (Gd2O3). The total conductivity of the films increased with increasing lattice parameter of the dopant oxides. An electrostatic Gouy−Chapman model is not sufficient to explain this behavior because all of the dopants theoretically contribute identical interfacial oxygen vacancy concentrations. Therefore, extensions beyond a Gouy−Chapman model are suggested, with consideration for the effects of interfacial strain on the concentration and mobility of vacancies from dopant oxides in the space charge region.

■

INTRODUCTION

Recent work by the authors has investigated heterogeneously doped CeO2 films consisting of alternating layers of 1 nm thick pure Y2O3 and 19 nm thick pure CeO2.18 The conductivities of those films remained essentially constant across a range of oxygen partial pressures, indicating a dominant ionic conduction. Pure Y2O3 exhibits the cubic bixbyite crystal structure, which can be described as a “pseudo-fluorite” structure with an ordered array of one of every four oxygen sites unfilled.19 Figure 1 illustrates a schematic anion sublattice near the Y2O3/CeO2 interface. Strictly speaking, the unfilled sites are interstitial locations, and the migration of an oxygen ion from the CeO2 into one of these sites in the Y2O3 lattice is described as

CeO2-based materials are archetypal solid oxide electrolyte materials for use in fuel cells and other devices.1−4 With the addition of certain aliovalent oxides, such as Y2O3, oxygen vacancies are created to compensate the charge imbalance when Ce4+ host cations are substituted by the lower-valence solute Y3+. The conduction of oxygen ions in Y-doped CeO2 and, in most cases, other solid oxide-ion electrolytes occurs via a highly thermally activated vacancy-hopping mechanism.5 Pure, nominally undoped CeO2 cannot be used as the electrolyte material because of the relatively low concentration of intrinsic vacancies. When the dopant concentration is gradually increased, the oxygen ion conductivity of the doped CeO2 increases initially, but reaches a maximum value at a dopant concentration of about 15−20%.6,7 The mechanism behind this maximum conductivity at a relatively low dopant concentration is complex. 8−12 Dopant−vacancy association has been suggested as the main reason for reduced conductivity at high dopant concentration,8,13−15 though vacancy−vacancy interactions have also been proposed.16 At low temperatures, positively charged vacancies become trapped near the negatively charged dopants because of both Coulombic attraction and local lattice strain fields caused by the different ionic radii of the host and dopant atoms.8,13−15 Gd is one of the most widely used dopants in CeO2 because of the similarity in ionic radius between Ce and Gd compared to that of, for example, Y or La.8,17 © 2014 American Chemical Society

CeO2

OO +

Y2O3

Vi → CeO2 V •• O +

Y2O3

O″i

(1)

where the text in superscript preceding each defect indicates the lattice in which the defect formally resides. This simple picture, though, is complicated by the fact that the interface has some chemical resemblance to a 50% doped ceria. Thus, this migration event may proceed to a greater extent than bulk measurements of defect formation energies in Y2O3 would suggest. The space charge regions created by the spatially dependent defect formation provide a counteracting Coulombic energy to limit the extent of the above reaction. Similar Received: July 1, 2014 Revised: September 8, 2014 Published: September 10, 2014 22904

dx.doi.org/10.1021/jp506554z | J. Phys. Chem. C 2014, 118, 22904−22912

The Journal of Physical Chemistry C

Article

of La2O3 was prepared by pressing La2O3 powders (99.999%, Inframat Advanced Materials) to a pressure of 20 MPa at room temperature. The pressed powder target was then sintered for 12 h at 1450 °C in air. All targets were reactively sputtered in an oxygen-rich environment to create oxidized films. In the operation of a typical film deposition, the background pressure of the sputtering chamber was roughly 1.33 × 10−5 Pa (10−7 Torr). Working gases of Ar and O2 were introduced into the sputtering chamber with a total gas flow rate of 20 sccm and an Ar-to-O2 flow rate ratio of 9:1. The deposition pressure was 1.33 Pa (10 mTorr), maintained using downstream pressure control. Single-crystal Al2O3 with (0001) orientation (MTI Corporation, Richmond, CA) was used as substrates (10 × 10 × 0.5 mm3). During the deposition process, the substrates were heated to 650 °C. The deposition rate from each target was adjusted using the applied RF sputtering power. A more detailed description of this fabrication process was reported previously in ref 32 with direct imaging of 1 nm compositional modulations given in ref 18. Films with modulated composition were sputtered sequentially using a computer to control the power applied to each target and the toggling of target shutters. The film thickness was measured by an optical interferometer (Veeco Wyko NT9100, Plainview, NY). X-ray diffraction (XRD) measurements were performed on a Philips X’Pert diffractometer with Cu Kα radiation using typical θ-2θ geometry to determine the orientations of the films and Al2O3 substrates. Platinum interdigitated electrodes for conductivity measurements were fabricated on the top surface of the films by lift-off photolithography. The width of the electrodes and the spacing between the electrodes were both 100 μm. The total perimeter length of the interdigitated electrodes was ≈115 mm. Electrochemical impedance spectroscopy (EIS) was measured in air with a Novocontrol AlphaA analyzer (Novocontrol Technologies, Montabaur, Germany) at temperatures between 300 and 650 °C. The frequency range was between 1 Hz and 3 MHz with a signal voltage of 10 mV. Equivalent circuit analysis was carried out using Zview software (Version 3.3, Scribner Associates, Southern Pines, NC).

Figure 1. Schematic anion superlattice structure at a Y2O3/CeO2 interface. Cations are not pictured. Gray circles represent oxygen in the CeO2 lattice, while black circles represent oxygen in the Y2O3 lattice. The open squares represent unfilled sites in Y2O3. The arrow indicates the movement of an oxygen ion from CeO2 into Y2O3, creating an oxygen ion interstitial at the previously unfilled site in Y2O3 and leaving behind an oxygen vacancy in CeO2.

reactions across La2O3/CeO2 interfaces have been examined as a means of reducing oxygen vacancies in La2O3 gate dielectrics.20−23 Oxygen vacancies in the space charge regions in undoped CeO2 layers transport with relatively high mobility. This design is somewhat analogous to high electron mobility transistors (HEMTs) in which electrons produced in a doped AlGaAs layer possess higher mobility when trapped in neighboring regions of undoped GaAs.24 A Gouy−Chapman model quantified the conductance induced in the CeO2 space charge regions,25,26 which was similar to the experimentally measured total conductance of the heterogeneously doped films. This conductance was much larger than that of the pure Y2O3 and pure CeO2 layers considered independently. In this work, the theory of heterogeneously doped electrolyte films is further developed by adjusting the net dopant content and the dopant identity. Increased dopant content is accomplished by decreasing the thickness of the pure CeO2 layers from ≈19 nm to ≈4 nm. Because the total film thickness stayed ≈200 nm, the decreased CeO2 layer thickness yields an increased number of space charge regions and amount of dopant, from 5% to 20%. Dopant oxides examined here included Y2O3, Gd2O3, and La2O3. These three rare earth metal oxides (RE2O3) adopt the same bixbyite crystal structure but with lattice parameters of 10.604, 10.813, and 11.327 Å, respectively.19,27 The lattice parameter of the bixbyite structure is roughly twice that of fluorite because of the reduced symmetry, but the actual bond lengths and geometry are highly similar. The fluorite-equivalent lattice parameters for Y2O3, Gd2O3, and La2O3 can be estimated as 5.30, 5.41, and 5.66 Å, respectively. Compared with the lattice parameter of bulk CeO2 (a ≈ 5.41 Å28), these dopant oxides provide different lattice mismatch at the interfaces. The lattice mismatch between Gd2O3 and CeO2 can likely be neglected, but for films composed of CeO2 with Y2O3 or La2O3, a biaxial compressive or tensile strain, respectively, may be created in the CeO2 layers near the interfaces. The effects of strain on ion mobility in heterostructures have been of intense interest lately.29−31

■

RESULTS AND DISCUSSION Three different sets of samples composed of alternating layers of CeO2 with Y2O3, Gd2O3, or La2O3 were prepared, as summarized in Table 1. The total thickness of these samples Table 1. Configuration of Three Different Sets of Heterogeneously Doped CeO2 Films Composed of Alternating Layers of RE2O3 and CeO2a samples

RE2O3 layer thickness (nm)

CeO2 layer thickness (nm)

number of bilayers

A B C

1 1 1

19 9 4

10 20 40

a

Total film thickness was maintained at ≈200 nm.

was kept constant at ≈200 nm. Within each sample set, the individual layer thickness of the rare earth metal oxide (RE2O3) was maintained at ≈1 nm, while the layer thickness of CeO2 decreased from ≈19 nm to ≈4 nm to increase the number of heterogeneous interfaces. Compositional modulation through the thickness at singlenanometer resolution is a challenge to this work. A serious concern could be the cation diffusion, with the dopants mixing

■

EXPERIMENTAL SECTION Thin films were deposited in a multitarget, custom-built magnetron sputtering machine (PVD Products, Wilmington, MA). Single-element, metallic targets of Ce, Y, and Gd were purchased from ACI Alloys (San Jose, CA). A sputtering target 22905



Article

with the neighboring undoped CeO2 layers during film deposition and subsequent conductivity measurements. This concern, however, would be highly unlikely to become an issue in this work because of the extremely low diffusion coefficients of cations in ceria at temperatures of 650 °C and below. Rochenhäuser et al. investigated the cation interdiffusion between thin Gd2O3 films on CeO2 substrates.33 On the basis of that work, the Gd cation diffusion coefficient in CeO2 can be estimated at about 2.8 × 10−27 m2/s at 650 °C. A similar value between 10−26 m2/s and 10−27 m2/s for the cation diffusion coefficient for Y or La in CeO2 at 650 °C could also be estimated from refs 34 and 35. The total time for sample preparation and conductivity measurement was less than 35 h; for only a small portion of the time were samples heated to the maximum of 650 °C. Ignoring that the diffusivity is greatly (exponentially) reduced at lower temperatures, the diffusion length for the dopants at 650 °C is calculated to be less than 0.7 Å even after 35 h. Previous work using cross-sectional transmission electron microscopy (TEM) confirmed the desired compositional heterogeneity in a sample with bilayers of 1 nm thick Y2O3 and 19 nm thick CeO2.18 Still, diffusion of dopants at the grain boundaries may have occurred becuase the films adopted a columnar microstructure. The presence of grain boundaries did not visibly interrupt the compositionally layered structure. The one-dimensional model of vacancy concentration distribution at the interfaces is believed to remain valid. Any segregation of dopants to the grain boundaries in CeO2 that occurred would locally decrease the mobility of oxygen vacancies.36 In our previous analysis, the vacancy mobility in the CeO2 space charge regions was assumed to be the same as that in the CeO2 bulk regions. Quantitative adjustment to the vacancy mobility parameter within the model could be made if greater precision is desired. Figure 2a shows the θ-2θ XRD patterns of the heterogeneously doped CeO2 films with Y2O3. A single-layer, 105 nm thick CeO2 film is also included for comparison. From the figure, we can see that the samples with 19 and 9 nm thick CeO2 layers exhibit orientations along both (111) and (100) directions. Although there is a difference in lattice parameters between CeO2 and Y2O3, the CeO2(111) and (100) peaks are dominant in these two samples without any visible peaks corresponding to the Y2O3. This finding may be attributed to the small amount of Y2O3 within the samples. For the sample with CeO2 layer thickness of 4 nm, the CeO2(111) and (100) peaks are still dominant, but a small additional peak at 2θ ≈ 26.73° appears. This may be a satellite peak for the asymmetric superlattice formed in this film. Figure 2b shows the XRD patterns of the films composed of alternating layers of Gd2O3 and CeO2. As with the Y2O3 samples, the films are all predominantly oriented along the CeO2(111) and (100) directions. As the layer thickness of CeO2 decreases, an additional small peak begins to emerge on the right shoulder of the CeO2(111) peak. This small peak may also be a satellite superlattice peak, but becuase the fluorite-equivalent lattice parameter of Gd2O3 is very close to that of CeO2, it is very difficult to distinguish. The XRD patterns of the films composed of alternating layers of La2O3 and CeO2 are shown in Figure 2c. Similar to the previous samples, no separate peaks corresponding to the diffraction planes of La2O3 are identified in the films with 19 and 9 nm thick CeO2. As the layer thickness of CeO2 decreased to 4 nm, a superlattice XRD pattern with an average structure peak SL0 surrounded by two very small first-order satellite peaks, SL+1 and SL−1, is observed

Figure 2. θ-2θ XRD patterns of heterogeneously doped CeO2 films composed of alternating layers of CeO2 with (a) Y2O3, (b) Gd2O3, or (c) La2O3 on Al2O3 (0001) substrates. A single-layer CeO2 film with thickness of ≈105 nm is also included for comparison.

for the (111) and (222) diffraction peaks. The appearance of satellite peaks indicates the formation of high-quality interfaces between the adjacent layers.37 Two possible issues can complicate the ability to quantify the conductivity of a thin film using electrodes placed on the top surface. The first regards the relative resistance of the current path through the film relative to the substrate, which has much lower conductivity but much greater thickness. To ensure that the film and not the substrate properties were being measured, preliminary work included measurement of the impedance of nominally identical electrodes placed directly on bare Al2O3 substrates. The resistances measured were larger than those of the samples reported here by more than 1 order of magnitude. The second possible issue regards whether the current path proceeded through the entire thickness of the film or remained confined to the top surface layer. Schweiger et al. used two different electrode patterns to measure the impedance of Ce0.9Gd0.1O2−δ/Er2O3 multilayers.38 When the electrodes were deposited on the top surface of multilayers, the electrical current was found to not penetrate into the whole multilayer. 22906



Article

Figure 3. (a) Impedance spectra of a heterogeneously doped film composed of bilayers of 1 nm thick Gd2O3 and 4 nm thick CeO2 at the temperatures indicated. The equivalent circuit used to fit the data is inset. Arrhenius plots of the electrical conductivity of heterogeneously doped films composed of CeO2 with (b) Y2O3, (c) Gd2O3, or (d) La2O3. The symbols indicate whether the films consist of (■) 10 bilayers of 1 nm RE2O3/ 19 nm CeO2, (●) 20 bilayers of 1 nm RE2O3/9 nm CeO2, or (▲) 40 bilayers of 1 nm RE2O3/4 nm CeO2. The conductivities of single-layer Ce0.90Y0.10O2−δ (YDC), Ce0.80Gd0.20O2−δ (GDC), and Ce0.85La0.15O2−δ (LDC) thin films with thicknesses of 192, 173, and 256 nm, respectively, are also included for comparison.

Figure 3a shows typical impedance spectra from a film composed of bilayers of 1 nm thick Gd2O3 and 4 nm thick CeO2. The spectra exhibit three main features with a semicircle located at high frequency, a highly depressed semicircular arc at intermediate frequency, and the first portion of a semicircle at low frequency. Three R-CPE elements connected in series formed the equivalent circuit to fit the corresponding impedance spectrum. The capacitance corresponding to lateral conduction through a thin film on an insulating substrate is immeasurably low and is typically dwarfed by stray capacitances. Thus, the measured capacitance from this kind of experiment rarely corresponds to the geometry of the sample. On the other hand, the electrode polarization terms typically have capacitances that are larger by orders of magnitude. The equivalent capacitance values of CPE1 were extremely low, of order 10−11 F, which represents the lowest values we have measured with our experimental setup. For the CPE2 and CPE3 at low frequency, the capacitances were larger than 10−7 F. Such large capacitance values cannot be explained by the geometry of conduction through the sample and gave strong evidence that resistance R1 was due to the conduction through the film, while R2 and R3 were due to electrode effects. The electrical conductivity of each film, σ, was calculated according to w σ= R1(td) (2)

Rather, the current was confined to the topmost Ce0.9Gd0.1O2−δ layer because of the large resistance of the Er2O3 layer directly beneath. In the work reported here, the thickness of each RE2O3 layer was always ≈1 nm. The number of bilayers varied from 10 to 40; therefore, the total thickness of RE2O3 in a multilayer was at most ≈40 nm. On the other hand, the lateral conduction path length between the two electrode fingers was 100 μm. Oxygen ion conductivity in the RE2O3 layers is expected to be very small. Ando et al. measured the oxygen selfdiffusion coefficient in single-crystal Y2O3 and found that the value was at least 5 orders of magnitude less than that of Zr0.85Ca0.15O2−δ at 1100 °C.19 Wirkus et al. observed that the electrical conductivity of Gd2O3 was on the order of 10−6 S/cm at 727 °C.39 On the other hand, the electrical conductivity of pure CeO2 at 650 °C is normally of order 10−3 S/cm.40 The electrical conductivity of the RE2O3 layers was, therefore, about 3 orders of magnitude less than that of pure CeO2. The electrical resistance through the thickness of the 40 nm thick RE2O3 is calculated to be about 1 order of magnitude less than that of the 100 μm lateral conduction through the CeO2. Thus, the current is not likely to be confined to the top CeO2 layer, yet the contribution from the parallel RE2O3 layers to the total electrical conductivity measured laterally across the multilayer (parallel to the interfaces) can nevertheless be neglected. Moreover, as will be shown below, the measured impedance of the multilayers generally scaled in linear proportion to the number of RE2O3/CeO2 interfaces. A linear increase in the conductivity as a function of the number of the interfaces provides very strong evidence that the measured electrical current penetrates through the entire thickness of the multilayer.

in which R1 is the resistance determined from the equivalent circuit fit to the data, w (= 100 μm) the distance between each pair of interdigitated electrode fingers, t (≈ 115 mm) the total perimeter length of the interdigitated electrode, and d the 22907



Article

thickness of the conducting film. The contribution of RE2O3 to the overall oxygen ion conduction likely can be neglected. We thus set d to be the total thickness of CeO2 in each film. In a multilayer composed of 10 bilayers of 1 nm thick RE2O3 and 19 nm thick CeO2, d = 190 nm; in a multilayer composed of 20 bilayers of 1 nm thick RE2O3 and 9 nm thick CeO2, d = 180 nm; and in a multilayer composed of 40 bilayers of 1 nm thick RE2O3 and 4 nm thick CeO2, d = 160 nm. The actual d values used to calculate conductivity were adjusted slightly based on the measured thickness of each sample. The total range of d values across all samples is within ±15%, which is not sufficient variability to strongly affect any conclusions. Figure 3b shows the Arrhenius plots of the electrical conductivity of the Y2O3/CeO2 films in air. The conductivity of the sample with 9 nm thick CeO2 (20 bilayers) is somewhat increased compared to the sample with 19 nm thick CeO2 (10 bilayers). Further decreasing the layer thickness of CeO2 to 4 nm (40 bilayers), however, yields conductivity similar to that of the 20 bilayer sample. The homogeneously doped Ce0.90Y0.10O2−δ (YDC) single-layer film, however, exhibits larger conductivity by almost 1 order of magnitude compared to that of the heterodoped Y2O3/CeO2 samples. Panels c and d of Figure 3 show the conductivities of the Gd2O3/CeO2 and La2O3/CeO2 films, respectively. The conductivities of the single-layer homogeneously doped Ce0.80Gd0.20O2−δ (GDC) and Ce0.85La0.15O2−δ (LDC) films are also included for comparison. At all temperatures, the conductivities of the Gd2O3/CeO2 and La2O3/CeO2 films monotonically increased as the layer thickness of CeO2 decreased, and the number of interfaces between RE2O3 and CeO2 correspondingly increased. At the same time, the activation energy remains nearly constant. The conductivity of the heterodoped sample composed of 1 nm thick La2O3 and 4 nm thick CeO2 approaches that of the homogeneously doped LDC film with optimal 15% dopant concentration. The conductivities of the heterodoped films increased in the order La2O3 > Gd2O3 > Y2O3, which is notably not the behavior for these as homogeneously distributed dopants in ceria.17,41,42 For the Gd2O3/CeO2 and La2O3/CeO2 samples, the conductivity scales roughly linearly with the number of heterogeneous interfaces. Figure 4 shows the isothermal electrical conductivity of Y2O3/CeO2, Gd2O3/CeO2, and La2O3/CeO2 films as a function of the number of interfaces. For the films composed of Y2O3 and CeO2, the conductivity has low dependence on the number of interfaces, but for the Gd2O3/CeO2 and La2O3/CeO2 samples, conductivity is roughly linearly dependent on the number of interfaces between RE2O3 and CeO2. In our previous work, the conduction behavior of heterogeneously doped Y2O3/CeO2 was understood by deconvoluting the conduction mechanism into three independent, parallel pathways.18 The first two paths were conduction through the CeO2 and Y2O3 bulk regions. The third path was conduction through the space charge regions in CeO2 near the interfaces. Because of the extremely small conductivity of the Y2O3 layers, the contribution from Y2O3 bulk regions to the overall oxygen ion conduction was neglected. The total conductivity of the heterogeneously doped film was thus determined by the conduction through the bulk and space charge regions of CeO2 layers. The oxygen vacancy concentration in the CeO2 space charge regions was modeled with a Gouy−Chapman model in which electrostatic effects balance diffusive transfer of vacancies away

Figure 4. Isothermal electrical conductivity of the heterogeneously doped films composed of CeO2 with (a) Y2O3, (b) Gd2O3, and (c) La2O3 as a function of the number of interfaces at three different temperatures. Note that average measurement temperatures for the three samples are given.

from the interface. The total conductance of the space charge regions in excess to that of bulk CeO2 was derived according to refs 25 and 26 as t GTotal,SC = (2n − 1)μv 2ε0εrkBTc v(0) (3) w in which GTotal,SC represents the electrical conductance in space charge regions; n is the number of CeO2 layers in the heterogeneously doped film; μv denotes the mobility of the oxygen vacancies in the space charge regions; εo is the permittivity of free space, εr the relative dielectric constant, kB Boltzmann’s constant, and T the absolute temperature; cv(0) represents the concentration of oxygen vacancies at the interface; and t and w denote the total perimeter length and separation distance, respectively, of the electrodes. Interestingly, the conductance of the space charge regions does not depend on the bulk vacancy concentration in CeO2, but only on the vacancy concentration in CeO2 at the RE2O3 interface. As mentioned previously, RE2O3 adopts a “pseudo-fluorite” 22908



Article

bixbyite structure with one of every four oxygen sites unfilled. In order to create a quantifiable model, here the unfilled sites in RE2O3 at the interface are treated as vacancies. The value of cv(0) was set equal to the concentration of unfilled sites in RE2O3. The oxygen vacancy mobility in the space charge regions was assumed to be the same as that in the bulk regions of CeO2. When the parallel conductance from the bulk regions of the CeO2 layers and the space charge regions in the CeO2 layers were compared, oxygen ion conduction in the whole film was found to be dominated by vacancies trapped in CeO2 space charge regions. On the basis of eq 3, the conductivity of heterogeneously doped films should be linearly dependent on the number of interfaces and the square root of interfacial vacancy concentration. The three different RE2O3 used in this work adopt the same bixbyite crystal structure and therefore theoretically possess the same concentration of vacant oxygen sites. Thus, a purely electrostatic model such as Gouy− Chapman cannot account for the finding here that heterogeneously doped films composed of CeO2 with Y2O3, Gd2O3, or La2O3 exhibit different conductivity values when they have the same number of interfaces. Moreover, the Gouy−Chapman model cannot explain why the films heterodoped with Y2O3 exhibited conductivity values that did not linearly scale as the number of interfaces increased. The most obvious difference between the three dopant oxides examined here is their lattice parameter. The significant change in conductivity behavior across these three dopants suggests that lattice mismatch strain may be influencing the oxygen ion conductivity in the space charge zones. The effect of lattice strain on oxygen ion conduction behavior has been extensively studied in nanometric multilayered systems using doped conductors.37,43−49 A number of studies37,43,45−48with notable exceptions44,49 have found that biaxial tensile strain induced in zirconia- or ceria-based solid electrolyte films increased the total oxygen ion conductivity, whereas compressive strain had the opposite effect. For systems involving highly doped conductors, the space charge regions are inconsequentially small;18 thus, the concentration of oxygen vacancies is normally fixed by the dopant concentration. The increased conductivity was, therefore, attributed to increased vacancy mobility in tensile-strained regions due to reduced migration energy. These assumptions were later verified by the numerical work carried out by Kushima et al.50 in YSZ and De Souza et al.51 in CeO2. The conductivity of ionic conductors normally can be expressed as

Figure 5. Activation energy of conduction of the heterogeneously doped films composed of CeO2 with Y2O3, Gd2O3, or La2O3 as a function of the number of interfaces.

of a 26% doped sample is 1.1 eV.8 The constancy of the activation energy over the large range of dopant content for the heterogeneously doped samples reported here indicates that increased dopant content does not lead to more deeply trapped vacancies. This result is as expected from the space charge model, in which the mobile defects are in the pure ceria layers. Any effects of dopant−vacancy association are thus reduced for these defects. A practical benefit of this effect may be that large total mobile defect concentration may be possible in solid electrolytes without coincident high activation energy using heterogeneous doping. Figure 6 shows the activation energy of the three different types of heterogeneously doped CeO2 films as a function of the

(4)

Figure 6. Activation energy of conduction of the heterogeneously doped CeO2 films as a function of the lattice parameter mismatch between the dopant and CeO2. Note that the lattice mismatch is calculated based on the formula [(aRE2O3 − aCeO2)/aCeO2] in which a represents the lattice parameter. The dotted line represents the activation energy of very lightly doped CeO2 with 0.05 mol % Y2O3.8

in which σ0 is a pre-exponential factor and Ea represents the activation energy for ionic conduction. A linear fit of ln(σT) as a function of 1/T allows the calculation of Ea.14 Figure 5 plots the activation energy of each of the three different types of heterogeneously doped CeO2 films. For each of the dopants, the activation energy is weakly and nonmonotonically dependent on the concentration of dopant in the film (i.e., the number of heterodopant/CeO2 interfaces). Specifically, the activation energy does not significantly increase as the net dopant concentration increases from 5% to 25%. This behavior is different from that of homogeneously doped ceria.1 For example, in homogeneously distributed yttria-doped ceria, the activation energy of a 5% doped sample is 0.77 eV, whereas that

lattice mismatch between RE2O3 and CeO2. The activation energy of the Gd2O3/CeO2 samples was about 0.85 eV, and the activation energy of the La2O3/CeO2 samples was about 0.87 eV. De Souza et al. numerically found that the vacancy migration energy would decrease from ≈0.59 eV to ≈0.2 eV when the lattice mismatch increased from near-zero to +4.6%.51 The similarity in these activation energies with each other and with that of very lightly doped ceria (≈0.90 eV)8 in this work gives strong indication that either the lattice mismatch between La2O3 and CeO2 is not creating a significant amount of tensile strain in the CeO2, or alternatively, that the migration energy of vacancies in the space charge zones is not strongly affected by

σ=

⎛ E ⎞ σ0 exp⎜ − a ⎟ T ⎝ kBT ⎠

22909



Article

increased it. A similar change in vacancy formation energy here would provide a driving force for the transfer of oxygen vacancies from the compressive-strained La2O3 into the tensilestrained CeO2.

tensile strain. On the other hand, the activation energy of the Y2O3/CeO2 samples was about 1.0 eV, significantly larger than that of the other dopants. Thus, the lattice mismatch in the yttria-doped samples seems to have created biaxial compressive strain in the space charge zones and has significantly increased the migration energy of oxygen vacancies therein. Figure 7 shows the conductivity per interface of the three different types of heterogeneously doped CeO2 films as a

■

CONCLUSIONS The effects of lattice mismatch strain on heterogeneously doped ceria was studied here using films with alternating layers of CeO2 with 1 nm thick dopant oxides of Y2O3, Gd2O3, or La2O3. The previously used Gouy−Chapman model is no longer sufficient to fully explain the ion conduction behavior of heterodoped films becuase the film conductivity increased with increasing lattice parameter of the dopant. Two effects of the lattice mismatch are suggested. First, compressive strain in CeO2 decreases the oxygen vacancy mobility in Y2O3/CeO2 films, as evidenced by an increased activation energy of conduction. Second, strain gradients at the interfaces are suggested to cause more oxygen vacancies to be transferred into the CeO2 space charge regions in the La2O3/CeO2 films. More work is needed to clarify the mechanism(s) behind the conductivity increase in CeO2 heterodoped with La2O3 relative to Gd2O3 and to see whether further increases in the conductivity of heterogeneously doped CeO2 are possible. Such investigations are merited by the fact that the conductivity of these heterogeneously doped systems is already approaching the conductivity of optimal, homogeneously doped systems. Taking advantage of strain gradients and high net dopant content in heterogeneously doped materials may indicate a route to significantly increase solid electrolyte conductivity over what is achievable with traditional doping strategies. Similarly engineered electrolyte materials can potentially be used in micro solid oxide fuel cells or other devices operating at intermediate temperature range (≤650 °C) where cation diffusion is not a concern.

Figure 7. Conductivity per interface obtained from the slope of linear fit of the conductivity as a function of interfaces in Figure 4

function of the temperature. The value of the conductivity per interface was obtained from the slope of the linear fit of the conductivity as a function of interfaces in Figure 4. From the figure, it can be clearly seen that the total conductivity increased from Y2O3 to Gd2O3 to La2O3. On the basis of the analysis of the activation energy in Figure 6, the migration energy and the mobility of the vacancies in space charge regions adjacent to Gd2O3 or La2O3 are likely to be similar. The conductivities of La2O3/CeO2 films, however, were consistently about 2 times greater than those of the Gd2O3/CeO2 films. If these two types of films indeed possess the same vacancy mobility, then there must be ≈2 times more oxygen vacancies in the space charge regions adjacent to La2O3. The actual mechanism behind an increased vacancy concentration in La2O3 space charge regions remains unclear. It may be that the migration of oxygen across the RE2O3/CeO2 interface is more facile for La2O3 relative to Gd2O3. This may result from ion−ion interactions or lattice dynamical effects.16,52 Another clear difference between La2O3 and Gd2O3 is their lattice parameters. For Gd2O3/CeO2 films, there is almost no lattice mismatch at the interfaces, but for La2O3/CeO2, strain is expected: tensile in the CeO2 and compressive in the La2O3. An increase in charge carrier concentration is possibly due to greater strain gradient across the RE2O3/CeO2 interfaces. It has been well-established that the creation of a vacancy in fluorites yields an expansion of the lattice.53 Therefore, a negative lattice strain gradient, with the RE2O3 under compression and CeO2 under tension, provides a driving force to transfer vacancies across the interface into the ceria. Because the CeO2 layers were thicker than the 1 nm thick RE2O3, the strain may be mostly borne by the RE2O3, pushing more oxygen vacancies from the highly compressive strainedLa 2 O 3 into the CeO 2 layers, increasing the vacancy concentration in space charge regions. A similar explanation was also proposed by Fronzi et al. in a numerical study of defect formation at ZrO2−CeO2 interfaces.54 In that work, tensile strain in ZrO2 significantly decreased the vacancy formation energy, whereas the compressive strain in CeO2 significantly

■

AUTHOR INFORMATION

Corresponding Author

*Tel.: +1 302 831 8452. Fax: +1 302 831 3619. E-mail: hertz@ udel.edu. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This research was supported by the U.S. Department of Energy, Office of Basic Energy Science, Division of Materials Science and Engineering under Award DE-SC0005403.

■

REFERENCES

(1) Inaba, H.; Tagawa, H. Ceria-Based Solid Electrolytes. Solid State Ionics 1996, 83, 1−16. (2) Mogensen, M.; Sammes, N. M.; Tompsett, G. A. Physical, Chemical and Electrochemical Properties of Pure and Doped Ceria. Solid State Ionics 2000, 129, 63−94. (3) Fergus, J. W. Electrolytes for Solid Oxide Fuel Cells. J. Power Sources 2006, 162, 30−40. (4) Wachsman, E. D.; Lee, K. T. Lowering the Temperature of Solid Oxide Fuel Cells. Science 2011, 334, 935−939. (5) Minh, N. Q. Ceramic Fuel Cells. J. Am. Ceram. Soc. 1993, 76, 563−588. (6) Zha, S.; Xia, C.; Meng, G. Effect of Gd (Sm) Doping on Properties of Ceria Electrolyte for Solid Oxide Fuel Cells. J. Power Sources 2003, 115, 44−48. (7) Ou, D. R.; Mori, T.; Ye, F.; Takahashi, M.; Zou, J.; Drennan, J. Microstructures and Electrolytic Properties of Yttrium-Doped Ceria 22910



Article

Electrolytes: Dopant Concentration and Grain Size Dependences. Acta Mater. (Weinheim, Ger.) 2006, 54, 3737−3746. (8) Wang, D. Y.; Park, D. S.; Griffith, J.; Nowick, A. S. Oxygen-Ion Conductivity and Defect Interactions in Yttria-Doped Ceria. Solid State Ionics 1981, 2, 95−105. (9) Murray, A. D.; Murch, G. E.; Catlow, C. R. A. A New Hybrid Scheme of Computer Simulation Based on Hades and Monte Carlo: Application to Ionic Conductivity in Y3+ Doped CeO2. Solid State Ionics 1986, 18−19 (Part 1), 196−202. (10) Ou, D. R.; Mori, T.; Ye, F.; Kobayashi, T.; Zou, J.; Auchterlonie, G.; Drennan, J., Oxygen Vacancy Ordering in Heavily Rare-EarthDoped Ceria. Appl. Phys. Lett. 2006, 89. (11) Kilner, J. A. Defects and Conductivity in Ceria-Based Oxides. Chem. Lett. 2008, 37, 1012−1015. (12) Grope, B. O. H.; Zacherle, T.; Nakayama, M.; Martin, M. Oxygen Ion Conductivity of Doped Ceria: A Kinetic Monte Carlo Study. Solid State Ionics 2012, 225, 476−483. (13) Adler, S. B.; Smith, J. W. Effects of Long-Range Forces on Oxygen-Transport in Yttria-Doped Ceria: Simulation and Theory. J. Chem. Soc., Faraday Trans. 1993, 89, 3123−3128. (14) Kilner, J. A. Fast Oxygen Transport in Acceptor Doped Oxides. Solid State Ionics 2000, 129, 13−23. (15) Rushton, M. J. D.; Chroneos, A.; Skinner, S. J.; Kilner, J. A.; Grimes, R. W. Effect of Strain on the Oxygen Diffusion in Yttria and Gadolinia Co-Doped Ceria. Solid State Ionics 2013, 230, 37−42. (16) Burbano, M.; Norberg, S. T.; Hull, S.; Eriksson, S. G.; Marrocchelli, D.; Madden, P. A.; Watson, G. W. Oxygen Vacancy Ordering and the Conductivity Maximum in Y2O3-Doped CeO2. Chem. Mater. 2012, 24, 222−229. (17) Omar, S.; Wachsman, E. D.; Jones, J. L.; Nino, J. C. Crystal Structure−Ionic Conductivity Relationships in Doped Ceria Systems. J. Am. Ceram. Soc. 2009, 92, 2674−2681. (18) Shen, W.; Jiang, J.; Ni, C.; Voras, Z.; Beebe, T. P., Jr.; Hertz, J. L. Two-Dimensional Vacancy Trapping in Yttria Doped Ceria. Solid State Ionics 2014, 255, 13−20. (19) Ando, K.; Oishi, Y.; Hase, H.; Kitazawa, K. Oxygen SelfDiffusion in Single-Crystal Y2O3. J. Am. Ceram. Soc. 1983, 66, C-222− C-223. (20) Kouda, M.; Umezawa, N.; Kakushima, K.; Ahmet, P.; Shiraishi, K.; Chikyow, T.; Yamada, K.; Iwai, H. Charged Defects Reduction in Gate Insulator with Multivalent Materials. 2009 Symposium on VLSI Technology, Digest of Technical Papers; 2009; pp 200−201. (21) Wong, H.; Iwai, H.; Kakushima, K.; Yang, B. L.; Chu, P. K. XPS Study of the Bonding Properties of Lanthanum Oxide/Silicon Interface with a Trace Amount of Nitrogen Incorporation. J. Electrochem. Soc. 2010, 157, G49−G52. (22) Wong, H.; Yang, B. L.; Kakushima, K.; Iwai, H. Subthreshold Characteristics of Mos Transistors with CeO2/La2O3 Stacked Gate Dielectric. IEEE Electron Device Lett. 2011, 32, 1002−1004. (23) Wong, H.; Yang, B. L.; Dong, S. R.; Iwai, H.; Kakushima, K.; Ahmet, P. Current Conduction and Stability of CeO2/La2O3 Stacked Gate Dielectric. Appl. Phys. Lett. 2012, 101, 233507. (24) Meneghesso, G.; Paccagnella, A.; Haddab, Y.; Canali, C.; Zanoni, E. Evidence of Interface Trap Creation by Hot-Electrons in AlGaAs/GaAs High Electron Mobility Transistors. Appl. Phys. Lett. 1996, 69, 1411−1413. (25) Guo, X. X.; Matei, I.; Jamnik, J.; Lee, J. S.; Maier, J. Defect Chemical Modeling of Mesoscopic Ion Conduction in Nanosized CaF2/BaF2 Multilayer Heterostructures. Phys. Rev. B 2007, 76, 125429. (26) Guo, X. X.; Maier, J. Comprehensive Modeling of Ion Conduction of Nanosized CaF2/BaF2 Multilayer Heterostructures. Adv. Funct. Mater. 2009, 19, 96−101. (27) Paranthaman, M.; Lee, D. F.; Goyal, A.; Specht, E. D.; Martin, P. M.; Cui, X.; Mathis, J. E.; Feenstra, R.; Christen, D. K.; Kroeger, D. M. Growth of Biaxially Textured RE2O3 Buffer Layers on Rolled-Ni Substrates Using Reactive Evaporation for HTS-Coated Conductors. Supercond. Sci. Technol. 1999, 12, 319. (28) Bhuiyan, M. S.; Paranthaman, M.; Sathyamurthy, S.; Aytug, T.; Kang, S.; Lee, D. F.; Goyal, A.; Payzant, E. A.; Salama, K. MOD

Approach for the Growth of Epitaxial CeO2 Buffer Layers on Biaxially Textured Ni−W Substrates for YBCO Coated Conductors. Supercond. Sci. Technol. 2003, 16, 1305−1309. (29) Cavallaro, A.; et al. Electronic Nature of the Enhanced Conductivity in YSZ-STO Multilayers Deposited by PLD. Solid State Ionics 2010, 181, 592−601. (30) Hinterberg, J.; Zacherle, T.; De Souza, R. A. Activation Volume Tensor for Oxygen-Vacancy Migration in Strained CeO2 Electrolytes. Phys. Rev. Lett. 2013, 110, 205901. (31) Yildiz, B. “Stretching” the Energy Landscape of Oxides-Effects on Electrocatalysis and Diffusion. MRS Bull. 2014, 39, 147−156. (32) Jiang, J.; Shen, W.; Hertz, J. L. Fabrication of Epitaxial Zirconia and Ceria Thin Films with Arbitrary Dopant and Host Atom Composition. Thin Solid Films 2012, 522, 66−70. (33) Rockenhäuser, C.; Butz, B.; Schichtel, N.; Janek, J.; Oberacker, R.; Hoffmann, M. J.; Gerthsen, D. Microstructure Evolution and Cation Interdiffusion in Thin Gd2O3 Films on CeO2 Substrates. J. Eur. Ceram. Soc. 2014, 34, 1235−1242. (34) González-Romero, R. L.; Meléndez, J. J.; Gómez-García, D.; Cumbrera, F. L.; Domínguez-Rodríguez, A.; Wakai, F. Cation Diffusion in Yttria-Zirconia by Molecular Dynamics. Solid State Ionics 2011, 204−205, 1−6. (35) Izuki, M.; Brito, M. E.; Yamaji, K.; Kishimoto, H.; Cho, D.-H.; Shimonosono, T.; Horita, T.; Yokokawa, H. Interfacial Stability and Cation Diffusion across the LSCF/GDC Interface. J. Power Sources 2011, 196, 7232−7236. (36) Guo, X.; Sigle, W.; Fleig, J.; Maier, J. Role of Space Charge in the Grain Boundary Blocking Effect in Doped Zirconia. Solid State Ionics 2002, 154, 555−561. (37) Sanna, S.; Esposito, V.; Tebano, A.; Licoccia, S.; Traversa, E.; Balestrino, G. Enhancement of Ionic Conductivity in Sm-Doped Ceria/Yttria-Stabilized Zirconia Heteroepitaxial Structures. Small 2010, 6, 1863−1867. (38) Schweiger, S.; Kubicek, M.; Messerschmitt, F.; Murer, C.; Rupp, J. L. M. A Microdot Multilayer Oxide Device: Let Us Tune the StrainIonic Transport Interaction. ACS Nano 2014, 8, 5032−5048. (39) Wirkus, C. D.; Berard, M. F.; Wilder, D. R. Oxygen Diffusion in Gd2O3. J. Am. Ceram. Soc. 1969, 52, 456. (40) Kosacki, I.; Anderson, H. U. Microstructure−Property Relationships in Nanocrystalline Oxide Thin Films. Ionics 2000, 6, 294−311. (41) Gerhardt-Anderson, R.; Nowick, A. S. Ionic Conductivity of CeO2 with Trivalent Dopants of Different Ionic Radii. Solid State Ionics 1981, 5, 547−550. (42) Yoshida, H.; Deguchi, H.; Miura, K.; Horiuchi, M.; Inagaki, T. Investigation of the Relationship between the Ionic Conductivity and the Local Structures of Singly and Doubly Doped Ceria Compounds Using EXAFS Measurement. Solid State Ionics 2001, 140, 191−199. (43) Azad, S.; et al. Nanoscale Effects on Ion Conductance of Layerby-Layer Structures of Gadolinia-Doped Ceria and Zirconia. Appl. Phys. Lett. 2005, 86, 131906. (44) Garcia-Barriocanal, J.; Rivera-Calzada, A.; Varela, M.; Sefrioui, Z.; Iborra, E.; Leon, C.; Pennycook, S. J.; Santamaria, J. Colossal Ionic Conductivity at Interfaces of Epitaxial ZrO2: Y2O3/SrTiO3 Heterostructures. Science 2008, 321, 676−680. (45) Korte, C.; Peters, A.; Janek, J.; Hesse, D.; Zakharov, N. Ionic Conductivity and Activation Energy for Oxygen Ion Transport in Superlattices−The Semicoherent Multilayer System YSZ (ZrO2 + 9.5 mol % Y2O3)/Y2O3. Phys. Chem. Chem. Phys. 2008, 10, 4623−4635. (46) Li, B.; Zhang, J. M.; Kaspar, T.; Shutthanandan, V.; Ewing, R. C.; Lian, J. Multilayered YSZ/GZO Films with Greatly Enhanced Ionic Conduction for Low Temperature Solid Oxide Fuel Cells. Phys. Chem. Chem. Phys. 2013, 15, 1296−1301. (47) Schichtel, N.; Korte, C.; Hesse, D.; Zakharov, N.; Butz, B.; Gerthsen, D.; Janek, J. On the Influence of Strain on Ion Transport: Microstructure and Ionic Conductivity of Nanoscale YSZ| Sc2O3Multilayers. Phys. Chem. Chem. Phys. 2010, 12, 14596−14608. (48) Shen, W.; Jiang, J.; Hertz, J. L. Reduced Ionic Conductivity in Biaxially Compressed Ceria. RSC Adv. 2014, 4, 21625−21630. 22911



Article

(49) Pergolesi, D.; Fabbri, E.; Cook, S. N.; Roddatis, V.; Traversa, E.; Kilner, J. A. Tensile Lattice Distortion Does Not Affect Oxygen Transport in Yttria-Stabilized Zirconia-CeO2 Heterointerfaces. ACS Nano 2012, 6, 10524−10534. (50) Kushima, A.; Yildiz, B. Oxygen Ion Diffusivity in Strained Yttria Stabilized Zirconia: Where Is the Fastest Strain? J. Mater. Chem. 2010, 20, 4809−4819. (51) De Souza, R. A.; Ramadan, A.; Horner, S. Modifying the Barriers for Oxygen-Vacancy Migration in Fluorite-Structured CeO2 Electrolytes through Strain: A Computer Simulation Study. Energy Environ. Sci. 2012, 5, 5445−5453. (52) Buckeridge, J.; Scanlon, D. O.; Walsh, A.; Catlow, C. R. A.; Sokol, A. A. Dynamical Response and Instability in Ceria under Lattice Expansion. Phys. Rev. B 2013, 87, 214304. (53) Marrocchelli, D.; Bishop, S. R.; Tuller, H. L.; Yildiz, B. Understanding Chemical Expansion in Non-Stoichiometric Oxides: Ceria and Zirconia Case Studies. Adv. Funct. Mater. 2012, 22, 1958− 1965. (54) Fronzi, M.; Cereda, S.; Tateyama, Y.; De Vita, A.; Traversa, E. Ab Initio Investigation of Defect Formation at ZrO2-CeO2 Interfaces. Phys. Rev. B 2012, 86, 085407.

22912


Beneficial Lattice Strain in Heterogeneously Doped Ceria - The

Recommend Documents