The Effect of Tartaric Acid and Citric Acid as a Complexing Agent on Defect Structure and Conductivity of Copper Samarium Co-Doped Ceria Prepared by a Sol-Gel Auto-Combustion Method

Materials Science and Nanotechnology Program, Department of Physics Faculty of Science, Khon Kean University, Khon Kaen 40002, #ailand Institute of Nanomaterials Research and Innovation for Energy (IN-RIE), NANOTEC-KKU RNN on Nanomaterials Research and Innovation for Energy, Khon Kaen University, Khon Kaen 40002, #ailand National Metal and Materials Technology Center (MTEC), 114 #ailand Science Park, Paholyothin Rd., Klong 1, Klong Luang, Pathumthani 12120, #ailand National Nanotechnology Center (NANOTEC), 111 #ailand Science Park, Paholyothin Rd., Klong 1, Klong Luang, Pathumthani 12120, #ailand


Introduction
A traditional electrolyte of solid oxide fuel cell (SOFC) is made of yttrium-stabilized zirconia (YSZ), which can efficiently operate at 800-1000°C [1]. Such high operating temperatures come with drawbacks, such as material degradation, a thermal mismatch between fuel cell components, and material incompatibility [2]. Since the oxygen ionic conductivity of ceria-based materials is higher than that of YSZ, ceria-based materials have been recognized as one of the most interesting ionic conductors that operate at intermediate temperatures ranging from 500 to 700°C [1,3]. To enhance the total conductivity of doped ceria materials, the conductivities of both the grain and the grain boundary must be tuned. Grain conductivity mainly relates to oxygen vacancies and densification, whereas grain boundary conductivity mainly corresponds to oxygen vacancies, densification, grain size, segregation, and impurities [4,5]. e oxygen vacancies (V •• O ) were increased by doping ceria materials with rare-earth (RE) cations (Gd 3+ , Sm 3+ , Nd 3+ , etc.) [6][7][8][9]. e formation of V •• O can be described by the Kröger-Vink notation as follows: e V •• O is responsible for ionic conductivity because the oxygen ion (O 2− ) transport through V •• O in the ceria lattice by a hopping mechanism within a grain region and then O 2− ion will move along or across the grain boundary region [10,11].
e V •• O increases with increasing dopant concentration until it reaches an optimal dopant concentration. Anjaneya et al. [4] reported that 20 mol% Sm doped ceria (20SDC) has the highest total conductivity (∼6 × 10 − 5 S/cm at 500°C) and the lowest E a (0.84 eV) value, as compared to 10SDC (∼3 × 10 − 5 S/cm, 0.87 eV) and 30SDC (∼3 × 10 − 6 S/ cm, 1.08 eV). e lowest total conductivity was observed for 30SDC due to an increase in the association of defect (Sm Ce ′ : V •• O ) complex and impurity phases such as SiO 2 and Sm 2 O 3 , resulting in decreased oxygen ion mobility across grain and grain boundaries, respectively. In addition, the conductivity of 10SDC is lower than that of 20SDC because 10SDC has a lower V •• O and its vacancy sites are further apart, making it difficult for oxygen ions to migrate effectively. e processing methods affect both grain and grain boundary conductivity [3,12]. For example, Cioaterȃ et al. [3] compared Gd/Sm co-doped ceria synthesized using Pechini and modified Pechini methods (added Triton X-100 as a surfactant). ey reported that using Triton X-100 increased bulk ionic conductivity by preventing the formation of defect complex clusters during sintering. is effect also increases V •• O even in the space-charge layers, which leads to an increase in grain-boundary conductivity. Kim et al. [12] prepared the aliovalent (Gd 3+ , Sm 3+ , and Nd 3+ ) doped ceria with various dopant contents and various preparation methods, including coprecipitation, combustion, solid-state, and hydrothermal methods. Among these methods, the sample obtained from the combustion synthesis has the highest grain and grain boundary conductivities because it has the highest density due to its porous spongy structure made from connected thin-flake nanocrystallite. Moreover, the grain boundary conductivity of the sample from the combustion method is the highest, which could be due to the fewer impurities. Arabaci and Serin [13] prepared SDC by the Pechini method as compared to the cellulose template method, both of which were sintered at 1400°C. ey found that the sample from the cellulose template method has higher total ionic conductivity (0.0342 S/cm at 750°C) than that from the Pechini method (0.0302 S/cm) due to the larger grain sizes and higher density. Moreover, the cellulose template method produces a sample with lower activation energy (E a ) at all temperatures, which could be attributed to high V •• O diffusion. Khakpour et al. [14] studied the effect of sintering temperature (1300 and 1450°C) and dopant concentration (20SDC and 30SDC) on the sample conductivities.
ere are no significant changes in grain conductivity with different sintering temperatures or grain sizes. However, grain boundary conductivity decreases with increasing sintering temperature/grain sizes. Moreover, the higher dopant concentration increases grain conductivity by increasing V •• O , and it can also increase grain boundary conductivity due to the scavenging of SiO 2 by the Sm 2 O 3 segregation. Also, Pérez-Coll et al. [5] reported that grain and grain boundary conductivities increase with increasing densification. However, grain and grain boundary conductivities only slightly increase when the relative densities are above 91.9% and 93.9%, respectively. e sol-gel auto-combustion method, also known as solgel combustion or modified sol-gel method [15][16][17], is widely used due to simple preparation, short time-toproduct, low cost, higher homogeneity, and fine crystal size with high surface areas in the resulting materials [17,18]. e sol-gel combustion method includes precursors and a complexing agent with/without additional fuels. Note that complexing agents themselves also function as fuel. e different complexing agents in the sol-gel combustion method could impact synthesized materials in terms of purity, homogeneity, and surface area [19]. e complexing agents widely used in sol-gel combustion mainly contain a carboxylic group such as tartaric acid (TA), citric acid (CA), oxalic acid, etc., or a hydroxyl group such as ethylene glycol (EG), glycerol (GL), etc. Notably, EG was commonly used with carboxylic complexing agents to improve complexing capabilities [20,21]. Nonetheless, the doped ceria could also be prepared by a single complexing agent. As demonstrated by Mahata et al. [22], GDC nanoparticles were synthesized by the sol-gel combustion method using CA. Furthermore, it was found that the sample pellets sintered at 1250°C achieved 94-97% density. Note that such a low sintering temperature was due to the addition of a polyvinyl alcohol (PVA) binder as a sintering aid. However, the conductivity of the samples was not reported. Another case was reported by Accardo et al. [18] who also prepared GDC powders by sol-gel combustion method using only CA, and their samples were sintered at 1500°C. eir conductivity of 20GDC measured at 600°C is 0.19 S.cm − 1 , which is higher than that of 10GDC (0.01 S.cm − 1 ) due to higher densification. e effect of different complexing agents on the electrical conductivity of GDC was reported by Zarkov et al. [1]. A complexing agent 2 Advances in Materials Science and Engineering in comparison was TA, ethylene glycol (EG), or glycerol (GL), each of which was added to mixed metal nitrate solution. eir samples were sintered at 1300°C. It was found that the sample from the GL route has the maximum conductivity (0.0037 S/cm) due to its highest pellet density. For an additional body of knowledge, TA should be compared to CA because CA also has functional groups similar to TA but differs in molecular structure and the number of coordination ligands.
To increase densification, high sintering temperatures (such as 1300-1650°C) were used to obtain RE-doped ceria materials. If no sintering aids are used and sintering is conducted at high temperatures, this leads to high manufacturing costs, segregation of defect clusters, failure of co-firing with the cathode supported electrode, and reduction of Ce 4+ to Ce 3+ [23,24]. Moreover, the reduction of doped ceria materials at high temperatures as well as low oxygen partial pressure results in mixed ionic and electronic conductivities (MIEC) [25]. e strategies to reduce sintering temperature have been widely investigated such as tuning particle size/shape and adding the sintering aids [20,26]. e sintering aids with low melting temperatures, such as Cu [20,25], Li [27], Sr [28], and Zn [29], have been proposed to reduce the sintering temperature. Among sintering aids, copper is the most promising candidate as it is inexpensive; moreover, copper metal or copper oxide nanoparticles (which may coexist with copper) have low melting temperatures that promote the effect of liquid-phase sintering [30], resulting in the reduction of sintering temperatures down to 900-1100°C [20,25]. High grain mobility during liquid-phase sintering allows grains to rearrange, merge, and then coarsen at lower temperatures. It was also reported that a ternary phase, Gd 2 O 3 -CeO 2 -CuO, could also form and contribute to the sintering process [25]. To date, there have not been many studies involving Cu/Sm codoped ceria (Cu-SDC) prepared by the sol-gel auto-combustion method. Toor and Croiset [25] produced Cu-20SDC powders with the copper content of 0.1, 0.5, 1.0, 3.0, and 5.0 mol% by the sol-gel combustion method using the only glycine as a complexing agent/fuel. Adding Cu content below 1 mol% results in a low CuO amount at the grain boundaries, which leads to a high value of grain boundary conductivity at low temperatures. eir total conductivities of 0.5Cu-20SDC and 1Cu-20SDC are ∼0.007 and ∼0.006 S.cm − 1 at 600°C, respectively. Lü et al. [31] and Dong et al. [15] prepared 1Cu-20SDC by the sol-gel combustion method with PVA as a complexing agent/fuel. ey observed that the total conductivities are 0.0146 and 0.0154 S.Cm − 1 at 600°C, respectively.
is study examined the effects of TA and CA used as a single complexing agent for preparing Cu0.01Sm0.19-Ce0.80O2-d powder (CSDC), which could reveal additional useful information, particularly changes in thermal behaviors, as-synthesized morphologies, densities, V •• O , segregations, and electrical conductivities. Moreover, CSDC powder without additional complexing agents/fuels has never been reported.

Powder Synthesis.
e composition of CSDC powders in this work was Cu 0.01 Sm 0.19 Ce 0.80 O 1.90 , which was synthesized by the sol-gel auto-combustion method via three different routes. Among these were the use of pure deionized water (CSDC/DI) as a reference sample and an additional complexing agent, either L-(+)-tartaric acid (CSDC/TA) or citric acid (CSDC/CA). Metal nitrate precursors included Ce(NO 3 ) 3 .6H 2 O (99.9%, Sigma-Aldrich), Sm(NO 3 ) 3 .6H 2 O (99.9%, Sigma-Aldrich), and Cu(NO 3 ) 2 (99.9%, Sigma-Aldrich). A complexing agent, either tartaric acid (TA, 99.9%, Sigma-Aldrich) or citric acid (CA, 99.9%, Sigma-Aldrich) was used. All metal nitrates were dissolved in de-ionized water at a total concentration of 0.1 M. e mixed metal nitrate solution is CSDC/DI. For CSDC/TA or CSDC/CA, TA or CA was added to the mixed metal nitrate solution at a molar ratio of metals: complexing agent of 1 : 1. After being thoroughly dissolved, each mixed solution was heated to 80°C and continuously stirred until the yellow gel was formed. Each gelled samples was heated in a muffle furnace at 300°C. en, the as-synthesized powder was transferred into a crucible and calcined at 800°C for 3 h. e calcined powder was ground by dry ball milling for 3 h using zirconia balls with a diameter of 3 mm.

Pellet Preparation.
Each calcined CSDC powder with the weight of 0.5000 ± 0.0005 g was subjected to a uniaxial compression of 5 MPa by using a hydraulic press. e green CSDCs pellets with 13 mm diameter were sintered at 1100°C in a muffle furnace at a heating rate of 2°C/min, with a holding time of 3 h and then cooled to 500°C at a rate of 10°C/min. After reaching 500°C, the furnace was shut down and allowed to cool to room temperature. A summary of sample preparation steps is shown in Figure 1.

Characterization.
e resulting gel samples were analyzed by using simultaneous thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) (TGA/ DSC1, Mettler-Toledo, Switzerland) from room temperature to 1100°C at a constant heating rate of 10°C/min with air--zero flow rates of 20 ml/min. e reaction enthalpy can be determined by integrating the area of the reaction peak. Scanning electron microscope (SEM) (SNE: 4500-M, SEC, South Korea) was used to characterize the morphology of the as-synthesized samples and sintered pellets. e calcined CSDC powders were characterized using X-ray diffraction (XRD) (X'Pert3, Malvern Panalytical, USA) with a 2θ scan range of 20-90°, Fourier transform infrared spectroscopy (FT-IR) (TENSOR27, Bruker, USA) over a range of 4000-600 cm − 1 , Raman spectroscopy (T64000 Triple monochromator, JOBIN YVON HORIBA, France) with a wavelength of 532 nm, high-resolution transmission electron microscope (HRTEM) (TecnaiTM G2 20, FEI, USA and X-ray photoelectron spectroscopy (XPS)) (AXIS Ultra DLD, Kratos Analytical, UK). In addition, the sintered CSDC powders were assessed in terms of defect structure and chemical composition using Raman spectroscopy and XPS, respectively. e electrical properties of CSDC pellets were estimated from impedance spectroscopy (PAR4000, Princeton Applied Research (PAR), USA) with a scanning range from 4 MHz to 0.1 Hz.
Advances in Materials Science and Engineering e relative density (r.d.) was calculated from the relationship between bulk density (ρ m ) and theoretical density (ρ th ). e theoretical density was calculated using the following equation [15,31]: where a is a lattice parameter calculated from XRD, N A is Avogadro's number (6.022 × 10 23 ), x is Sm content (Sm = 0.19), and y is Cu content (Cu = 0.01). e Archimedes method was used to determine the bulk densities of sintered CSDC pellets. For conductivity measurement, silver paste (Ag-paste) was painted on both sides of the sample surfaces in a defined area of 0.28 cm 2 and then the painted pellets were dried at 80°C on a hot plate. e Ag-wires were connected on both sides of the Ag-paste. e samples were placed in the holder and heated in a tube furnace at 850°C for 30 min. e conductivities were measured in a high-temperature range from 800°C to 400°C with an increment of 100°C and a lowtemperature range from 400°C to 250°C with an increment of 50°C. e total resistivity (ρ tot ) was calculated from the combination of grain and grain boundary resistivity (ρ tot = ρ g + ρ gb ) and the conductivity was derived from the following equation [15,31,32]: where σ tot is the total conductivity (S.cm − 1 ), ρ tot is the total resistivity (Ω.cm), L is sample thickness (cm), R is resistance from impedance spectra (Ω), and A is an active area (cm 2 ). e activation energy (E a ) was determined from the slope of the Arrhenius plot of ln (σT) versus 1/T following equation [8,31]: where σ 0 is the preexponential factor, k B is Boltzmann's constant (8.617 × 10 − 5 eV.K − 1 ), and T is the absolute temperature in Kelvin.

Results and Discussion
3.1. ermal Analysis. e thermal behavior of Cu-Sm-Ce-O synthesized gels was analyzed using simultaneously TGA/ DSC. e results described a complexing agent effect on the solid solution formation of the CSDC compound, which may be different.
is thermal behavior is related to the energy released during heating the precursor gels, as shown in Figure 2. Differential thermogravimetric (DTG) analysis was used to confirm the mass loss during the various stages in the TGA curve. Interestingly, the decomposition of metal nitrate hydrate is unique for the fact that metal hydroxide can form without any alkaline additives, which was analyzed by the TGA-MS technique [33,34]. Herein, we also added the DSC technique to confirm the thermal behavior of metal nitrate decomposition. For TGA and DTG curves, the CSDC/DI gel (Figure 2(a)) shows a mass loss of 16.1% in the 1 st step over the temperature range of 63-130°C. is is due to the evaporation of water in the gel, which is related to an endothermic peak (i) with heat flow of 623.5 mJ and enthalpy change (ΔH = − 134.4 J/g) on DSC curves as shown in Figure 2(d). Afterward, the integral area of the 2 nd small exothermic peak (ii) corresponding to heat flow of 109.1 mJ (ΔH = 23.5 J/g) is observed while the mass loss on the TGA curve slightly decreases. e thermal behavior is possibly caused by the formation of metal nitrate hydroxide compound as well as M(OH) b (NO 3 ) (z-b) .H 2 O (a-b-c) and other byproducts including HNO 3 and H 2 O. A further mass loss of the 2 nd step on the TGA curve is 10.5% (198-207°C), which corresponds to an endothermic peak (iii) of about 571.1 mJ -c) to N x O y gases (x and y = 1 or 2), coordinated water, and oxygen. For the last TGA step, a mass loss of 33.7% (286-328°C) corresponds to an endothermic peak (iv) of heat flow of 3038.7 mJ (ΔH = − 654.9 J/g), which can be ascribed to the decomposition of most hydroxides as well as some metal oxide formation. e TGA/DTG result of CSDC/TA is shown in Figure 2 intermolecular complexation of some metal-TA (crosslinking), while the TGA curve still slightly decreases in the temperature range of 139 to 170°C. e cross-linking of the metal-TA complex leads to the formation of an organometallic framework as depicted in Figure 3(a). e mass loss of the 2 nd step on the TGA curve is 10.6% (175-195°C), which results from the decomposition of the nitrate and organic compounds from some metal-TA complexes. At the same time, the large exothermic peak (iv) with a heat flow of 847.9 mJ (ΔH = 175.9 J/g) is observed in this step. Despite the fact that N x O y gases are not combustible, they are a strong oxidizer that increased combustion energy, and the heat release is attributed to the combustion reaction of carbon and O 2 with N x O y gases as an accelerant under wet conditions. For the 3 rd step on the TGA curve, the mass loss is 12.1% (284-307°C) due to the decomposition of organic residue, remnant nitrate, and most hydroxide. Moreover, this step is accompanied by a weak exothermic peak (v) with a heat flow of 69.0 mJ (ΔH = 14.3 J/g), which may be caused by the combustion of carbon residue and remnant N x O y accelerants under drier conditions. e last step on the TGA curve in the temperature range from 326 to 373°C shows a mass loss of 3.9%, and the DSC curve shows a small endothermic peak (vi) with a heat flow of 179.4 mJ (ΔH = -37.2 J/g). is thermal behavior may be caused by the thermal decomposition of the CSDC compound due to oxidation phenomena of Ce 3+ to Ce 4+ , as reported by Ambrozini et al. [35]. is DSC stage, on the other hand, could be associated with the decomposition of some copper hydroxy-carbonate (malachite) into copper oxide, water, and carbon dioxide [36].
For the CSDC/CA gel (Figure 2(c)), the TG/DTG curve shows a mass loss at the 1 st step of 37.9% (68-102°C) due to residual water evaporation, which shows an endothermic peak (i) with a heat flow of 2883.0 mJ (ΔH = − 613.4 J/g). Afterward, the small endothermic peak (ii) of about 114.9 mJ (ΔH = − 24.5 J/g) is observed in the temperature range from 129 to 138°C, while the TGA curve slightly decreases simultaneously due to dehydration of water molecules. e 2 nd and 3 rd steps on the TGA curve are overlapping due to the partial consecutive losses of metal-CA complex compound [35], which involves 11.3% (139-166°C) and 7.3% mass loss (195-217°C).
ey are related to the removal of coordinated water molecules and the decomposition of nitrate and an organic compound. Both steps exhibit exothermic behaviors on the DSC curve ). e first exothermic peak (iii) has a heat flow of about 249.9 mJ (ΔH = 53.2 J/g) which may be caused by metal-citrate formation (metal-CA complex) and metal-CA complex cross-linking. A possible illustration of the metal-CA cross-linkage is shown in Figure 3(b). e 3 rd step on TGA curve is related to the exothermic peak with a heat flow of 1124.9 mJ (ΔH = 239.4 J/g), which could be attributed to heat release from the combustion reaction between carbon and O 2 with N x O y gas accelerant under wet conditions.
For the final step on the TGA curve, a mass loss of 14.5% is observed in the temperature range from 292 to 319°C due to the decomposition of organic residue, remnant nitrate, and hydroxide, and the release of carbon as well as metal oxide formation, which corresponds to an exothermic peak with a heat flow of 2613.8 mJ (ΔH = 556.1 J/g) and to the combustion reaction between carbon with N x O y gas accelerant under dry conditions. When CSDC/TA gel is compared to CSDC/CA gel, the integral of the exothermic peak (iii) at ∼150°C for CSDC/ TA is lower than that of CSDC/CA because CSDC/TA has a lower degree of cross-linking than CSDC/CA. is leads to lower heat release at ∼200°C on the DSC curve of CSDC/TA and lower combustion temperature, as well. Because TA has a lower cross-linking degree, it releases fewer N x O y gases. Moreover, the amount of carbon and oxygen in TA is lower than that of CA, resulting in less combustion energy. e final heat release (v) at ∼300°C of CSDC/TA is much lower than that of CSDC/CA at ∼325°C due to the lower amount of carbon residue as some carbon in CSDC/TA transforms into carbonate species such as copper hydroxycarbonate (malachite), which exhibits a weak endothermic peak at ∼365°C.

FT-IR Analysis.
FT-IR was used to confirm the presence of carbon-oxygen residues after calcination. e FT-IR spectra of calcined CSDCs prepared by DI, TA, and CA routes are presented in Figure 4. All samples show vibration bands at ∼3500 cm − 1 , ∼1650 cm − 1 , 1500 cm − 1 , and ∼1300 cm − 1 , which are attributed to the O-H stretching of water, O-H stretching, O-C-O stretching, and N-O stretching, respectively [15,18,39,40]. e O-H vibration mode of water molecules is observed due to moisture adsorption or possibly Ce-OH at sample surfaces [41]. Moreover, the O-H vibrational modes come from the amorphous phases of hydroxyl species, such as Ce(OH) 3 and Ce(OH) 4 or CeO 2 .2H 2 O [42]. e probable formations of Ce(OH) 3 and Ce(OH) 4 as well as the derivation of CeO 2 nanoparticles are as follows [43]: However, the C-O stretching is also observed in calcined CSDC/DI, which results from CO 2 absorption at the particle surfaces from the environment [41]. Moreover, the carbonate groups in the FT-IR spectra may have been caused by the reaction of nitrate and CO 2 (from burning complexing agents) during the combustion stage according to the following reaction [18]: Moreover, the vibration band at 850-400 cm − 1 is the characteristic of Ce-O bonds [18,44]. Although CSDCs were calcined at 800°C, the FT-IR spectra still show -OH vibration, carbon, and carbonate residues from burnt complexing agents that may be trapped along grain boundaries of nanoparticles during calcination. Notably, the residues of an organic compound can still be detected after calcination at temperatures of up to 1000°C, which has been reported by Kalyk et al. [45].  Advances in Materials Science and Engineering Figure 5 shows various morphologies of as-synthesized CSDCs that were prepared using three different nitrate precursors: pure DI, or TA complex, or CA complex. e as-synthesized CSDC/DI has both a layered structure (yellow arrow) and small agglomerated particles (green arrow), as shown in because the bubbles in CSDC/CA can be extensively generated and expanded more than those of CSDC/TA during the gas-release stage. CA can stretch more than TA when bubbles enlarge because CA has a higher viscosity. In addition, the CA has a branch structure and more coordination, resulting in more cross-links in the CA route. Moreover, CA has a higher carbon content than TA. erefore, the as-synthesized CSDC/ CA exhibits stronger combustion than the as-synthesized CSDC/TA, which leads to a higher energy release that is confirmed by the DSC result.

Transmission Electron Microscopy (TEM).
TEM was used to characterize the crystallinity and particle sizes of ungrounded samples after the calcination of nanoparticle powders. It was found that geomertry of all samples are mostly polyhedral and the particles demonstrate agglomeration as seen in Figures 6(a)-6(c). Because CSDC/ DI nanoparticles form more metal hydroxide than CSDC/ TA and CSDC/CA nanoparticles, it shows higher agglomeration with lower dispersion than the others. Moreover, TA and CA can form an organometallic complex, which improves metal ion dispersion in solution. In addition, CA has a branched molecule and more coordination ligands, resulting in greater dispersion of nanoparticle sizes than TA. e particle sizes of CSDC samples are shown in Figures 6(d)-6(f ), while the histograms in the inserted figures show the particle size distribution. e CSDC/TA and CSDC/CA have greater combustion than CSDC/DI; however, the particles of all samples are nonuniform in size. Moreover, the formation of organometallic compounds can inhibit nanoparticle agglomeration during the auto-combustion process, as well as increase the degree of cross-linking, which leads to more combustion energy. e average particle sizes of CSDC/DI, CSDC/TA, and CSDC/CA are approximately 35.5 ± 9.9 nm, 35.0 ± 10.3 nm, and 35.2 ± 11.1 nm, respectively. Nevertheless, the average particle sizes are close to one another because only dry suspended nanoparticles were measured using the TEM technique. e crystallinities of CSDC nanoparticles are estimated by selected area electron diffraction (SAED). e SAED patterns of CSDC nanoparticles are shown in Figures 6(g)-6(i). e CSDC/DI nanoparticles have poor crystallinity as shown in Figure 6(g), which may have been caused by the presence of amorphous phases of hydroxyl species such as Ce(OH) 3 and Ce(OH) 4 or CeO 2 .2H 2 O at the particle surfaces [42], as well as amorphous CuO.
is leads to decreased crystallinity. e broad reflection rings of CSDC/TA and CSDC/CA nanoparticles show eight rings that can be ascribed to (111), (200), (220), (311), (222), (400), (331), and (420) [15]. e CSDC/TA and CSDC/CA nanoparticles have higher crystallinity than the CSDC/DI nanoparticles because the complexing agent can produce metal-complexes, which decrease the coordinated -OH, and another reason is the existence of an exothermic behavior, which results in faster metal oxidation. According to TEM analysis, the complexing agents can aid in crystallinity improvement.

XRD Patterns.
e XRD patterns of all calcined CSDCs show an FCC fluorite CeO 2 phase (JCPDS Card No. 00-034-0394) with no detected Sm 2 O 3 or CuO, which are typical impurity phases as shown in Figure 7(a). However, it is difficult to identify CuO impurity phase because the level of Cu is lower than the detection limit of the XRD technique [21]. Because the calcined CSDC/CA has the highest peak intensity, it has higher crystallinity than the calcined CSDC/ TA and CSDC/DI, which confirms the SEAD data. In addition, the XRD pattern can be used to calculate a lattice parameter by Bragg's relationship. e change in 2θ peak shift and lattice parameters depends on doping efficiency and dopant size. Higher doping efficiency and larger dopant size should lead to a decrease in 2θ peak positions and an increase in lattice parameters. us, the substitution of Sm 3+ into Ce 4+ sites should increase the lattice parameter and 2θ peaks shift to lower angles because the cationic radius of Sm 3+ (0.108 nm) is larger than Ce 4+ (0.097 nm) [8]. Apart from the dopant effect, the complexing agents also have a clear influence on the shift of 2θ peaks in calcined SDC powders without copper (SDC/DI < SDC/TA < SDC/CA), as shown in Figure S1. As a result, the complexing agent affects the solubility of Sm in ceria. When copper is added to SDCs, the trend in the 2θ peak shift is uncertain, except for the (220) peak, as shown in Figure  S2.
e lattice parameters of calcined CSDC powders are averaged from all peak positions, as shown in Figure 7(b). e average lattice parameters of calcined CSDC/TA and CSDC/ CA are slightly higher than that of calcined CSDC/DI. When calcined CSDC/CA is compared to calcined CSDC/TA, the Advances in Materials Science and Engineering average lattice parameter of calcined CSDC/CA is close to that of CSDC/TA, which may be due to more substitution of Cu + (0.074 nm) and/or Cu2 + (0.071 nm) into Ce4 + [46]. e crystallite sizes (D) are estimated from Scherrer's equation as follows [8,47]: where λ is X-ray wavelength (Cu K α1 = 1.5418Å), θ is the Bragg diffraction angle of peak max, and β is full-width half max (FWHM) in radians. e average crystallite sizes were calculated from all peak locations and it was found that CSDC/ DI ( (111) is the highest due to its strongest exothermic reaction, whereas the crystallite size of calcined CSDC/TA is slightly higher than that of calcined CSDC/DI because the calcined CSDC/TA has less hydroxide formation due to the metal-TA complexing and it also exhibits exothermic reaction.
3.6. Visible Raman Spectroscopy. Visible Raman spectroscopy is a highly sensitive technique that can be used to analyze the local structure of nanomaterials, which relates to both bulk and surface defects [48]. us, Raman spectroscopy can detect V •• O and distinguish the F-structure (fluorite of CeO 2 ) and C-structure (cubic of Sm 2 O 3 ) [49,50]. e Raman spectra of calcined and sintered CSDCs shown in Figure 8 exhibit main peaks at around 250, 460, 550, and   Figure 6: e HRTEM images (a-c), high magnification of HRTEM images with particle size distribution histograms (d-f ), and SAED patterns (g-i). e left side represents CSDC/DI nanoparticles, the center represents CSDC/TA nanoparticles, and the right side represents CSDC/CA nanoparticles.   10 Advances in Materials Science and Engineering assigned to SmO 8 -type complex in this case [48]. In addition, a very weak Raman band at ∼350 cm − 1 in calcined and sintered CSDCs is assigned to copper segregation such as CuO [59] could overlap with the C-structured Sm 2 O 3 (∼370 cm − 1 ) [56]. Because the XRD peak shift cannot clearly explain the dopant solubility into the ceria lattice for calcined CSDCs, Raman spectroscopy is analyzed and used to extract solubility information based on F 2g characteristics (intensity, peak shift, broadening, and shoulders), V •• O , and segregation. e F 2g intensities of calcined and sintered CSDCs, as shown in Figures 8(a) and 8(b), are in the following order: CSDC/ DI > CSDC/TA > CSDC/CA. When TA or CA was added, the F 2g intensity decreases because there are fewer Ce-O symmetry vibrations and an increase in lattice disorder [52] due to the improvement of Sm and Cu substitution. To confirm the efficacy of TA or CA in doping, the F 2g shift should be analyzed. Because of the radius of Sm 3+ > Ce 4+ > Cu 2+ , the replacement of Ce 4+ with Sm 3+ causes F 2g to shift to lower wavenumber (red-shift), resulting in lattice expansion, whereas Cu 2+ substitution causes F 2g to shift to higher wavenumber (blue-shift), resulting in lattice contraction. However, the shift of F 2g induced by the Cu 2+ ions is insignificant because of the low Cu concentration of only 1 mol%. e F 2g peak positions of calcined and sintered CSDCs follow the same trend: CSDC/CA < CSDC/ TA < CSDC/DI. Moreover, the sintered CSDCs exhibit a slight red-shift in F 2g compared to the calcined CSDCs due to the removal of Cu and hydroxide species (especially CSDC/DI). e interpretation of F 2g intensity and peak shift confirms that the addition of TA and CA affects doping efficiency. e CA could facilitate more metal-CA formation than metal-TA, so the substitution of Sm and Cu into ceria can be the best in the CA case. e intensities of all Raman spectra have been normalized to compare the broadening of F 2g . e broadening of F 2g is caused by the decrease in lattice symmetry [58] and/or the decrease in crystal sizes as well as an increase in defects [60]. Moreover, the left shoulder of F 2g at ∼404 cm − 1 is related to lattice distortion [48], whereas the right shoulder of F 2g is S-mode (∼470 cm − 1 ) which could be ascribed to a surface mode appearing in very small crystallites like CeO 2-δ [61] or relates to the nonequivalence of cation-O bonds in doped or reduced ceria [62]. From Figure 8(c), the S-mode broadening of calcined CSDC/CA and CSDC/TA are higher than that of calcined CSDC/DI, indicating the increase of Sm and/or Cu solubility in ceria. However, the S-mode of sintered CSDCs decreases (Figure 8(d)) with increasing particle size and oxygen intake, which results in the improvement of lattice ordering [61].
From Figure 8, the calcined and sintered CSDCs have a very weak Raman band at ∼350 cm − 1 , which may be attributed to the Raman band of Cu 2+ ions segregating such as CuO. Rashad et al. [59] reported the Raman bands of CuO shown at 282, 330, and 616 cm − 1 . It is worth noting that the Raman band at 616 cm − 1 cannot be detected because it is situated within the D2 band. e calcined CSDC/DI has more CuO segregation because of the dehydration of Cu(OH) 2 and partial transformation into CuO when the temperature is higher than 70°C [63]. It is possible that Cu 2+ in TA and CA routes is not only in the form of CuO but also copper carbonate (CuCO 3 ) due to reaction between carbon and Cu 2+ during the calcination process since the C-O bond is observed in the FT-IR. In addition, a peak of the copper segregation phase at ∼354 cm − 1 may overlap with that of Sm 2 O 3 (∼370 cm − 1 ), if the C-structured Sm 2 O 3 segregate exists.
e extrinsic V •• O (D1) is generated from the substitution of Ce 4+ with Sm 3+ and Cu 2+ and/or Cu + (equations (8) and (9)), whereas intrinsic V •• O (D2) are generated from the partial reduction of Ce 4+ to Ce 3+ (equation (10)) due to high temperature/low oxygen partial pressure [50,52], nonstoichiometry of ceria [57,64], and redox reaction of Cu + + Ce 4+ ↔ Cu 2+ + Ce 3+ above 450°C [65], which leads to lattice expansion. In addition, the D2 peak can also be assigned to Frenkel-related defects (equations (11) and (12)). e formation of extrinsic , and Frenkel defects written in Kröger-Vink notations were summarized as follows [57,64]: Extrinsic oxygen vacancy from Sm 3+ : Extrinsic oxygen vacancy from Cu ions: Intrinsic oxygen vacancy and Frenkel defect: e relative oxygen vacancy concentration, [48], which is calculated after Gaussian fitting [54,58]. Besides, the relative concentration of SmO 8 , [SmO 8 ], is calculated from A SmO8 /A F 2g . Note that F 2g areas of calcined and sintered CSDCs include the area of F 2g shoulders (∼450 and ∼470 cm − 1 ). e relative concentrations of defects are investigated by comparing these area ratios, as shown in Figure 9.  Figure S3. Since the ratio of Sm 2 O 3 to F 2g peak ( Figure S4(a)) is small, the Sm 2 O 3 may segregate in the form of nanodomains phase rather than microdomains phase. e C-structure of Sm 2 O 3 nanodomains was also reported for the Sm concentration of Advances in Materials Science and Engineering ∼20 mol%, which was detected by μ-Raman spectroscopy [66]. Moreover, the earlier work from Artini et al. [56] showed the obvious segregation of C-structured Sm 2 O 3 as determined by μ-Raman spectroscopy and Synchrotron X-ray Diffraction when the Sm 3+ concentration was 30 , is shown in Figure 9 is increased due to increasing nonstoichiometry, which results from the decrease in the nanoparticle size [67]. In the presence of Cu 2+ , the effect of the redox reaction of Cu + + Ce 4+ ⟶ Cu 2+ + Ce 3+ could produce Cu 2+ which exists in three forms, including CuO or Cu-Sm-Ce solid solution, all of which can coexist. Comparing calcined CSDC/TA with calcined CSDC/CA, the calcined CSDC/TA has the higher [V •• O ] int. because it has smaller crystallite sizes according to the XRD result ( Figure 7(b)). e TA route shows a smaller crystallite size because crystal growth is impeded by the presence of CuO and CuCO 3 . More of these extra phases appear in the TA route than in the CA route because TA contains fewer coordination ligands, resulting in more free metal ions that readily transform into copper segregation, as shown in Figure S4(b). is implies that TA is not as good as CA in promoting the formation of the Cu-Sm-Ce solid solution. However, calcined CSDC/DI has the highest concentration of copper segregation, which implies the least amount of the formation of Cu-Sm-Ce solid solution.  Figure 9: e relative concentration of (a) extrinsic . Another type of segregate mentioned in literature is the SmO 8 -type complex, which causes the presence of microdomains in the microstructure with a predominant metastable C-type phase [47]. In this study, the relative concentration of the SmO 8 -type complex, [SmO 8 ], is shown in Figure 9( 8 ] as compared to the calcined CSDC/CA (from 0.0048 to 0.01), which may be caused by the diffusion of Cu ions from Cu-Sm-Ce solid solution into CuO along grain boundary regions during the sintering process. e possible explanation is that a redox reaction: Cu + + Ce 4+ ↔ Cu 2+ + Ce 3+ has occurred, which corresponds to the reduction of Ce 4+ , resulting in the increasing of [V •• O ] int. (Figure 9(b)), as well as CuO segregation ( Figure S4(a)) for sintered CSDC/CA. However, CuO amount in sintered CSDC/CA is closed to those in other samples because a certain degree of CuO is evaporated during the sintering process, thereby increasing the metastable SmO 8 (as well as Sm 2 O 3 nanodomains) of sintered CSDC/CA.

XPS Spectra.
e XPS was a technique used to evaluate the surface chemical composition and oxidation states of the materials. Since Raman spectroscopy was used to characterize the materials at a deeper depth than XPS, the trend of Ce 3+ and Cu ion concentration from XPS may be different from that of the Raman results. Moreover, the XPS technique was measured under ultra-high vacuum which induced reduction conditions so that the Ce 3+ content can be overestimated. Cerium has two oxidation states, Ce 3+ and Ce 4+ , which can be observed in high-resolution Ce3d spectra extending from 875 to 925 eV. e detected Ce 3+ can exist in the Ce 3+ -V •• O and Ce 2 O 3 and/or Ce(OH) 3 forms. e formation of Ce 3+ -V •• O is not only generated by external doping but also created under high temperature and/or low partial oxygen pressure. e coexistence of Ce 3+ and Ce 4+ as present in Figures 10(a) and 10(b) is confirmed by the high-resolution Ce3d XPS spectra of calcined and sintered CSDCs. e different peaks of Ce d 5/2 and Ce d 3/2 spin-orbital components are attributed to v and u [71]. In addition, the v, v ″ , v ″′ , u 0 , u″, and u‴ refer to the Ce 4+ final state, which is ascribed to Ce 4+ -3d 9 4f 1 O2p 4 for the doublet v/u, Ce 4+ -3d 9 4f 1 O2p 5 for the doublet v ″ /u″, and Ce 4+ -3d 9 4f 0 O2p 6 for the doublet v ″′ /u‴, whereas u‴ is not observed in pure CeO 2 [72]. Doublet v 0 , v ′ , u 0 , and u′ refer to the Ce 3+ final state and the binding energy doublet v ′ /u′, and v 0 /u 0 corresponds to Ce 3+ -3d 9 4f 1 O2p 6 and Ce 3+ -3d 9 4f 2 O2p 5 , respectively [73]. However, v 0 and u 0 could not be fitted because they have such low intensities. e relative concentration of Ce 3+ , [Ce 3+ ], is calculated from the ratio of the total area of Ce 3+ and that of Ce 4+ as follows [57]: where  Table 1 is decreased when adding TA or CA because Sm 3+ and Cu ions can substitute into Ce 4+ sites, which leads to the formation of Cu-Sm-Ce solid solution. Since CA has a high carbon source, the oxidation reaction of Ce 3+ + Cu 2+ ⟶ Ce 4+ + Cu + during CO release [65] could occur in the CA route more than other routes, which leads to decreasing [Ce 3+ ] value. However, the V •• O at the particle surface could be depleted by CO 2 adsorption from the atmosphere. Chang et al. [69] reported that the reaction involving CO 2 reduction by CeO 2-δ could occur without heat or with a low CO 2 partial pressure, and that if a catalyst metal such as Cu is present on Ce 3+ , dissociation of CO 2 to CO will occur rapidly. Note that the Ce 3+ from the XPS result is not only related to the reduction of Ce 4+ to Ce 3+ from cation replacement but also the segregation of Ce 2 O 3 and/or Ce(OH) 3 at the surface. If interpretation is based on the Ce 3+ segregation, then CA is confirmed to be a better complexing agent than TA as CA induces the least segregation.
For sintered CSDCs, there are no significant differences in [Ce 3+ ] values. e values of [Ce 3+ ] are 20.57%, 21.85%, and 20.26% for CSDC/DI, CSDC/TA, and CSDC/CA, respectively. When calcined and sintered CSDCs are Advances in Materials Science and Engineering compared, the [Ce 3+ ] value of sintered CSDC/DI is found to be lower than that of calcined CSDC/DI due to the partial transformation of Ce(OH) 3 to CeO 2 . However, the [Ce 3+ ] values of sintered CSDC/TA and CSDC/CA are more than those of calcined ones because of the reduction of Ce 4+ to Ce 3+ and the oxidation of Cu + to Cu 2+ at 1100°C. e Ce 3+ of sintered CSDC/TA and CSDC/CA concentrates at the surface because the Cu-O and Ce-O bonds in the Cu-Sm-Ce solid solution are weaker than those that are present in CeO 2 , so the metal-oxygen bonds tend to break and produce reactive oxygen species [74]. e oxygen peaks provide information about the oxidation states of cerium and confirm the coexistence of Ce 4+ and Ce 3+ . e O1s XPS spectra of calcined and sintered CSDCs are shown in Figures 10(c) and 10(d). Deconvolution of O1s core level spectra yields four peaks. e first peak (i) at ∼528 eV is attributed to the lattice oxygen of Ce 4+ -O [73,75]. e overlapping peak (ii) at ∼529 eV is attributed to Ce 3+ -O [55,58,73] because Ce 3+ -O has a stronger bond than Ce 4+ -O, so the binding energy of Ce 3+ -O is higher than that of Ce 4+ -O [55,68,73]. e third peak (iii) at ∼531 eV corresponds to Sm 3+ -O because Sm has a higher electronegativity than Ce on the Pauling scale (Sm = 1.71 and Ce = 1.12) [55,57], leading to higher binding energy. e last peak (iv) at ∼533 eV is related to water adsorption [58] or oxygen species adsorption such as O − , OH − , and CO 2− 3 [58,75], resulting from reactions with V •• O at the  14 Advances in Materials Science and Engineering nanoparticle surfaces [76] or nonstoichiometric oxygen [77]. e relative oxygen concentration of Ce 3+ (O 1s Ce 3+ ) as present in Table 1 was calculated from the peak area of O1s spectra as follows [58]: where O 1s Ce 4+ , O 1s Sm 3+ , and O 1s ads. are oxygen concentrations of Ce 4+ , Sm 3+ , and oxygen adsorption, respectively. e Ce 3+ from XPS spectra may exist either in the Ce 2 O 3 form or the Ce 3+ -V •• O form. e Ce 3+ -V •• O form created by the dopant cations substitution into Ce 4+ ions can be verified by comparing the stoichiometry ratio (χ) with actual stoichiometry (χ′), which are calculated following equations (16) and (17), respectively [57]. If χ′ is more than χ, the Ce 3+ -V •• O is formed without the presence of Ce 2 O 3 ; however, if χ′ is less than χ, Ce 2 O 3 and Ce 3+ -V •• O co-exist [57,78]. Table 1 lists the calculated value of χ and χ′ for calcined and sintered CSDCs.
where O 1s denotes the total area under Ce peaks, Ce 3d denotes the total area under Ce peaks, and S Ce = 7.399 and S O = 0.711 denote sensitivity factors of Ce and O atoms, respectively. For all sintered CSDCs, Ce 3+ -V •• O (related to CeO 2-δ ) is predominant in the sintered CSDC/DI and CSDC/CA since χ′ is more than χ. Considering [Ce 3+ ] values, the sintered CSDC/DI decreases from that of the calcined one because of the removal of hydroxide species, so Ce 3+ turned into Ce 4+ , and Ce 4+ could subsequently partially reduce to Ce 3+ at high temperature; thus, Ce 3+ -V •• O is predominant in the sintered CSDC/DI. In contrast, the [Ce 3+ ] of sintered CSDC/CA is higher than that of the calcined one because the Cu ions diffuse from solid solution and could have reacted with Ce 4+ (Cu + + Ce 4+ ⟶ Cu 2+ + Ce 3+ ). Moreover, this effect takes place along with the thermally induced partial reduction of Ce 4+ to Ce 3+ , resulting in the domination of Ce 3+ -V •• O . Although Ce 2 O 3 and Ce 3+ -V •• O coexist in sintered CSDC/TA because χ′ is less than χ, Ce 3+ -V •• O could still be a major species rather than Ce 2 O 3 due to the decrease of Ce 3+ -O after sintering.
e high-resolution Sm 3d spectra of calcined and sintered CSDCs as shown in Figures S5(a) and S5(b) show the same characteristic, which consists of Sm 3d 5/2 (∼1082 eV) and Sm 3d 3/2 (∼1109 eV). e major peaks at ∼1082 eV and ∼1109 eV result from ionization of Sm 3+ . e shoulder at ∼1080 eV and ∼1102 eV results from the strong charge transfer of unpaired electrons in 4f of Sm 3+ [54], or Sm 3+ segregation [79], or the result of electrostatic interactions between a core level and a partially filled 4f level [80]. In our work, the shoulders of calcined and sintered CSDCs are possibly metastable SmO 8 -type complex or Sm 2 O 3 nanodomains. e C1s core level spectra at ∼288-290 eV as shown in Figures S5(c) and S5(d) represent the presence of carbonate at the surface [58] and the C1s peaks at ∼284-286 eV are ascribed to carbon [80]. All calcined and sintered CSDCs consist of carbonate species at ∼290 eV, which is due to the adsorption of CO x in ambient air or CO x gases from complexing agents during combustion, in addition to cerium oxide since Cu dope CeO 2 also has been used in CO catalyst applications [69,81,82]. Furthermore, all of them have the C1s peaks at ∼286 eV, which are assigned to carbon and are related to carbon trapped from complexing agents on nanoparticle surfaces. e signal of Cu2p core level is fuzzy for both calcined and sintered CSDC powders, as shown in Figure 11(a) and 11(b). Curran et al. [83] reported that Cu = 0.01 could not be observed due to the detection limit of XPS. However, Cu species can only be detected in calcined CSDCs but not in sintered CSDCs due to Cu release at sintering temperature as well as Cu evaporation induced by the XPS laser. e Cu2p core level spectra of all calcined CSDCs are shown in Figure 11(c). e fitted peaks of Cu2p 3/2 and Cu2p 1/2 are at around ∼933 eV and ∼955 eV, respectively. Moreover, the Cu (II) satellite peaks at ∼940 to ∼945 eV are revealed. e Cu2p 3/2 peak at ∼933 eV is ascribed to Cu 2+ while the one at ∼930 eV is characteristic of Cu + and/or Cu 0 [84]. Note that the presence of Cu 2+ is also confirmed by the Raman result. However, the presence of Cu + and/or Cu 0 may arise from the reduction of Cu 2+ , similar to the transformation of Ce 4+ to Ce 3+ under the testing conditions.

SEM Images.
e cross-sectional morphology of pellet samples after sintering was characterized using SEM as shown in Figure 12. After sintering, the pores in the figure are (1) pores from intergranular cracking and (2) pores in the bulk which are the combination of (i) interparticle voids and (ii) voids left behind by the evaporation of the volatile phase (copper segregation) [85]. Pore reduction requires the good redistribution of liquid copper segregation along grain boundaries, which starts at ∼750°C for 1%mol Cu-20SDC [26]. e movement and merging of liquid Cu 2+ promote particle mobility and particle rearrangement [25], which results in grain coarsening, especially when the particles are in contact, allowing the atoms at the surface of particles to diffuse and bridge, resulting in densification. However, the merging of the liquid phase in the voids of interparticle can also generate large pores when the liquid phase evaporates during sintering at 1100°C. erefore, copper segregation could either increase or decrease densification for a certain sintering temperature. Despite having the highest amount of copper segregation in calcined CSDC/DI (based on Raman results), the CSDC/DI pellet is the least dense sample after sintering ( Figure S5(b)). In the case of CSDC/DI, it may be   Figure 11: High-resolution Cu 2p XPS spectra of (a) calcined CSDCs raw data, (b) sintered CSDCs raw data, and (c) fitted calcined CSDCs.
because the portion of copper segregation in calcined CSDC/ DI could evaporate earlier than copper in calcined CSDC/ TA and CSDC/CA due to less Cu-Sm-Ce solid solution formation, resulting in CSDC/DI particles that cannot be properly rearranged. Furthermore, the amount of hydroxide phase, e.g., cerium hydroxide (based on Ce 3+ in XPS results) in calcined CSDC/DI is higher than the others, which may prevent grain growth during sintering, leading to porosity development. e CSDC/CA pellet is the densest, yet it has the least amount of copper segregation in the calcined sample but the highest one in the sintered sample, which could be related to the diffusion of copper from solid solution when the temperature is raised above the calcination temperature. As a result, there is a large difference in the amount of copper segregation present in the calcined CSDC/ CA as compared to that present in the sintered one ( Figure S5(b)). From the Archimedes test, the relative densities (r.d.) of CSDC pellets are calculated from equation (1), which are 82.8 ± 2.4%, 95.5 ± 1.8%, and 97.8 ± 0.9% for CSDC/DI, CSDC/TA, and CSDC/CA, respectively. Both r.d. and grain size can contribute to the change in grain and grain boundary conductivities, which was reported by Pérez-Coll et al. [5]. ey reported that grain conductivity mainly depends on the r.d., while grain boundary conductivity was influenced by both r.d. and grain size. Generally, both conductivities of grain and grain boundary increase rapidly with increasing r.d. when the r.d. is lower than 91.9%. However, grain and grain boundary conductivities slightly arise when the r.d. is above 91.9% and above 93.9%, respectively, In addition, both grain and grain boundary conductivity slightly increases for grain sizes above 1.27 μm (93.9% r.d.). erefore, grain and grain boundary conductivities of the sample are impacted by other factors when the r.d. exceeds 93.9%. Herein, the CSDC/TA and CSDC/CA samples are higher than 94% r.d., so the relationship of CSDCs conductivity with segregation and V •• O is being investigated. Figure 13(a) shows an Arrhenius plot of ln (σT) vs. 1000/T for grain conductivity. e grain conductivity of CSDC/CA is slightly higher than that of CSDC/TA and CSDC/DI at 350, 400, and 500°C because it has a greater relative oxygen vacancy concentration than the others. However, the grain conductivity of CSDC/CA drops slightly lower than that of CSDC/TA at a temperature lower than 350°C because the V •• O could be trapped by cation doped defects (Sm Ce ′ , Ce Ce ′ , Cu ″ ′ Ce , and Cu Ce ″ ) and be blocked by the metastable SmO 8 complex and Sm 2 O 3 nanodomains in the grain. As previously shown in Raman results, the relative amount of metastable SmO 8 complex is the highest in CSDC/CA after sintering, whereas the CSDC/DI has the lowest gain conductivity because it does not only have the lowest oxygen concentration and the lowest r.d. but also has high Sm 2 O 3 nanodomains ( Figure S4(a)).

Total Grain Boundary Conductivity.
Copper segregation helps increase r.d., but it can also reduce grain boundary conductivities of samples. However, copper segregation could increase the total grain boundary conductivity due to the scavenging of SiO 2 contamination at grain boundary regions by copper [11]. e total grain boundary conductivities of all CSDCs (Figure 13(b)) are not significantly different above 300°C. At 250°C and 300°C, CSDC/DI has lower total grain boundary conductivity than CSDC/TA and CSDC/CA because CSDC/DI has the lowest r.d. and high content of Sm 2 O 3 nanodomains despite having the lowest CuO segregation.

Specific Grain Boundary
Conductivity. Specific grain boundary conductivity (σ sp gb ) is commonly utilized to characterize the grain boundary conductivity behavior of polycrystalline materials based on the brick-layer model. e C g /C gb normalization yields a specific grain boundary conductivity that is less dependent on grain size [32]. It can be calculated as follows [14,86]: where L is sample thickness, A is an active area, R gb is grain boundary resistance, C g is grain capacitance, C gb is grain boundary capacitance, and σ total gb is total grain boundary conductivity. Advances in Materials Science and Engineering e σ sp gb of GDC without sintering aids is slightly affected by grain size and r.d. if the r.d. is greater than 75% because it is only affected by geometrical effect [5]. e geometrical effects are explained by the formula: C gb = ε 0 ε gb (A/L gb ) [86], where C gb increases with a shorter effective pathway and larger effective cross section area per grain and with decreasing σ sp gb . Since the C gb is attributed to SiO 2 contamination, CuO and Sm 2 O 3 segregation, and depletion regions, the σ sp gb decreases as segregations and depletion regions increase. e σ sp gb , as presented in Figure 13(c), are estimated at temperatures below 400°C due to the existing C g and C gb . Herein, the grain boundary capacitances are in the range of 9-30 nF as shown in Table 2. e C gb values are arranged in the following order: CSDC/DI < CSDC/TA < CSDC/CA. e high C gb of CSDC/CA could be an indication of segregations or a wider depletion region in the samples. Considering the segregations from Raman results after sintering ( Figure S4) Figure 13: Arrhenius plot of (a) grain conductivity, (b) total grain boundary conductivity, (c) specific grain boundary, and (d) total conductivity. 18 Advances in Materials Science and Engineering the grain boundaries.
e black color on the surface of CSDC/CA pellet and in the bulk can be a CuO precipitate. According to Raman results ( Figure 4S), sintered CSDC/DI and CSDC/TA have more Sm 2 O 3 nanodomains than sintered CSDC/CA. However, the amount of CuO is higher than that of Sm 2 O 3 nanodomains, around tenfold. As the result, it is possible to conclude that CuO segregation mainly affects σ sp gb rather than Sm 2 O 3 nanodomains. Note that even though CSDC/CA has the highest σ sp gb , its total grain boundary conductivity is the same as CSDC/DI and CSDC/ TA. e difference between the σ tot gb and σ sp gb may be that the true area of grain boundary is not used for calculating σ tot gb .

Total
Conductivity. e total conductivities of CSDCs were measured at temperatures ranging from 250 to 800°C, as shown in Figure 13(d).
e total conductivity is the summation of grain and grain boundary conductivities, which are affected by the concentration of V •• O , CuO segregation, r.d., and Sm 2 O 3 nanodomains. In our work, grain conductivity mainly depends on V •• O , whereas grain boundary conductivity is influenced by CuO segregation and r.d. e activation energy (E a ) is correlated to the slope of the Arrhenius plot of total conductivity, as shown in Table 3. e E a can be extracted from two regimes: the high-temperature range (500-800°C) and the low-temperature range (250-400°C). Moreover, the E a is related to enthalpy of migration (ΔH m ) in the high-temperature range, whereas Ea corresponds to both ΔH m and enthalpy of association (ΔH a ) in the low-temperature range.
e ΔH m involves the vacancies jumping to neighboring sites, while the ΔH a corresponds to trapped vacancies or immobilized dopant ions at the low-temperature range [10]. e samples with high total conductivity may need a low E a value. Despite having a higher CuO segregation than CSDC/DI, the CSDC/CA and CSDC/TA have higher total conductivities than CSDC/DI due to higher V •• O and r.d. However, the segregation of CuO along grain boundaries slightly affects the total conductivity at temperatures above 400°C, which is caused by the increasing mobility of V •• O . At the high-temperatures range, the sum of effects contributed by the difference in V •• O , CuO segregation, and r.d. lead to the insignificant difference in the ΔH m . Furthermore, the ΔH a of CSDC/DI, CSDC/TA, and CSDC/CA are 0.16, 0.19, and 0.23 eV, respectively. e difference in ΔH a values is caused by the various amounts of segregated CuO along the grain boundaries, as well as the defect association between V •• O and the cation dopant [4]. However, the total conductivity of CSDC/CA is the highest because it has the highest grain conductivity. e total conductivities of CSDC/DI, CSDC/TA, and CSDC/CA are 0.0176, 0.0228, and 0.0271 S/cm 2 at 600°C, respectively. Table 4 shows a comparison of total conductivities at 600°C between this work and other works.

Conclusions
e CSDC samples were successfully synthesized via a solgel auto-combustion method with and without complexing agents. CA or TA was a complexing agent used for this study, which also served as fuel. e as-synthesized morphology of CSDC/DI shows both layered structure and small agglomerate particles, while those of CSDC/TA and CSDC/CA reveal dense-xerogel and porous xerogel, respectively. e porous structure observed in as-synthesized CSDC/CA is the result of bubble generation and expansion during the gasrelease stage, since the CA has a higher degree of crosslinking. Moreover, calcined CSDC/CA has higher extrinsic oxygen vacancies than calcined CSDC/TA, indicating greater solubility of Sm 3+ and Cu 2+ in ceria because CA can complex with metal ions better than TA. Such improved complexing results in well-dispersed metal ions. e total oxygen vacancies in sintered CSDCs are lower than those of calcined CSDCs due to the partial oxidation of Ce 3+ to Ce 4+ . e trend of relative total oxygen vacancy concentrations of sintered CSDCs is as follows: CSDC/CA > CSDC/ TA > CSDC/DI. e calcined CSDCs also follow the same pattern. From Raman analysis, the presence of copper segregation, Sm 2 O 3 nanodomains, and metastable SmO 8 complex are observed in calcined and sintered CSDCs. e segregation of Cu 2+ is the key factor rather than that of Sm 2 O 3 nanodomains and metastable SmO 8 complex in determining the grain boundary behavior because the   relative concentration of Cu 2+ is tenfold higher than those of other factors. Although calcined CSDC/DI has the highest amount of copper segregation, it has poor densification as some liquid copper especially in the interparticle space could evaporate before the CSDC/DI particles could properly rearrange. On the contrary, calcined CSDC/CA has the lowest amount of copper segregation yet it has the highest densification, which could be due to the diffusion of Cu ions from Cu-Sm-Ce solid solution during the particle arrangement stage. e relative densities of CSDC/DI, CSDC/ TA, and CSDC/CA are 82.8 ± 2.4%, 95.5 ± 1.8%, and 97.8 ± 0.9%, respectively. e highest CuO segregates are also observed in sintered CSDC/CA, implying the diffusion of Cu from Cu-Sm-Ce solid solution during the sintering process. Specific grain boundary conductivity, which is governed by the segregation of CuO along grain boundaries, can well describe the true behavior of grain boundary conductivity. e highest value of ΔH a of CSDC/CA at the low-temperature range, which coincides with their lowest specific grain boundary conductivity, indicates that grain boundaries are covered by CuO. It is worth noting that the CuO segregation has little effect on the total grain boundary conductivity of CSDC/CA at temperatures above 400°C. e overall conductivity of CSDC/TA is slightly lower than that of CSDC/CA. e total conductivities at 600°C are 0.0176, 0.0228, and 0.0271 S/cm for CSDC/DI, CSDC/TA, and CSDC/CA, respectively.

Data Availability
e raw data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
e authors declare that they have no conflicts of interest.