Decoupling the Chemical and Mechanical Strain Effect on Steering the CO2 Activation over CeO2-Based Oxides: An Experimental and DFT Approach.

Doped ceria-based metal oxides are widely used as supports and stand-alone catalysts in reactions where CO2 is involved. Thus, it is important to understand how to tailor their CO2 adsorption behavior. In this work, steering the CO2 activation behavior of Ce-La-Cu-O ternary oxide surfaces through the combined effect of chemical and mechanical strain was thoroughly examined using both experimental and ab initio modeling approaches. Doping with aliovalent metal cations (La3+ or La3+/Cu2+) and post-synthetic ball milling were considered as the origin of the chemical and mechanical strain of CeO2, respectively. Experimentally, microwave-assisted reflux-prepared Ce-La-Cu-O ternary oxides were imposed into mechanical forces to tune the structure, redox ability, defects, and CO2 surface adsorption properties; the latter were used as key descriptors. The purpose was to decouple the combined effect of the chemical strain (εC) and mechanical strain (εM) on the modification of the Ce-La-Cu-O surface reactivity toward CO2 activation. During the ab initio calculations, the stability (energy of formation, EOvf) of different configurations of oxygen vacant sites (Ov) was assessed under biaxial tensile strain (ε > 0) and compressive strain (ε < 0), whereas the CO2-philicity of the surface was assessed at different levels of the imposed mechanical strain. The EOvf values were found to decrease with increasing tensile strain. The Ce-La-Cu-O(111) surface exhibited the lowest EOvf values for the single subsurface sites, implying that Ov may occur spontaneously upon Cu addition. The mobility of the surface and bulk oxygen anions in the lattice contributing to the Ov population was measured using 16O/18O transient isothermal isotopic exchange experiments; the maximum in the dynamic rate of 16O18O formation, Rmax(16O18O), was 13.1 and 8.5 μmol g-1 s-1 for pristine (chemically strained) and dry ball-milled (chemically and mechanically strained) oxides, respectively. The CO2 activation pathway (redox vs associative) was experimentally probed using in situ diffuse reflectance infrared Fourier transform spectroscopy. It was demonstrated that the mechanical strain increased up to 6 times the CO2 adsorption sites, though reducing their thermal stability. This result supports the mechanical actuation of the "carbonate"-bound species; the latter was in agreement with the density functional theory (DFT)-calculated C-O bond lengths and O-C-O angles. Ab initio studies shed light on the CO2 adsorption energy (Eads), suggesting a covalent bonding which is enhanced in the presence of doping and under tensile strain. Bader charge analysis probed the adsorbate/surface charge distribution and illustrated that CO2 interacts with the dual sites (acidic and basic ones) on the surface, leading to the formation of bidentate carbonate species. Density of states (DOS) studies revealed a significant Eg drop in the presence of double Ov and compressive strain, a finding with design implications in covalent type of interactions. To bridge this study with industrially important catalytic applications, Ni-supported catalysts were prepared using pristine and ball-milled oxides and evaluated for the dry reforming of methane reaction. Ball milling was found to induce modification of the metal-support interface and Ni catalyst reducibility, thus leading to an increase in the CH4 and CO2 conversions. This study opens new possibilities to manipulate the CO2 activation for a portfolio of heterogeneous reactions.

Introduction

In heterogeneous catalysis, strained surfaces can be found in both metal-supported catalysts and multi-elemental metal oxides due to a lattice constant mismatch between metal–support and host–guest interactions, respectively.[1] Such lattice strain enhances the chemisorption properties of the catalytic surface significantly,[2] either for the adsorbate or for the intermediate species formed under reaction conditions. As a strain normally manipulates the surface ability to form bonds, there is a great possibility in using a strain as a catalyst reactivity modifier/descriptor.[2]

Ball milling is a mature mechanical activation technique that can lead to the fabrication of nanocatalysts with anomalous properties compared to their bulk counterparts. Through mechanical activation, solid-state chemical reactions can be initiated/accelerated while causing various transformations and reactions, such as grain boundary disordering, amorphization, defect generation/migration, polymorphic transformations, ion coordination sphere change, and reduction in particle size.[3] The main function that takes place during ball milling is that the material’s potential or its stored mechanical energy is enhanced in the presence of ball milling forces.[4] This is translated into defects (point, line, and volume ones), surface and interface formation, strain and structural disorder, and changes in electronic states. Some of the parameters in the ball milling process that can alter the impact of mechanical activation in terms of the final catalyst properties (e.g., surface area, active metal dispersion, and binding strength of the intermediates) are the ball size, number of balls used, ball-to-powder ratio, rotational speed, milling time, and milling atmosphere.[5,6] It is well known that the ceria morphology and thus its properties can be tuned through synthesis. A variety of synthetic methods have been used spanning from mild hydrothermal to form ceria nanotubes[7] and nanorods with different types and distributions of oxygen vacancies[8] to template-free microwave-assisted hydrothermal synthesis and urea homogeneous precipitation; the latter ceria materials were used for the preparation of Pt/CeO2 catalysts for the WGS reaction,[9] demonstrating the synthesis impact on the electronic/catalytic properties.

In the context of sustainable development, there is an increasing interest in the utilization of the captured CO2 from a flue gas to form valuable chemicals and products. In this sense, two of the catalytic reactions of most interest are dry reforming of methane (DRM) and CO2 hydrogenation. The DRM is very relevant in this context as it simultaneously tackles the abatement of two greenhouse gases (CH4 and CO2) while leading to the production of hydrogen (energy carrier). The CO2 hydrogenation leads to the valorization of CO2 into different value-added products. However, it is well known that the inert property and stability of the CO2 molecule pose a challenge in both aspects: thermodynamics and kinetics.[10] Upon CO2 adsorption on metal and metal oxide surfaces, the molecule’s stability is reduced. The adsorption configuration is highly dependent on the surface Lewis acidity/basicity. In an ideal scenario, a bent configuration is favored by charge transfer (CT) from the surface (Lewis base, electron reservoir) to the CO2 molecule. There are reports showing that both CO2 linear and bent configurations on the ceria surface are favored.[11,12] It has also been reported that reduced ceria surfaces are favorable for carbon dioxide reduction reactions due to the involvement of polarons (Ce3+–Ov pairs) in the CO2 molecular adsorption at oxide surfaces, thus enhancing CT that is the initiator of such chemical reactions. It is therefore important to obtain insights into the surface chemical reactivity and its CO2-philicity, as well as how this can be steered. Acidic surfaces favor the formation of linear-type CO2 adsorption configuration, whereas basic surfaces favor the CO2– formation (bent species and reactive ones).[13]

From the aspect of catalyst design for COactivation, some criteria need to be considered, such as (a) the use of metal oxide supports with high basicity (e.g., MgO and La2O3) can enhance the dissociative adsorption of CO2, which inhibits carbon formation by creating a higher number of oxygen atoms around the catalyst-active metal surface.[14] The improvement of CO2 dissociation on catalysts by increasing the surface basicity can similarly deactivate the catalyst.[14] It has been verified lately that excessive surface basicity/acidity can cause deactivation due to carbon formation in reactions such as DRM,[15] where the importance of moderate acidity and basicity along with the homogeneous dispersion that ideally defines catalytic conversion in the DRM reaction and the long-term stability of supported metal catalysts was demonstrated. (b) Oxygen vacancies (Ov) can be classified as intrinsic, naturally existing in the material (e.g., due to the presence of Ce4+/Ce3+), or extrinsic, induced by the doping of CeO2 lattice with aliovalent metal cations for better ionic conductivity.[16,17] Among the doped ceria catalysts, the Ce–Cu–O system exhibits the (i) high redox properties and the ability to switch between Ce3+/Ce4+ and Cu2+/Cu1+, (ii) increasing population of oxygen vacant sites (Ov), and (iii) increase of labile surface and bulk oxygen species compared to the single-phase oxide.[18] Therefore, it can be stated that Ov has a critical role in the CO2 dissociation on ceria surfaces and thus is expected to have a leading role in catalytic reactions where CO2 is a reactant or a co-reactant.[19] It is also well established that strain can modify the defect (e.g., Ov) structure and electronic properties.[20] While the tensile strain leads to surface vacancies having polarons as the next neighbor (NN), the compressive strain is associated with subsurface vacancies having polarons as NN, whereas dimer vacancies are also favored over compressive strain.

Mechanical forces are used either to prepare or to actuate ceria-based catalysts and devices. For instance, in the preparation of ceria-based catalysts, the latter used for environmental and energy applications has been reported in many review articles using mechanochemical methods.[21] Demonstration of the superiority of Pd/CeO2 ball-milled catalysts for the methane oxidation reaction at low temperature, above 95% conversion, over the traditional wet impregnation catalysts has been discussed by Trovarelli’s research group even at conditions which are challenging for the particular reaction.[22,23] Furthermore, there is extensive literature on the impact of different types of external stimuli on thin films of ceria; for example, the electrochemomechanical effect has been reported by Lubomirsky and his colleagues[24,25] to produce stress that can potentially deteriorate them.[22,23] Furthermore, strain engineering has been reported to increase the CT and electronic conductivity.[26] However, the effect of applying compressive or tensile stress on doped ternary ceria (111) surfaces has not being comprehensively investigated in the literature yet in the context of steering of its catalytic chemistry and adsorption behavior. The oxide surfaces (Ce–La–Cu–O) studied in the present work exhibit a versatile catalytic functionality that spans from reforming[27] to oxidation chemistry,[18] given their noble-metal-free nature is worthy of more attention. Finding a way to engineer the vacancy population and structure, and, hence, the CO2–surface interaction, can be an additional tool toward catalyst design for CO2 activation, thus contributing to a more rational design of catalysts for reactions such as DRM and CO2 hydrogenation.

In the present work, microwave-prepared Ce–La–Cu–O ternary metal oxides were subjected into mechanochemical activation under dry and wet ball milling targeting their intrinsic property modification, namely, structure, crystallite size, specific surface area, redox properties, CO2 activation energy, and pathway. The main emphasis was given on exploring how chemical strain, εC (doping effect), and mechanical strain, εM (compressive or tensile, originated by ball milling), can impact the oxygen vacancy formation and population [Ov] and their chemical reactivity toward CO2 activation. Experimental and density functional theory (DFT) studies were performed in an effort to deconvolute the εC and εM impact. A versatile toolbox including X-ray powder diffraction, Raman spectroscopy, transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), 18O2-transient isothermal isotopic exchange (TIIE), in situ CO2 diffuse reflectance infrared Fourier transform spectroscopy (DRIFT) spectroscopy, H2-temperature-programmed reduction (H2-TPR), and CO2-temperature-programmed desorption (CO2-TPD), along with intensive and systematic ab initio calculations was employed to fully characterize and analyze the performance of the solids. The effect of supports during ball milling on the performance of Ce–La–Cu–O supported Ni catalysts toward the DRM reaction was evaluated to provide evidence for the value that mechanochemistry can bring into catalysis.

Materials and Methods

Preparation of Ternary Metal Oxides

A microwave-accelerated reaction system (MARS-6) was used to synthesize the catalysts through microwave-assisted reflux synthesis. The microwave system had a power output of 0–1800 W ±5% (IEC 705 Method-1988).[18] Materials were prepared using precursor salts Ce(NO3)3·6H2O (Aldrich 99.95%), La(NO3)3·6H2O (Aldrich 99.95%), Sm(NO3)3·6H2O (Aldrich 99.95%), and Cu(NO3)2·3H2O (Aldrich 99.95%) and dissolved in distilled water. The prepared mixed metal oxides had 10 at. % Cu content, while the Ce/M (M: La and Sm) ratio was maintained at unity. The total molar ratio was maintained at 0.03 mol in all cases. The complexing agent ethylene glycol (EG) was added to the solution with a ratio of EG to water retained at 2; that is, for each 100 mL of EG, 50 mL of distilled water was added.

The final solution was added to an open 1000 mL round-bottom flask in an open vessel mode, equipped with an oval magnetic stirring bar (length 20 mm and diameter 6 mm) made with polytetrafluoroethylene coating. A reflux system was attached to the open round flask of the MW system, allowing water to pass through the reflux for condensation. Parameters, such as stirring, reaction temperature, MW power, and heating/cooling ramp, were all adopted from a previous study.[1] The temperature of the reaction was monitored by built-in fiber optic probes. The solution was heated in the microwave reactor at two stages: (i) 130 °C for 2 h and (ii) 170 °C for 1 h at 800 W of magnetron power. Following microwave heating, all synthesized materials were calcined at 500 °C for 6 h under ambient conditions to form the mixed metal oxide catalyst. In what follows, the wet ball-milled and dry ball-milled samples were coded as wet ball milling (WBM) and dry ball milling (DBM), respectively.

Ternary Oxides Post-Synthesis Mechanochemical Treatment (Ball Milling)

The synthesized materials were ball-milled using a planetary ball mill (Planetary Mill PULVERISETTE 5, Fritsch) under ambient conditions. A relevant amount of catalysts was mixed with distilled water, maintaining a weight ratio of 12:1. The same catalysts were ball-milled in a dry medium to track any structural changes. The catalysts were milled at 250 rpm at different milling times in a zirconia jar with a mass ratio of balls to solid powder equal to 100. The balls were made of zirconia and had a mass of around ∼3 g each. The ball-milled catalysts under wet condition were dried at 150 °C for 3 h. Wet milling took place for 4 and 10 h, whereas DBM took place from 0 to 10 h at an interval of 2 h.

Supported Ni Catalyst Preparation

Pristine Ce–La–Cu–O oxides along with the ones following DBM (Ce–La–Cu–O) and WBM (Ce–La–Cu–O) were used as supports for the deposition of Ni metal phase (supported Ni catalysts) following the wet impregnation method (5 wt % loading) as described in our previous work.[27]

Characterization of Ternary Metal Oxides

Powder X-ray diffraction (XRD), Raman spectroscopy, electron paramagnetic resonance (EPR), scanning electron microscopy, TEM, H2-TPR, CO2-TPD, and XPS were employed to study the structural, textural, redox, and CO2 adsorption properties in the pristine (following calcination) and ball-milled (wet and dry) ternary metal oxides. The instrumentation and the experimental protocol are provided in the Supporting Information (see Section S1.3).

18O/16O TIIE

The surface and bulk oxygen mobility/diffusion in the mixed metal oxides, particularly Ce–La–10Cu–O, was investigated using 18O2-TIIE experiments. The step-gas isotopic switch, 2 mol % 16O2/2 mol % Kr/Ar/He (T, 30 min) → 2 mol % 18O2/Ar/He (T, t), was conducted over a 20 mg sample with a total volume flow rate of 50 N mL/min. The sample was first pre-calcined under 20 vol % 16O2/He gas flow at 800 °C for 2 h, then Ar was passed over the sample for 10 min, and the temperature was then decreased to 350 °C in Ar gas flow, followed by the TIIE step-gas switch. During the TIIE experiment, the dynamic evolution of the rates of oxygen exchange between the gas-phase oxygen, lattice oxygen, and oxygen vacant sites was recorded.

The transient response curves of the three oxygen isotopic gases and of Kr tracer (inert gas) were continuously recorded by an on-line mass spectrometer (Balzers, Omnistar, 1–300 amu) for the mass numbers (m/z) 32, 34, 36, and 84, corresponding to the 16O2, 16O18O, 18O2, and Kr signals, respectively. Following the surface 16O/18O exchange, the lattice oxygen (16O) in the solid diffuses from the bulk to the surface, and the 18O lattice oxygen diffuses from the surface to the bulk.

For the continuous stirred-tank reactor (CSTR) behavior considered in this study,[22] the transient rates (μmol/g s) of 16O18O(g) (eq ) and 16O2(g) (eq ) formation were calculated

In eqs and 2, FT is the total molar flow rate (mol/s) of the feed gas stream, y, y, and yKr are the mole fractions of 16O18O, 16O2, and Kr at the outlet of the CSTR microreactor, respectively, NT is the total number of moles in the CSTR reactor, and W is the sample’s mass (20 mg).[22] The total rate of 16O exchange with 18O and R(16O) (mol 16O g–1 s–1) and the transient evolution of the amount of 16O and N(16O) (mol 16O g–1) were estimated based on 16O-material balance eqs and 4where Z is the dimensionless response of gaseous species i, y is the mole fraction of gaseous species i, y(16O2) (t = 0) is the mole fraction before the step-gas switch (2 vol %), and yf(16O18O) is the mole fraction of 16O18O(g) present in the 2 mol % 18O2/Ar/He feed gas mixture used.[28]

The ag18(t) descriptor function was used to investigate both the surface 16O/18O exchange and the oxygen diffusion in the bulk via eqs and 6

In Situ Diffuse Reflectance Infrared Fourier Transform Spectroscopy

Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) studies were performed using a PerkinElmer Frontier Fourier transform infrared spectroscopy spectrometer (256 scans per spectrum, resolution 4 cm–1, and scan speed 2 cm/s) equipped with a high-temperature/high-pressure-controllable DRIFTS cell (Harrick, Praying Mantis) to investigate (i) the intermediate-adsorbed species formed under CO2/Ar gas treatment and (ii) the stability of the adsorbed intermediate species under Ar inert gas. DRIFTS spectra were recorded in 5 vol % CO2/Ar gas flow at 350 °C and after 30 min in Ar gas flow at 350 °C for various times in Ar flow. The catalyst sample (∼80 mg) in a very fine powder form was placed firmly into the ceramic cup of the DRIFTS cell and the temperature was increased to 500 °C in Ar gas flow and kept for 30 min. Then, the sample was cooled to 350 °C in Ar gas flow, and the spectrum of the solid was recorded at 350 °C. The latter spectrum was subtracted from the spectrum of the solid recorded in the CO2/Ar gas mixture or Ar gas flow at 350 °C. Deconvolution and curve fitting procedures of DRIFTS spectra were performed considering the Gaussian peaks. DRIFTS spectra when necessary were smoothed to remove the high-frequency noise and further analyzed using the software Spectrum10 for Windows.

Ab Initio DFT Calculations

Energy of Oxygen Vacancy Formation (EOf)

The first-principles spin-polarized calculations, based on the DFT,[29,30] were performed after employing the Vienna Ab Initio Simulation Package (VASP).[31−33] The electronic exchange was used under the generalized gradient approximation (GGA) of Perdew–Burke–Ernzerhof.[34] To include the strong correlation effects of 4f electrons, GGA with the Hubbard U correction method[35,36] was used with a value of U = 5.0 eV.[37−39] The DFT + U method accounts for the O 2p states and has been previously followed in DFT studies as a common practice to accurately describe the oxygen electronic behavior rising from the doping effect.[40,41]

The plane-wave cut-off energy was set to 400 eV[42] in all the calculations, and the projector-augmented wave pseudopotentials[43] were used to describe the core electrons. The (111) slab was built from the optimized CeO2 unit cell with a lattice parameter equal to 5.46 Å, in agreement with the literature.[44] The slab consisted of nine (9) layers terminating with oxygen with a vacuum of 15 Å in the vertical direction to avoid interaction between neighboring images due to periodic boundary conditions. A γ-centered k-point of 3 × 3 × 1 in the full Brillouin zone, with an energy convergence criterion of 10–6 eV, was employed. A (2 × 2) CeO2(111) surface was constructed to create a vacancy in pristine CeO2 and the doped CeO2 systems (Ce–La–O and Ce–La–Cu–O). The oxygen vacancies were created at different locations by removing one or two of the denoted oxygen atoms at a time. The three bottom layers were fixed in their bulk positions to avoid the change in the crystal structure under compressive (−5 to −1%) or tensile (1 to 5%) strain.

Two-Dimensional Planar Lattice Strain

Biaxial strain was isotropically applied by stretching or shrinking the simulation cell in the x, y directions while relaxing the cell along the z direction. The structure optimizations were performed until the energy-convergence criterion was satisfied, ca. 1 × 10–6 eV. The atoms in the top six layers were allowed to relax in all the three directions (x, y, and z). A biaxial strain of −5 to 5% was applied to investigate the effect of biaxial strain on the oxygen vacancy formation energy (E) in the case of doped/co-doped ceria surface layer of the slab. The E value was calculated using eq where is the total energy of the relaxed pure/doped surface with oxygen vacancies after applying the biaxial strain or without the biaxial strain, is the total energy of the pure/doped slab with the same applied strain of but without the oxygen vacancy, is the energy of , and n is the total number of oxygen vacancies created in the system. Negative (<0) and positive (>0) values of correspond, respectively, to spontaneous and nonspontaneous Ov formation.

CO2 Adsorption Energy

In order to understand how the doping (chemical strain) and the imposed biaxial mechanical strain affect the CO2 adsorption, pure CeO2 and doped/co-doped ceria (Ce–La–O and Ce–La–Cu–O) surfaces were used to calculate the adsorption energy (Eads) of CO2. Two scenarios were considered: (a) in the presence of oxygen vacancies (Ov) at zero applied biaxial mechanical strain and (b) in the absence of oxygen vacancies (Ov) but subjecting the slabs to biaxial strain. A linear configuration of a single CO2 molecule was adopted on the constructed slabs, and eq was implemented to extract the Eads values. is the total energy of the slab with the adsorbed CO2 molecule, is the total energy of the slab without the adsorbate (CO2), and is the total energy of the isolated CO2 molecule. As linear CO2 is not activated, the hypothesis was to study how the chemical and mechanical strain enhances the CT, which will be evident by the bending and activation of CO2 after relaxing the full configuration. A negative value of the adsorption energy means that the molecule is exothermically adsorbed (Eads < 0), while a positive value indicates that the molecule is endothermically adsorbed (Eads > 0); the more stable adsorption state is implied by the more negative value. The coordinates of the unit cell were obtained from materials project,[45] the slabs were viewed through VESTA,[46] and the CeO2(111) surface was cleaved using VESTA. All the above calculations were run multiple times to estimate the possible errors introduced from the VASP calculations; the error was <1%.

Results and Discussion

Microstructural Studies

Figure A compares the powder XRD patterns of the Ce–La–10Cu–O mixed metal oxide composition before (0 h) and after DBM for different milling times (ca. 2–10 h) with an interval of 2 h. A detailed presentation of the (111) diffraction peak position is presented in Figure B. For the reference oxides (CeO2 and Ce–La–O), the corresponding XRD patterns are provided in Figure S1. For the case of Ce–La–O due to the La content (50%) and anticipated heavy ceria doping, the La2Ce2O7 pyrochlore structure of fluorite type can be formed. Even though the present XRD pattern does not allow for its determination, more discussion is provided in the Raman section. The main ceria fluorite cubic lattice reflections of (111), (200), (220), (311), (222), and (400) at diffraction angles of 28.5, 33, 47.5, 56.3, 59.1, and 69.3° can be noticed in all the cases of Ce–La–Cu–O DBM oxide (Figure A,B). It is also important that for the pristine Ce–La–10Cu–O (absence of ball milling), a small peak corresponding to the presence of CuO can be observed at 38° 2θ (noted with *), which is vanishing with ball milling (Figure A). Figure C,D presents the Ce–O and Cu–O bond lengths, respectively, as those have been calculated through modeling approach; their trends are discussed shortly. Figure E presents the powder XRD patterns of the Ce–La–10Cu–O ternary oxide following 4 h of WBM. The ball milling process under wet atmosphere (water slurry) resulted in a successive appearance of a new crystalline phase, as indicated by the peaks at 26, 29, 31, 45, and 60 2θ, corresponding to the (100), (002), (101), (110), (112), and (201) diffraction planes of La2O3.

Figure 1

(A) XRD patterns of Ce–La–10Cu–O ternary oxides following DBM treatment for 0, 2, 4, 6, and 10 h; (*) denotes the CuO(111) phase impurity; (B) zoom in of the XRD patterns of (A) in the 26–30° 2θ region corresponding to the (111) diffraction plane; (C) Ce–O bond length (Å) in CeO2, Ce–La–O, and Ce–La–10Cu–O oxides calculated through the ab initio studies; (D) Cu–O bond length (Å) in Ce–Cu–O and Ce–La–10Cu–O oxides calculated through the ab initio studies; (E) XRD patterns of the Ce–La–10Cu–O ternary oxides following WBM treatment for 0, 4, and 10 h; (*) denotes the CuO(111) phase impurity, and (#) denotes the La2O3 phase impurity; and (F) EPR spectra of the Ce–La–10Cu–O ternary oxide following DBM and WBM treatment for 4 h.

From the detailed analysis of the (111) reflection shown in Figure B, a peak shift to higher 2θ angles can be observed for milling times up to 4 h due to a compressive strain developed along the Ce–O bond axis in the lattice and possible move of atoms to interstitial sites under the mechanochemical forces.[47] The compression of the Ce–O bonds was also supported through the ab initio calculations (see Figure C). It has to be mentioned that an opposite trend was noticed along the Cu–O bond (elongation, see Figure D). As the ball milling duration was increased beyond 4 h, either emerging of new peaks corresponding to impurity phases (La(OH)3 and La2O3) is noticed (WBM) or vanishing of the CuO heterophase peak (DBM) (Figure B,E). An increase of milling time is expected to lead to an increased misorientation between neighboring grains causing the formation of high-angle boundaries. As the high-energy milling continues, the crystallite size reaches a critical value. Further energy input to the critical size crystals leads to additional crystal deformation, energy accumulation at the surface, followed by some extent of amorphization.[2]

Figure F presents the EPR spectra obtained over the DBM and WBM samples. The sharp signal at ∼3400 mT is usually associated with [Ce3+–O––Ce4+] (S = 1/2)[48,49] species on the surface. Since the presence of Ce3+ is usually linked to the presence of Ov, it can be stated that Ce3+ can also be linked with high surface activity. The signal at 2600–3300 mT is linked to Cu2+ (S = 1/2 and I = 3/2) ions. The shape of the EPR line shows the Cu–Cu dipolar interactions, that is, in a distance of 10 ± 3 Å from each other.[50]

Based on the powder XRD studies, two types of crystallite size were calculated, and the obtained results are listed in Table . Using the Scherrer formula (broadening of the XRD peak originates from the crystallite size only), the crystallite size, DS (nm), was calculated after neglecting the strain. In addition, accepting strain-induced broadening of the XRD peaks, due to crystal imperfections and distortion, the Williamson–Hall (W–H) method was used.[51] In the W–H plot method, DW–H (nm) does not depend on 1/cos(θ), but it does change with tan(θ). This difference is important when small crystalline size and strain co-exist. The equation used for the W–H calculations represents the uniform deformation model (UDM), where a uniform strain across crystallographic orientations is assumed. It is observed that the UDM crystallite size systematically decreases as the milling time increases up to 8 h, and then saturation occurs that leads to an increase of the size after 10 h of milling. This trend can also be related with the equilibrium state of ball milling, where the particle size reduction is escorted by particle size enlargement. In such a case, smaller particles are agglomerated. This is quite consistent with the high-resolution TEM (HRTEM) analysis and the observed refinement of the solid solution nanodomain observed following DBM, as it will be discussed later.

Table 1

Textural Properties of the CeO2, Ce–La–O, and Ce–La–10Cu–O Solids Studieda

solid	BET (m2/g)	DS (nm)	DW–H (nm)	pore size (nm)	lattice parameter (Å)	lattice strain, ε
CeO2	36.6	7.8		8.7–10.1	5.459	0.0012
Ce–La–O	5.8	4.6		43–47.4	5.500	0.0038
Ce–La–10Cu–O-pristine	4.9	4.5		18.8–19.3	5.488	–0.0030
Ce–La–10Cu–O-WBM-4 h	14.5	4.5		11.4–12.2	5.440
Ce–La–10Cu–O-DBM-2 h		10.5	15.93		5.471	0.0047
Ce–La–10Cu–O-DBM-4 h	8.1	9.7	12.77	13.8–14.4	5.459	0.0025
Ce–La–10Cu–O-DBM-6 h		11.4	13.58		5.470	0.0026
Ce–La–10Cu–O-DBM-8 h		7.9	9.82		5.444	0.0019
Ce–La–10Cu–O-DBM-10 h	3.5	12.8	15.48	18.9–20.3	5.474	0.0044

DS was calculated based on the Scherrer formula. DW–H was calculated based on the W–H plots. Lattice strain was calculated based on the W–H plots.

It is also noticed that the Ce–La–10Cu–O lattice parameter is generally decreased in the case of WBM (5.488–5.440 Å) and DBM (5.488–5.454 Å), more likely indicating a greater extent of substitution of Ce4+ by La3+ and Cu2+ guest cations under the severe milling conditions. The observed change in the lattice parameter can be justified with a preferred uptake in the crystal lattice of Cu (decrease) or La (increase) with a concomitant change in the lattice parameter, depending on the applied mechanical forces. The energy input introduced through ball milling enhances the solubility of the two dopant elements (La and Cu) in the host ceria matrix,[3] but we need to keep in mind that the diffusion is highly dependent on the atomic size of the diffusing species, as well as of host matrix atoms.[2]

The W–H method was also used to study the lattice strain (ε) of the catalysts. The W–H plots are provided in Figure S2. It is observed that after doping ceria with La and La/Cu induces a different lattice strain, namely, from 0.0012 (pure ceria) to 0.0038 (tensile strain in Ce–La–O) and −0.0030 (compressive strain in Ce–La–Cu–O). Mechanical ball milling changed the overall lattice strain in the Ce–La–Cu–O oxide from compressive to tensile. The calculated W–H strain trends are in agreement with the sizes of the dopant elements (Cu2+, La3+ vs Ce4+, Ce3+). In addition, after increasing the time of DBM (0–10 h), this leads to a profile that includes an initial increase of the lattice strain (DBM 2 h), followed by decrease (4–8 h, DBM) and a final significant increase after 10 h of DBM. This trend/profile can be due to the high pressure and high temperature spots (“hot spots”) developed under ball milling, thus causing some relaxation in the lattice. The hot spots can also drive the process of phase transformations.[52] It is also known that during the milling process, a drop of milling effectiveness can occur. According to the ball milling fundamentals, during the milling process, three stages can take place, namely, the Rittinger stage (a), where particle interaction is minimal; the aggregation stage (b), where particle interaction leads to aggregation, and thus, the surface area produced does not correspond to the energy input; and the agglomeration stage (c), where the dispersion is reduced and then vanished.[53]

Raman spectroscopy studies were conducted to investigate the strain, phenomena such as phonon confinement and stoichiometry defects, based on the shifting and broadening of the spectral lines. In addition, they were conducted to probe the O sublattice (CeO8) distortions due to doping and mechanochemical treatment. The Raman spectra for Ce–La–10Cu–O, before (t = 0 h) and after DBM for different milling durations (t = 0–10 h), are shown in Figure A. The Raman spectra for the reference oxides (CeO2 and Ce–La–O) are provided in Figure S3. The F2g peak centered at 469 cm–1 (Figure A) is the characteristic band for the cubic fluorite lattice of ceria and reflects the vibrational mode of oxygen surrounding Ce4+ ions in the CeO8 coordination environment.[54] It is well documented that doping ceria, in the present study with La3+ and Cu2+ cations, results in an intensity attenuation and red shift of the F2g peak to lower values, namely, from 469 to 452 and 447 cm–1 for the case of Ce–La–O and Ce–La–10Cu–O, respectively (Figure S3A). This shift can be justified by the increased lattice strain due to the incorporation of the dopants, for example, due to La3+ addition (tensile), 2M2O3 + 3CeCe× + OO× → 4MCe′ + 2VO + 3CeO2, and/or Cu2+ addition (compressive), , that generated defects, such as oxygen vacancies and lattice distortion/expansion.[55] It has also been reported that in La-doped systems, the F2g shift is about 6 cm–1, whereas in heavily doped systems, the shift can reach up to 15 cm–1; such a case is found in the La2Ce2O7 pyrochlore type of structure.[56,57]

Figure 2

(A) Raman spectra obtained over the Ce–La–10Cu–O ternary oxides following DBM treatment for t = 0, 2, 4, 6, and 10 h; (B) Raman spectra of the Ce–La–10Cu–O ternary oxides following WBM treatment for t = 0, 4, and 10 h (*, ^ denote the CuO- and CeO2-based phases); (C) effect of ball milling conditions/atmosphere on the F2g position (Raman band), Ov/F2g ratio (Raman bands ratio), and lattice parameter, Å (based on the XRD); and (D) effect of dry milling time on the F2g position, Ov/F2g ratio, and lattice parameter (Å).

Doped ceria with metal cations can improve the oxygen mobility of the catalyst by lowering the barrier for oxygen migration and minimizing the activation energy for ceria reduction (Ce4+ → Ce3+),[58] as it will be discussed later (Section ). Furthermore, recent density of states (DoS) studies from our group[59] showed that doping ceria leads to the formation of energy states, which host the electrons left behind in the oxygen vacant site, thus facilitating the formation of Ov. For each Cu2+ ion introduced into the lattice, one oxygen vacancy is expected to be formed,[55] which is anticipated to cause tensile strain that leads to the elongation of M–O bond and therefore an increase in the number/density of mobile oxygen vacancies.[60,61] This has been demonstrated to apply in the case of Ce–La–Cu–O solid by elongated Cu–O and compressed Ce–O bonds in the tensile strain region (Figure C,D) according to the present DFT calculations (Table S1).

A closer look into Figure A (DBM) and Figure B (WBM) shows intense changes in the F2g band shape and size following the mechanochemical treatment. For both DBM and WBM, there is an optimum in the duration of the mechanochemical treatment, ca. 4 h, where either the F2g peak intensity is maximum (DBM) or there is no formation of phase impurity (WBM), in agreement with the XRD studies. Further increase of the mechanochemical treatment time either induces changes to the structure (WBM) or decreases the F2g band intensity (DBM). It can be noticed that 10 h treatment leads to almost vanishing of the F2g peak under DBM conditions, whereas under WBM, the peaks that correspond to the appearance of the hexagonal La(OH)3 phase are noticed; the latter is most likely due to the dehydroxylation of the oxide surface and a subsequent reaction of the OH entities with La3+ (La3+ + 3 OH– → La(OH)3) under milling forces.[53]

The bands in the region below <400 cm–1 (e.g., 250 cm–1) are usually assigned to oxygen defects in ceria due to Brillouin zone scattering.[61] The increase of the intensity of those bands up to a milling time of 4 h agrees with the above observations in agreement with the XRD interpretations. These bands are more evident under DBM, suggesting a larger amount of oxygen defects (Ov) present in DBM oxides compared to the WBM ones. A detailed analysis of the F2g band is presented in Figure C, showing the effect of lattice strain and phonon confinement on the band shape and oxygen stoichiometry.[62] It is worth noting that Raman band broadening implies the reduction in the size of the nanocrystallites, whereas a shift in Raman frequency (Δ cm–1) is associated with tensile or compressive strain when a red/blue shift, respectively, occurs. The combination of the phonon confinement and strain on the Ce–La–Cu–O nanocrystallites can make the interpretation hard. In this study, Ce–La–Cu–O oxides in the 4–12 nm size range were employed, making the phonon confinement effect significant in all the cases presented herein. However, the broadening and shift of Raman peaks are predominantly due to the strain and Ov defects originated from the milling treatment.

Figure C presents a drop in the lattice parameter in the presence of mechanical strain (ball milling, DBM/WBM), and this is in agreement with the Ce–O bond shrinkage estimated under tensile strain conditions (DBM was found to induce tensile strain, Table ). Only the cases of 4 h milling are presented as this was found to preserve to the highest degree of the fluorite structure. In addition, in Figure D, a general increase of FWHMF was observed with the milling time ranging from 0 to 10 h. This broadening of Raman peak full width at half-maximum (FWHM) coincides with the size reduction and strain development in the crystallites. At the same time, a shift in the peak position is observed with the milling duration.

Important from the catalysis perspective is the defect region (550–600 cm–1), which is associated with the LO mode (F1u symmetry). Even though this symmetry is Raman inactive, a relaxation of selection rules[63,64] makes the LO mode visible. The presence of defects causes a significant lowering in symmetry and thus relaxation in selection rules. It is anticipated that ball milling tunes the oxygen defects as it will be demonstrated through the ab initio calculations (Section ). Deconvolution of the Raman spectral defect region led to quantitative estimation of the Ov/F2g ratio for the pristine, DBM, and WBM Ce–La–10Cu–O oxides (Figure S4). This ratio is considered as a good descriptor of the Ov population in the oxide. In particular, the Ov/F2g intensity ratio was found to be 0.24 (0 h), 0.15 (2 h), 0.11 (4 h), 0.04 (6 h), 0.03 (8 h), and 0.38 (10 h) for different increasing DBM durations. In addition, Raman spectra deconvolution allowed us to spot the phase impurity (LaO8) as shown in Figures S3 and S4.

Figures S5 and 3 show the HRTEM (A), scanning TEM high-angle annular dark-field (STEM-HAADF) (B), red-green-blue (RGB) analysis (C), selected area (electron) diffraction (SAED) (D), and electron energy loss spectroscopy (EELS) (E) micrographs of the Ce–La–10Cu–O oxide before (Figure S5) and after DBM (t = 4 h, Figure ). The DBM oxide (t = 4 h) was chosen for this analysis due to its structural features, which make it most promising from the catalysis perspective (preservation of the fluorite ceria lattice and hence higher oxygen storage capacity). A spongy morphology with some extent of agglomeration was noticed. In the pristine ternary oxide (Figure S5), La seems to prefer to be segregated (RGB analysis, Figure S5C). Additionally, as shown in Figure S5C, La can be found as a dopant (green inside the bulk of the shown particle, area 3), as a layer surrounding the surface of the particle (decorator, area indicated by red arrows), and as a segregate (pure green upper left, area 1). The much lower Cu content (10 at. %) secures that the role of copper is mostly as a dopant into the ceria lattice. This is reflected through the purple color in the RGB image (purple as a result of the R (red) + B (blue) combination, area 2 in Figure S5C). Dopant distribution is most likely nonuniform. The above inhomogeneity in the nanoscale is in agreement with the results obtained from powder XRD and Raman; the latter showed the formation of the LaO8 phase impurity peak. Additionally, line-scan EDX data (two lines, (i) and (ii)) are presented in Figure S5C(i,ii) aligned with the above La and Cu role.

Figure 3

(A) HRTEM, (B) HAADF-STEM, (C) RGB mapping, (D) SAED, and (E) EELS obtained over the DBM, 4 h Ce–La–10Cu–O metal oxide.

Following ball milling, La and Cu dopants are fairly dispersed over the ceria catalyst. As demonstrated, all materials are polycrystalline with high crystallinity, based on the SAED studies (Figure D and Tables S2 and S3) and in agreement with the XRD studies. Crystallinity is also maintained after ball milling. Compared to the pristine one, after DBM, a more homogeneous mixing of Ce, La, and Cu is achieved based on the RGB analysis (compare Figures C and S5C). Ce M4,5 EELS edges (Figure E) also show a chemical shift to a lower energy and change of the fine structure corroborating the presence of Ce3+, while Ce4+ is predominant in the pristine oxide (Figure S5E). This is an important finding as it confirms the contribution of milling process to the M–O bond breaking and formation of oxygen vacancies (Ov), which always accompany the presence of Ce3+ oxidation state. This result is in agreement with the EPR results presented earlier, and with the DFT estimated Ce–O bond lengths under 0, −5, and +5% strain levels (Table S1). Analysis of the SAED ring structure (Tables S2 and S3) is in good agreement with the fluorite lattice.

An insightful look and analysis of the pristine and DBM oxides is presented in Figure , where the dislocations (Figure F areas enclosed in boxes), atomic mobility (Figure D, areas enclosed in circles), and local refaceting (compare Figure A vs Figure C) are demonstrated. Such features are more intense in the case of DBM oxide (Figure C,D). The surface steps shown in the image demonstrate the surface/atomic mobility. In the ball-milled sample (DBM), faceting appears as the particle size is small (<10 nm, i.e., 5–8 nm) with sharp edges (shown with yellow arrows), whereas the pristine sample is smoother (Figure A,B). Additionally, based on the SAED data, in the case of pristine oxide (Figure S5D), the complete rings showcase the presence of more crystalline material exhibiting larger in size crystals.

Figure 4

(A,B) HRTEM of the pristine CeLaCuO; (C,D) HRTEM of the DBM CeLaCuO; (E) fast Fourier transformed (FFT) analysis of (B); and (F) inverse FFT analysis of (B) across the planes (111).

Textural Studies

It is noticed that the mechanochemical treatment of Ce–La–10Cu–O oxide yields to at least 100% increase in the specific surface area (m2 g–1) following both DBM and WBM conditions (Table and Figure S6). The improvement of such textural property is consistent with previous studies, suggesting that the ball milling process facilitates the formation of more micropores and particle size reduction, particularly under dry conditions (harsh ones) due to the mechanical forces exerted on the grains.[65,66] The porosity trends for the reference oxides are provided in the Supporting Information (Figure S6).

Redox Properties

Figure A presents the H2-TPR profiles of Ce–La–10Cu–O before (t = 0 h) and after the mechanochemical treatment in dry (DBM) and wet (WBM) atmospheres (t = 4 h). Ball milling after 4 h was only investigated since these conditions showed the preferred structural characteristics (based on XRD and Raman studies) among the wet ball and dry ball conditions. A profile with multiple peaks in three reduction regimes is observed, namely, T < 250 °C (region α), 250 °C < T < 500 °C (region β), and 500 °C < T < 700 °C (region γ). Interestingly, the pristine sample (εc only) presents larger concentration of easily reduced oxygen by hydrogen (increased mobility) in the regions α and β, whereas DBM (εC + εM) increases the solid’s reducibility at higher temperatures (region γ), as reflected by the relative amounts of H2 consumed (mmol/g) (Figure C). This demonstrates that the chemical strain activates mostly the easily reducible oxygen species, whereas the presence of mechanical strain contributes to the activation of the hardly reducible ones. The latter is important as a criterion of catalyst design for reactions taking place in the T > 500 °C region (e.g., DRM). The enhancement of the redox properties following mechanochemical treatment is in agreement with the study of Yang et al.[67] who compared the as-received commercial MnO2 with a modified one in a top-down ball milling approach. Nonetheless, an increase in the concentration of oxygen vacancies was observed, which ultimately boosted the reactivity and mobility of surface lattice oxygen. The above results can be understood on the theoretical basis of the ball milling process, where it was reported that high temperatures (>1000 °C) can be developed for periods of time in the scale of 10–3 to 10–4 s due to the friction between the balls and the materials. This leads to the formation of local hot spots which can drive the displacement of Olattice species leading to the formation of Ov3.

Figure 5

(A) H2-TPR profiles of pristine-, DBM-, and WBM-treated Ce–La–10Cu–O oxides; (B) deconvoluted TPR profile of DBM-treated Ce–La–10Cu–O oxide; and (C) redox site distribution over the pristine, DBM, and WBM oxides in the low-, medium-, and high-temperature regime.

Deconvolution of the H2-TPR profile for the Ce–La–10Cu–O oxide following DBM for 4 h is presented in Figure B. The H2-TPR profiles for the reference samples are given in Figure S7. After calibrating the TCD signal in the H2-TPR experiments for the Ce–La–10Cu–O-pristine, Ce–La–10Cu–O (DBM, 4 h), and Ce–La–10Cu–O (WBM, 4 h) solids, the amount of H2 consumed was found to be 4.5, 11.1, and 3.2 mmol/g, respectively, demonstrating the increase of the redox sites when both chemical and mechanical strains are present (DBM solid). Considering the SSA (m2/g) of each solid, the density of the redox sites was estimated to be 0.92, 1.4, and 0.22 mmol/m2 for Ce–La–10Cu–O (pristine), Ce–La–10Cu–O (DBM, 4 h), and Ce–La–10Cu–O (WBM, 4 h), respectively.

Surface Analysis (XPS)

Figure presents the XPS Cu 2p (Figure A) and O 1s (Figure B) core-level spectra of the oxides before and after WBM and DBM treatment. The Ce 3d along with the O 1s deconvoluted spectra for pristine, DBM, and WBM oxides are presented in Figures S8 and S9. According to the literature,[68] Cu (Figure A) exists in the 2+ oxidation state, which is evident by the broad satellite peak of Cu 2p3/2 arising at the high binding energy side (940–945 eV), clearly noticed after DBM and WBM modification. In the case of Ce 3d, all the intense peaks, namely, v0, v′, and v″ of Ce 3d5/2 and u0, u′, and u″ of Ce 3d3/2 spectra (Figures S8 and S9), are attributed to Ce4+, indicating that the dominant species is Ce4+ in all solids. As ceria exhibits high redox properties, it is useful to understand the effect of doping and mechanochemical treatment on the formation of Ce3+ species. According to Ardelean et al.,[69] the shoulder peaks appearing at ∼ 885 and 898 eV, labeled as v and u, respectively, are associated with Ce3+ species, and their intensities relatively decrease upon doping. Interestingly, the intensities of Ce3+ peaks increase after the mechanochemical treatment, which is consistent with the fact that the mechanical forces induce defects, such as oxygen vacancies. For the O 1s spectra (Figure B,C), two components can be traced; the one at ∼529 eV corresponds to lattice oxygen,[68] whereas the one at ∼531.5 eV can be associated with surface hydroxyl species as well as carbonate or polarized O2– ions surrounding the O vacant sites.[59] There are several reports associating this peak with neighbor to vacancies, as vacancies themselves cannot be traced using XPS.[70,71] It is observed that after mechanochemical treatment, there is a slight shift in the 529 eV peak maximum due to the change of the environment of the lattice oxygen and the introduced lattice distortion (in agreement with XRD and Raman studies). The calculated area proportion at 531.5 eV for these materials follows the order: WBM > DBM > P (Figure D). Given the structural studies performed (XRD and Raman), it can be stated that in the case of WBM oxide, the predominance of oxygen species at 531.5 eV is originated from OH species rather than neighbor to vacancies species.

Figure 6

(A) Cu 2p core-level spectra for pristine and ball-milled Ce–La–10Cu–O oxides; (B) O 1s core-level spectra for the pristine and ball-milled Ce–La–10Cu–O oxides; (C) deconvoluted O 1s spectrum of the Ce–La–10Cu–O following DBM; and (D) calculated O 2p areas as per the deconvoluted spectra in (C).

Oxygen Mobility Studies

As the lattice oxygen mobility is a descriptor for the surface reactivity of an oxide and the potential of oxygen vacancy formation, it is important to get an insight into the lattice oxygen (OL) mobility (diffusivity). The oxides of this study were tested for their 16O/18O exchange through TIIE (18O2-TIIE) experiments conducted at 350 °C. The temperature was chosen based on the catalytic reactions’ operational window. The process is governed by the descriptor functions described in Section . Differences in the transient kinetic rate profiles associated with these descriptors are mainly due to the differences in their oxygen sublattice structure [118], as also traced using Raman and H2-TPR techniques. It should be noted, however, that 18O2-TIIE was recently illustrated to be more sensitive to subtle structural differences compared to Raman spectroscopy.[28,72]

Figure A presents the R16O2 (μmol g–1 s–1) transient response curves for which Rmax values of 14, 17, and 31 μmol g–1 s–1 were estimated for the WBM, DBM, and pristine samples, respectively. It can be seen clearly that the exchange starts immediately after the step-gas switch to 2% 18O2/Ar/He and reaches a maximum rate at a relatively short time; the tmax values for 16O2 dynamic evolution are 14.3, 16.1, and 15.7 s for pristine, DBM, and WBM, respectively. All these times are much shorter than the tmax of 16O18O (s) dynamic rate as discussed in what follows. These initial rates of 16O/18O exchange reflect the surface exchange, while bulk oxygen 16O/18O exchange via diffusion occurring at longer times is largely related to the tmax value recorded.

Figure 7

(A) Transient rates (μmol g–1 s–1) of 16O2 formation as a function of time estimated from the TIIE experiment: 2 mol % 16O2/2 mol % Kr/Ar/He (350 °C, 30 min) → 2 mol % 18O2/Ar/He (350 °C, t) for the Ce–La–10Cu–O oxides. W = 0.02 g; F = 50 N mL/min; (B) transient rates (μmol g–1 s–1) of 16O18O formation as a function of time estimated from the TIIE experiment: 2 mol % 16O2/2 mol % Kr/Ar/He (350 °C, 30 min) → 2 mol % 18O2/Ar/He (350 °C, 15 min) for Ce–La–10Cu–O solids. W = 0.02 g; F = 50 N mL/min; (C) evolution of the total amount of 16O exchange (N(16O), mmol g–1) as a function of time estimated from the TIIE experiment: 2 mol % 16O2/2 mol % Kr/Ar/He (350 °C, 30 min) → 2 mol % 18O2/Ar/He (350 °C, 15 min) for the Ce–La–10Cu–O oxides. W = 0.02 g; F = 50 N mL/min; (D) αg(18) descriptor parameter as a function of time (t) estimated from the TIIE experiment: 2 mol % 16O2/2 mol % Kr/Ar/He (350 °C, 30 min) → 2 mol % 18O2/Ar/He (350 °C, 15 min) for the Ce–La–Cu–O oxides. W = 0.02 g; F = 50 N mL/min.

Figure B presents the transient 16O/18O exchange rate toward the 16O18O formation (R16O18O) as a function of time at 350 °C. It is illustrated that the exchange rate starts immediately after the switch to the isotopic gas mixture and passes through a maximum at larger times (tmax > 100 s) compared to the 16O2 formation rate for all the oxides studied herein. This result indicates that bulk diffusion of 16O toward the surface to combine with 18O on the surface is slow, and this diffusion-limited process is clearly different among the three oxides studied (pristine, DBM, and WBM). For the pristine oxide, the Rmax(16O18O) value is 13.1 μmol g–1 s–1, while following DBM and WBM treatment, it takes the values of 8.5 and 4.8 μmol g–1 s–1, respectively. The 35% reduction in the Rmax(16O18O) value after DBM can be understood on the basis of defect formation that slow down the oxygen diffusion, acting themselves as mechanical boundaries/barriers. The tmax16O18O value is 147.2 s for pristine, 229.1 s for DBM, and 303.2 s for WBM. These results demonstrate that the oxygen binding strength of the Ce–O–La, Ce–O–Cu, and Cu–O–La local environments is modified after ball milling (co-presence of chemical and mechanical strain) most likely due to the structural disturbances introduced through the mechanical forces applied. The extent of this modification depends on the closest neighbor (Figure C,D).

Figure C displays the transient evolution with time of the total amount of 16O (N, mmol g–1) able to exchange with gaseous 18O2 at 350 °C. The observed differences in the N16O(t) descriptor function for the samples under study reflect the differences in the intrinsic transient kinetic rate of subsurface lattice 16O diffusion. The latter depends on the structure of the oxygen sublattice and the local energy barriers encountered for oxygen mobility, which are determined by the chemical composition of the solid. It has been reported[28,73] that oxygen diffusion via grain boundaries on ceria-based catalysts is much faster than the bulk oxygen diffusion. It is also noted that the total amount of 16O species, N(16O), exchanged varied in the 8.2–12.4 mmol O g–1 range (see Table ) for the pristine, DBM, and WBM oxides).

Table 2

Single-Value Parameters Derived from the Transient Response Curves Recorded during the TIIE Experiment: 2 mol % 16O2/2 mol % Kr/Ar/He (350 °C, 30 min) → 2 mol % 18O2/Ar/He (350 °C, t) over the Ce–La–10Cu–O Samplesa

	oxide
parameter	pristine	DBM	WBM
Rmax16O2 (μmol g–1 s–1)	32.2	18.7	14.2
Rmax16O18O (μmol g–1 s–1)	13.1	8.5	4.8
tmax16O2 (s)	14.3	16.1	15.7
tmax16O18O (s)	147.2	229.1	303.2
N(16O) (mmol16O g–1)	12.4	11.4	8.2

W = 0.02 g; F = 50 N mL/min.

Figure D presents the dynamic evolution of the αg(18)(t) descriptor parameter (see eq ) with time upon exposure of the solids in the 2 mol % 18O2 isotopic gas mixture. The higher the value of this parameter, the slower the exchange of lattice oxygen with 18O2, thus the smaller oxygen mobility (effective diffusivity) in the oxygen sublattice of the solid. It is shown that ball milling deteriorates oxygen mobility compared to the pristine Ce–La–10Cu–O metal oxide according to the αg(18)(t) transient response curves (see the first 300 s of the transient). These results are consistent with those of Figure B, where larger tmax is recorded for WBM compared to the other solids.[28] Furthermore, the pristine sample (chemically strained) appears to possess more labile surface and bulk lattice oxygen species (Figure D) in harmony with the H2-TPR results (Figure ), where the same solid showed the lowest concentration of high-temperature reducible lattice oxygen species.

CO2 Metal Oxide–Surface Interaction

Probing CO2-Philicity Using Thermal Desorption

Figure A presents the CO2-TPD profiles obtained over Ce–La–Cu–O pristine, DBM, and WBM, while those for the reference oxides are given in Figure S10. Generally, it is well known that CO2 adsorption and desorption probe the strength of the basic sites (Mδ+–Oδ− entities) on the surface. The peaks observed at temperatures below 200 °C correspond to weak basic sites, the peaks at 200–450 °C represent moderate basic sites, while peaks above 450 °C are linked to strong basic sites.[74]Figure depicts the CO2-TPD profiles over the pristine, DBM, and WBM Ce–La–10Cu–O oxides, aiming to probe the modifications induced by ball milling (chemical and mechanical strain) on the CO2-philic sites on the surface. Calibrating the CO2-TPD signal for the Ce–La–10Cu–O-pristine, Ce–La–10Cu–O (DBM, 4 h), and Ce–La–10Cu–O (WBM, 4 h) solids, the total amount of CO2 adsorbed on the surface was found to be 1.48 (0.30), 6.03 (0.74), and 3.78 (0.26) mmol/g (mmol/m2), respectively. The numbers in parentheses correspond to basic sites concentration per unit surface area. As can be seen, the DBM enriches the surface basicity overall, while the WBM induces milder modification at temperatures below 250 °C and a redistribution of the CO2-philic sites. In particular, the DBM increases the concentration of low- and high-strength basic sites, while it reduces that of the medium-strength basic sites (Figure B). It is anticipated that ball milling through the imposed mechanochemical forces causes reduction of the particle size, as well as structural disordering, thus creating more adsorption sites per gram basis, as proved in what follows.

Figure 8

(A) CO2-TPD profiles over pristine, DBM, and WBM Ce–La–10Cu–O oxides and (B) distribution of basic sites (%) in the low, medium, and high strength basicity regions over the pristine (P), DBM, and WBM metal oxides.

It has been reported that Lewis acidity (e– acceptor characteristic) and basicity (e– donor characteristic) have been correlated with the differences in powder processing method, as they can affect the triboelectric phenomenon,[75] powder flow,[76] and other surface and powder properties.[77] The milling process has been correlated with the increase of electron-donor characteristic of the surface and the increase of basic sites due to introduced defects, leading to higher oxidation activity.[78]

Vibrational Spectroscopy Tools

To monitor the CO2 activation pathway over the two surfaces of the most interest (pristine and DBM), in situ CO2-DRIFT spectra were recorded at 350 °C on the unreduced catalysts as shown in Figures and S11. The assignment of the adsorbed CO2 IR bands was based on the open literature. Figure A presents the DRIFTS spectra in K–M units (1750–1150 cm–1 range) recorded under the 5 vol % CO2/He gas mixture and after 30 min on stream over the Ce–La–10Cu–O-DBM and Ce–La–10Cu–O-pristine metal oxides. The K–M units allow for a quantitative interpretation of the bands’ intensities recorded.

Figure 9

(A) DRIFTS spectra recorded under 5 vol%CO2/He (30 min) gas mixture in the 1750–1150 cm–1 range over Ce–La–10Cu–O—DBM and Ce–La–10Cu–O—pristine solids. Deconvoluted IR spectra of Ce–La–10Cu–O—pristine (B) and Ce–La–10Cu–O—DBM (C) solids. (D) Ratio of area(t)/area(t = 0) related to peak 1 and peak 2 for both the pristine and DBM materials.

After spectra deconvolution, both samples show IR bands at 1590 cm–1 (peak 1) and 1554 cm–1 (peak 2), which correspond to bidentate carbonate species (Figure B,C). A closer look at the CO2 adsorption IR bands for the pristine oxide (chemically strained) (Figure B) gives the IR band recorded at 1554 cm–1 as the predominant one (areaPeak1/areaPeak2 = 0.32). Following the DBM process, where both chemical and mechanical strain co-exist, the surface chemistry apparently changes, and the predominant IR band becomes the one at 1590 cm–1 (Figure C; areaPeak1/areaPeak2 = 1.96). According to the literature, the IR band recorded at 1554 cm–1 over unreduced ceria shifts to higher wavenumbers (1590 cm–1) over a reduced CeO2, and after reoxidation, the IR band shifts back to 1554 cm–1.[78] The areas ratio of peak 1 to peak 2 for the pristine sample was found to be 0.32, and for DBM 1.96, indicating that the DBM process (addition of mechanical strain) leads to an increase in the surface concentration of defects/CO2 adsorption sites by ∼6 times compared to pristine (chemical strain only). The IR band at 1468 cm–1 (peak 3) and 1425 cm–1 (peak 4) can be assigned to polydentate carbonates and bicarbonate species, respectively.[79] The IR bands at 1322 cm–1 (peak 5) and 1285 cm–1 (peak 6) correspond to carbonates and bidentate carbonates, respectively.[80] These CO2 adsorption-DRIFTS experiments were performed at 350 °C, where medium-strength basic sites were present for both the pristine and milled solids. The concentration of medium-strength basic sites is higher for pristine compared to DBM (Figure B), in agreement with the DRIFTS results (Figure A).

The thermal stability of the adsorbed species formed after CO2 interaction at 350 °C can be studied based on the data reported in Figures D and S11. The IR bands for both materials decrease with time in Ar gas flow following a 30 min CO2 gas treatment (Figure S11). The ratio in the integral band area of peak 1 and peak 2 recorded after 30 min in CO2/Ar to that after 10 min in Ar gas flow remains the same for both samples, indicating that the IR band at 1590 cm–1 was not shifted to lower wavenumbers (1554 cm–1) and that both materials were not reoxidized. Figure D shows the dynamic evolution at 350 °C of the ratio of areas of peak 1 or peak 2 under Ar gas flow to that at time zero (under CO2/He gas mixture). As illustrated in Figure D, more stable carbonates are formed over the pristine (chemically strained) compared to the DBM (chemically and mechanically strained) solids. This thermal stability should be related to the CO2–surface interaction as described using the theoretical approach. Additionally, upon the CO2 adsorption on the surface, the O–C–O angle for 0, 5, and −5% was found to be 120.04, 119.88, and 124.75°, respectively. The value for the tensile strain (119.88°) is more realistic as discussed earlier, that upon DBM, tensile strain was developed in the solids (Table ). However, the difference between the two angles (120.04 vs 119.88°) is insignificant and hence not conclusive. Furthermore, the CCO–Osurface distance (Å) (Tables S19 and S21) was found to be shorter under 5% tensile strain compared to zero and compressive strain, corroborating for a higher interaction/CT in the tensile strain conditions. This difference is more pronounced in the case of CeO2 and Ce–La–O but not in the case of Ce–La–Cu–O solid. However, it is noteworthy that the two angles Osurf–C–O1 and Osurf–C–O2 (O1 and O2: the two CO2-derived oxygen atoms) exhibit the highest difference for Ce–La–Cu–O under 5% tensile strain compared to zero strain for the same solid (pristine). This inequality in the two angles supports the different contribution of the O1 and O2 atoms of the CO2 molecule in the adsorbed state in pristine and DBM solids, leading to different thermal stabilities of the formed carbonates. Furthermore, the energy of CO2 adsorption was found to be −1.33 and −1.15 eV for the zero and +5% (tensile) strain cases. This is a strong indication for the higher stability of carbonates over the chemically strained surface (Ce–La–Cu–O, pristine) and for the role of mechanical strain (tensile) to engineer the CO2 adsorption sites might be through structural reconstruction. In addition, the higher population of medium-strength basic sites on the pristine surface, over which the bidentate carbonates are formed, gives rise to a higher population of carbonates on the surface, thus prolonging their presence under thermal treatment.

DFT Studies

Ov Chemical Environment

Oxygen vacancies at different chemical environments (locations) in pure ceria and doped system (Ce–La–O and Ce–La–Cu–O) slab were created as explained in Section (see Figure and Table ) to have a qualitative comparison and deeper insights regarding lattice strain (doping effect) and biaxial strain (external stimulus) effects on the defective systems’ stability. The pure CeO2(111) depicted in Figure A was doped with La in the first layer and labeled Ce–La–O, which is shown in Figure B. Figure C illustrates the addition of copper (Cu) and lanthanum (La) to the ceria slab to resemble the oxides discussed in the experimental part (section 2.1). In this work, the same annotations as in our previous publication[59] was followed; the single vacancies on the surface layer are denoted as black, and the vacancies generated in the first subsurface layer are represented by cyan. The following labels were used for various vacancies: single surface (1,2,3,4), single subsurface (5,6,7,8), double of surface and subsurface (combination of the corresponding locations of the vacancies, e.g., 1,7), double of surface vacancies (combination of the corresponding locations of the vacancies, e.g., 1,2), and double vacancies both occurring at the subsurface (combination of the corresponding locations of the vacancies, e.g., 6,8). These distinct scenarios of surface defects were considered (see Table ), and their effect on the slabs’ stability was systematically investigated. All the obtained values are provided in Tables S4–S18.

Figure 10

Schematic representation of oxygen vacancy (Ov) created in the uppermost surface (black) and subsurface (cyan): (A) pure CeO2(111), (B) La-doped CeO2 (Ce–La–O(111)), (C) La and Cu co-doped CeO2 (Ce–La–Cu–O(111)), (D,E) first neighbor configurations of the possible oxygen vacancies in Ce–La–Cu–O(111), and (F) demonstration of the compressive and tensile biaxial strain.

Table 3

Configurations of Oxygen Vacancies (Ov) Considered for the DFT Calculations Performed in This Study

		vacancy site
type of Ov	notation	Ce–La–O system	Ce–La–Cu–O system
single surface (SSV)	1	1: La–□–Ce	1: La–□–Cu
	2	2: La–□–Ce	2: Cu–□–La
	3	3: Ce–□–Ce	3: Ce–□–Cu
	4	4: La–□–Ce	4: La–□–Ce
single subsurface (SSSV)	5	5: La–□–Ce	5: La–□–Cu
	6	6: La–□–Ce	6: Cu–□–La
	7	7: La–□–Ce	7: La–□–Ce
	8	8: Ce–□–Ce	8: Ce–□–Cu
double surface (DSV)	1,2	Ce–□–Ce and La–□–Ce	La–□–Ce and La–□–Ce
	3,4	Ce–□–Cu and La–□–Ce	La–□–Cu and Cu–□–La
double subsurface (DSSV)	6,8	La–□–Ce and Ce–□–Ce	Cu–□–La and Ce–□–Cu
	5,6	La–□–Ce and La–□–Ce	La–□–Cu and Cu–□–La
	7,8	La–□–Ce and Ce–□–Ce	La–□–Ce and Ce–□–Cu
	5,8	La–□–Ce and Ce–□–Ce	La–□–Cu and Ce–□–Cu
double surface and subsurface (DSSSV)	1,7	La–□–Ce and La–□–Ce	La–□–Cu and La–□–Ce
	4,7	La–□–Ce and La–□–Ce	La–□–Ce and La–□–Ce
	2,5	La–□–Ce and La–□–Ce	Cu–□–La and La–□–Cu

Single Surface Vacancies

Figure A shows that the oxygen vacancy formation energy (EOf) in the pure CeO2 was in the range of 2.64–2.66 eV for different Ov configurations at zero mechanical strain, which is in good agreement with the previous report for the zero, tensile, and compressive strain if the p(2 × 2) case is considered.[42] Doping with La (Ce–La–O surface) drops the EOf to the 1.09–1.30 eV range at zero mechanical strain, where the lowest EOf value was calculated at the single surface vacancy (SSV) 3 (Ce–Ov–Ce), near Ce atom, and the highest one at the La nearest neighbor (La–Ov–Ce, configuration 2). In the case of La and Cu co-doping (SSV1), the EOf value was further reduced to −1.62–0.56 eV. The oxygen vacancy that is located between Cu and La had the lowest energy (most stable), whereas the first neighbor vacancy to La (single surface vacancy 4 = La–Ov–Ce) exhibited the highest energy.

Figure 11

(2 × 2) CeO2(111) (red), Ce–La–O (blue), and Ce–La–Cu–O (green) oxygen vacancy energy of formation (EOf) under the compressive (−5 to 0%) and tensile (0–5%) biaxial strain for (A) single surface vacancy (SSV), (B) single subsurface (SSSV), (C) double surface (DSV), (D) double subsurface (DSSV), and (E) double surface and subsurface configurations (DSSSV).

Mechanical (Biaxial) Strain Effect on EOf

Under the biaxial strain of ±5% in pure and doped CeO2, the oxygen vacancies showed a similar effect and stability. With some exceptions, the EOf value decreases with increasing tensile strain, indicating more stability as we move from compressive (−5%) to tensile (+5%) strain. In the case of single oxygen vacancies in pure CeO2 at +5% strain, the EOf value ranged between 1.15 and 1.37 eV and between 3.75 and 3.89 eV for −5% strain. A similar monotonical trend of decrease was observed upon doping CeO2 with La, but with lower EOf values (more stable with increasing tensile strain, similar to pure CeO2). Contrary to the trend observed in pure ceria and La-doped systems, on the ternary system (Ce–La–Cu–O), the slab stability’s behavior presented an opposite trend; the EOf values were found to be favorable at compressive strain and zero strain conditions. Despite the fluctuating nature of EOf, the values are generally increasing with tensile strain.

Single Subsurface Vacancies

Figure B demonstrates the effect of chemical strain (doping) and biaxial strain (mechanical forces) on the single vacancies created in the first subsurface (SSSVs) of the three studied oxides. The profile of the SSSV is similar to that of the SSV, but the slope of the energy–strain curves for the SSSV is smaller (less steep) than for the SSV. For CeO2, under zero strain, the subsurface oxygen vacancies (SSSVs) are more favorable than the single surface vacancies (SSVs). The EOf values are in the range of 1.95–2.2 eV and decrease with increasing tension by 1% (1.11–1.53 eV). The stability of the different configurations of the single subsurface vacancies (SSSVs) at 0, +5, and −5% strain values can be arranged ascendingly as 5 < 7 < 6 < 8, 8 < 5 < 7 < 6, and 6 < 5 < 7 < 8, respectively. A similar behavior was also observed in the Ce–La–O slab’s case under zero strain but with lower formation energy values (0.53–1.11 eV). The highest EOf values were observed in the case of SSSV 7 (La–Ov–Ce), while the lowest ones were for SSSV 8 (Ce–Ov–Ce). The profile of EOf remains the same for the doped Ce–La–O and undoped CeO2 cases under compression and tension.

The Ce–La–Cu–O system exhibited the lowest EOf values at the subsurface sites, which indicates that oxygen vacancies are more likely to occur spontaneously after adding copper. It is well established that the EOf values can be a qualitative descriptor of the chemical activity of an oxide surface.[81] Among the SSSV sites 5, 6, 7, and 8, site 7 (La–Ov–Ce) showed better stability at +5% strain. The system’s behavior fluctuated in the strain span, ca. −2–0%, showed low stability, followed by a drop in EOf at 1% but deviating again at 2% strain. Generally, in the Ce–La–Cu–O oxide, the SSSVs 5, 6, and 8 exhibited higher stability in the compression range compared to the SSSV 7; this designates the important role of Cu being placed at an adjacent site to the Ov.

Double Surface Vacancies

Figure C depicts the double vacancy formation energy at the three oxides’ surface layers. Ce–La–Cu–O reached the highest stability at all the strains by achieving the lowest values of EOf compared to pure CeO2 and the binary system (Ce–La–O). As an example, the double surface vacancy (DSV) 1,2 and DSV 3,4 configurations in Ce–La–Cu–O under zero mechanical strain (εM = 0) showed energies of 0.03 and 0.18 eV, respectively. The same configurations in CeO2 and Ce–La–O had much higher energies, ca. 2.765, 2.764, and 2.05, 2.07 eV, respectively. As previously mentioned, the higher the % tension, the higher the Ov stability, and this is also supported by the Ce–O bond lengths (see Table S1 and Figure C,D). The double vacancies (DSV) in the surface and subsurface (DSSV) slabs suggest lower stability of CeO2 in all ranges of tensile and compression strains applied than the single surface vacancy. Moreover, by comparing between the double vacancy configurations in CeO2 (Figure C–E), the stability appears to be the highest when both vacancies are created in the subsurface (0%: 2.17–2.37 eV) and to be the lowest when a combination of surface and subsurface vacancies are suggested (zero strain: 2.72–3.03 eV). This indicates the difficulty and the low possibility to have two Ov released from two different oxygen layers; however, the energies drop by 0.7–1 eV at +5% strain.

Similarly, for Ce–La–O, the double subsurface vacancy (DSSV) formation energy under zero mechanical strain (εM = 0) lies between 1.49 and 1.65 eV, while the vacancy in surface and subsurface vacancies have higher formation energies (0%: 2.01–2.65 eV) but tends to stabilize as the tensile strain increases (5%: 1.35–1.89 eV). For Ce–La–Cu–O DSSVs, the EOf value is lower by approximately 2 eV at zero mechanical strain (εM = 0) compared to CeO2 and Ce–La–O (−0.10, −0.10, and 0.01 eV). It was also noticed that the behavior of the Ov formation energies at 5% tension was not consistent throughout the suggested configurations in the double surface (see Table S7). For instance, 1,7, 7,4, and 2,5 cases of vacancies exhibited low stability at +5% strain (0.17 eV), while 3,8, 1,5, and 2,6 showed the opposite behavior at −5% strain.

CO2–Surface Interaction and Chemistry of Activation

CO2 Adsorption Energy under Chemical Strain (εc) and Zero Mechanical Strain (εM = 0)

CO2 adsorption calculations were performed on the pure and doped CeO2 surfaces (Ce–La–O and Ce–La–Cu–O) since carbon dioxide has a preference to be adsorbed on the ceria/doped ceria support during catalytic reactions.[82] In this scenario, only the chemical strain was considered; the latter originates from the doping and the subsequent Ov presence. For these calculations, the oxygen vacancy on site 2, single surface vacancy (SSV 2: La–Ov–Ce or Cu–Ov–La) configuration was chosen based on the reported literature.[83] This part of the study aims to enlighten the chemical strain effect on the CO2 adsorption, which is reflected on the doping. As clearly shown in Figure A,B and Table , La doping drops the Eads from −0.97 eV (CeO2 defective structure) to −1.01 eV; La and Cu ceria co-doping further enhances the spontaneity of the CO2 adsorption process (Eads = −1.40 eV) (see Figure C). It should be noted that the adsorption energy in the presence of the oxygen vacancy in the CeO2 was approximately as that reported in the literature[83] (−0.91 eV), where CO2 is adsorbed to a surface oxygen (Os, shown in Figure A–C) near to the vacancy site and which ultimately forms carbonate species (CO32–).[83] Further calculations have been conducted on different sites which showed that CO2 adsorption on vacancy sites takes place with a low adsorption energy. Therefore, the rest of the calculations were established assuming the above chemical environment (adsorption site) since it was reported to be the most stable compared to the one where the CO2 is directly adsorbed onto an oxygen vacancy.[83] It should be mentioned at this point that the present study did not focus on optimizing all the possible CO2 adsorption sites on the surface, and thus, only the linear configuration of CO2 was studied.

Figure 12

(A) CO2 adsorption on defective CeO2(111); (B,C) CO2 adsorption on the doped-CeO2(111) surfaces at site 2 at zero strain (all the surfaces are considered with surface Ov); and (D) CO2 adsorption on CeO2(111) and doped-CeO2(111) surfaces under the strain effect in the absence of oxygen vacancies (Ov).

Table 4

Adsorption Energies and Geometric Parameters for the Activated CO2 Molecule on CeO2(111), Ce–La–O(111), and Ce–La–Cu–O(111); Bond Length (Å), C–Osurf (Å) Distance from the Surface, and O–C–O Angle (°) with the Presence of a Surface Oxygen Vacancy (Ov) at Zero Straina

oxide system	αO–C–O (deg)	dO=C (Å)	dC–Osurf (Å)	Eads (eV)
*CeO2(111)	128.77	1.28/1.26	1.37	–0.97
*Ce–La–O(111)	127.62	1.22/1.32	1.40	–1.01
*Ce–La–Cu–O(111)	124.71	1.25/1.31	1.34	–1.40

CO2 linear configuration (L) was considered.

Based on the EOf results presented above (Section ), it is expected that the CO2 adsorption is enhanced with the presence of La, Cu dopants due to the facility of the Ov formation (Figure A), which means activation of CO2 will be more efficient. Figure shows the activation of the linear CO2 molecule, which is accompanied by the decrease in the O–C–O angle and the elongation of the O=C bonds (Table ). In particular, the intramolecular C–O bonds of the adsorbed species appeared to be elongated to 1.22–1.32 Å compared to 1.13 Å in the free CO2 molecule; this corresponds to a strong deformation and CT to be discussed later.

CO2 Adsorption Energy without Oxygen Vacancies (Ov) under the Mechanical Strain Effect (εM≠ 0)

CO2 adsorption energy (Eads) calculations were also performed on the three surfaces (CeO2, Ce–La–O, and Ce–La–Cu–O) in the absence of oxygen vacancies. The CO2 adsorption was considered under one effect at a time (e.g., doping effect vs external strain effect) in order to decouple the combined effects of each of these two factors and their role on the CO2 adsorption. In the herein calculations, the chosen adsorption site was also on top of a surface oxygen atom (Osurf) as in the previous case. The calculations were carried out regarding this specific site, starting from adsorbing the CO2 on the clean CeO2 for which a good agreement with the literature (−0.36 eV) was found (formation of carbonate species). Figure D shows that the adsorption of the CO2 molecule on CeO2 (111) and Ce–La–O (111) surfaces follows the same behavior, whereas as the biaxial tensile strain (external stimulus) increases, the Eads value drops, indicating stronger adsorption. Introduction of the dopant in the ceria lattice facilitated the adsorption significantly. A fluctuating behavior of the adsorption energy was observed, where the Eads value varied between a decrease (more negative value) as the compressive strain decreased (−3 to −1%), to reach its strongest binding at 0% strain, and then an increase (less negative value) as the tensile strain increased. It is also important to point out that the Eads values at zero strain were more negative with Ov present on the surface compared to the cases where Ov were absent (as can be observed in Figure ). This is due to the enhanced CT from the surface to the CO2 molecule.[84] In other words, the high basicity of the surface (CO2-TPD earlier studies) is indicative of the CT from the surface to the CO2 molecule in order for the latter to be activated; the CT occurs due to the electrons residing from the O 2p shell upon creating the Ov.[85] The presence of dopants weakened the M–O bond (see Table S1) and facilitated the generation of oxygen vacancies that strengthened the Eads of CO2.[86]

Figure 13

CO2 adsorption on CeO2(111) and doped CeO2(111) surfaces at the 0, 5, and −5% strain level in the absence of oxygen vacancies (Ov).

Charge Transfer

To gain understanding on the adsorbate/surface charge distribution upon CO2 adsorption on the best performing surface of doped CeO2(111) with lanthanum and copper, Bader charge analysis along with charge density difference has been performed. Results of the Ce–La–Cu–O surface considering first the effect of oxygen vacancies at zero strain followed by the Ce–La–Cu–O surface at different levels of strain in the absence of oxygen vacancies are shown in Figure .

Figure 14

(A) Bader charge analysis, charge density difference (B) top view, and (C) side view of CO2 adsorbed on Ce–La–Cu–O with oxygen vacancy at site 2 at zero strain. Bader charge analysis with the strain effect on CO2 adsorbed on Ce–La–Cu–O at (D) 0, (E) +5, and (F) −5% strain in the absence of the oxygen vacancy (Ov). Yellow and blue regions in charge density difference plots denote charge accumulation and depletion, respectively. An isosurface value of 0.0025 eÅ–3 is used.

Effect of oxygen vacancy: The effect of oxygen vacancy (Ov) on the CO2-derived carbon atom (Cδ+) and the CO2-derived oxygen atoms (Oδ−) and charge changes on the surface are discussed next. Upon CO2 adsorption on the Ce–La–Cu–O surface and in the presence of Ov, the charge associated with the CO2-derived carbon atom was found to be +2.10 |e−| (Figure A), whereas in the absence of Ov was calculated to be +2.20 |e| (Figure D), both being compared to the 0% strain condition. Thus, the presence of Ov enhances the electron accumulation on the carbon atom (Cδ+) as its value is found to be less positive. In Figure B,C, the charge accumulation region on the C atom is designated with the blue region.

In the case of CO2-derived oxygen atoms, one of the oxygen atoms (O1) shows similar values (−1.04 and −1.03 |e|) in both cases, whereas the second oxygen atom (O2) had distinct charges of −1.07 and −0.82 |e| in the case of the Ce–La–Cu–O surface with and without oxygen vacancy, respectively. Therefore, the O2 atom in the adsorbed CO2 had higher charge accumulation in the presence of Ov; hence, in the presence of Ov, a stronger electric field is formed due to the excess electrons surrounding the oxygen atoms belonging to CO2. Additionally, the charge density difference plot reveals the region of electron accumulation around the O2 in the CO2 molecule (Figure B,C), anticipating the participation of this oxygen atom in charge-exchange events with the surface (surface acting as the electron donor).

Effect of applied strain: Comparing the tensile (+5%) and compressive (−5%) strain to the zero strain condition in the case of Ce–La–Cu–O, it can be seen that +5% strain has a negligible effect on the CO2 electron transfer toward/from the surface based on similar values of charges on the adsorbed carbon and oxygen atoms (2.22 for carbon, −1.04 and −0.84 |e| for both oxygens) compared to the 0% strain case (2.22 for carbon and −1.03 and −0.87 |e| for both oxygens). In the above reported values, the O atoms with the low charge value are the ones pulled upward along the z direction. However, upon applying compressive strain (−5%), both oxygens are pulled toward the surface having more negative charge than in the 0 and +5% strain cases.

The CO2-derived C atom (Cδ+) is also more enriched with electrons having a charge of 2.12 |e| (Figure F) compared to that of 2.22 |e| (Figure D,E). Thus, these charges under −5% applied strain suggest the formation of covalent bonds within the CO2/La–Ce–Cu–O system, thus indicating strong CO2 adsorption.

Electronic DOS: Influence of Chemical Strain (εC) and Mechanical Strain (εM) on the Electronic Structure of Doped CeO2(111) Surfaces

In this section, the role of Ov as well as that of tensile and compressive strain on the DOS of the pristine and doped (La)/co-doped (La, Cu) ceria is discussed. The relative position of the CB and VB is important when adsorbent–adsorbate interactions have a covalent characteristic. In such a case, strain can alter the energetic positions of the bands and thereby the adsorption properties. Furthermore, covalent interaction as that dictated by the VB–CB splitting means higher dependence of basicity from the strain.[87] The calculated DOS for clean and defected—single and double oxygen vacancies—CeO2(111), Ce–La–O(111), and Ce–La–Cu–O(111) are given in Figures S13, and S14.

Figure 15

DOS of Ce–La–Cu–O(111) under −5, 0, and +5% strain level with different configurations of (A) clean surface, (B) reduced surface, single oxygen vacancies, and (C) reduced surface, double oxygen vacancies. The dashed vertical line represents the Fermi level.

CeO2(111) Surface

The main contribution to the CeO2(111) valence band (VB) and conduction band (CB) configurations is expected to originate from the O 2p and Ce 5d orbitals allocated below and above the Fermi level, respectively. The Fermi level is represented by a dashed vertical line in all DOS plots. It is important to note that in most cases, the value of the band gap (Eg) calculated using DFT is underestimated, mainly due to the exchange–correlation derivative discontinuity.[88]

In the case of clean CeO2(111) (absence of oxygen vacancies), no gap states were observed in the wide band gap region between the VB and CB (O 2p and Ce 5d).[89,90] Such a DOS profile is characteristic of an insulator with empty f-orbitals, the case of no surface defect formation.

However, in the reduced CeO2 case, it is noted that the presence of both single Ov (Figure S13B) and double Ov (Figure S13C) results in an observable reduction of the O2p–Ce5d band gap. This may be due to the fact that oxygen vacancies increase the number of readily available excess electrons in both surface and subsurface layers. This in turns induces charge compensation by Ce atoms to minimize the bonds between neighboring oxygen atoms that ultimately result in the smaller O2p–Ce5d gap. Prominently, in the case of reduced ceria surfaces with single and double oxygen vacancies, the −5% applied strain reveals the greatest reduction in band gap energy. Besides, a high peak of spin-up defected state at around 0.4 eV is witnessed with −5% applied compression for surfaces with single oxygen vacancy, while in the cases of double oxygen vacancies, the highest intensity of 0% strain is observed in the band gap region.

Doped CeO2(111) Surfaces

To provide an in-depth understanding of the effect of lanthanum and copper doping (chemical strain) on the electronic structure of ceria coupled with the mechanical strain (induced by the imposed biaxial strain), the DOS of doped CeO2(111) along with surface defects and imposed biaxial mechanical strain applied are presented for La-doped-CeO2 (Figure S14) and for La, Cu co-doped CeO2 (Figure ). First, it can be seen that the presence of La and Cu dopants causes more complex splitting of the Cu 3d states and La 4f states formed from the effect of the dopant compared to that of pure CeO2 surface (Figure S14).

Ce–La–O(111) Surface

In this case, coupling of both lattice strain (Ov) and mechanical biaxial strain effects leads to the following noteworthy results: (a) −5% compression strain has been observed to result in the lowest band gap for both cases of reduced Ce–La–O(111) surface; single and double oxygen vacancies (Figure S14B,C green curve). Moreover, the single and double oxygen vacancies in the Ce–La–O-doped surface incurred splitting in the suspected Ce 4f- and La 4f-related states, with greatest splitting intensities obtained for the case of −5% for single Ov and 0% for double Ov (Figure S14B,C). These mixed-occupied Ce 4f and La 4f states within the band gap are interpreted as an increased CT efficiency of the surface, a factor usually promoting surface activity in catalytic reactions.[91]

Ce–La–Cu–O(111) Doped Surfaces

In this case, it is observed that the CB shifts toward the VB (or closer to the Fermi level) as we move from the clean surface to the reduced surface with one Ov to the reduced surface with 2 Ov. The shift is significant in the case of the reduced Ce–La–CuO(111) surface with 2 Ov. Therefore, among the Ce–La–Cu–O configurations studied, the incorporation of double Ov results in the smallest band gap (Figure C). This smallest band gap could be the result of Ce 4f with Cu 3d- and La 4f-related states formed via the Ce–La–Cu–O doping along with the promoted double vacancies. Such impurity states are expanded in order to connect with the O 2p band, which narrows the band gap width in the DOS of ceria, promoting enhanced oxygen mobility and leading to a higher catalytic activity.[91] Upon applying compressive strain in the Ce–La–Cu–O reduced surface (Figure C, red curve), the CB band shift to the Fermi level is the most pronounced compared to the zero strain and tensile strain conditions, causing the highest reduction in band gap. These results are consistent with the above energy of oxygen vacancy formation (EOf) in which the reduction in the band gap found after ceria surface doping is consistent with the decrease of EOf for Ce–La–O and Ce–La–Cu–O CeO2(111) surfaces. From the projected DOS calculations (Figure S15) performed at 0% strain, only for the case of Ce–La–Cu–O oxide of interest, it was found that Cu 3d-induced impurity states with new peaks extended from the O 2p band toward the Fermi level, which narrows the band gap in total DOS (Figure ).

On the Strain Tuning of the CO2 Activation over the Ce–La–Cu–O Surface through the DFT and Experimental Tools

Metal Oxide Role As Acid/Base

A metal oxide can behave as a Lewis acid (electron acceptor); in such a case, the M sites interact with the CO2-derived oxygen forming M–OCO entities. In the case where the metal oxide acts as a Lewis base (electron donor), the O2–surface site interacts with the CO2-derived carbon atom forming O2–surface–CCO entities.[13]

CO2 can react with a metal oxide acid/base surface toward the formation of carbonate or bicarbonate species through the participation of Osurface or OHsurface, respectively, as well as linear adsorbed species, the latter being perpendicular or parallel to the surface. It has been reported that the surface with acidic characteristic promotes the linear-type adsorbed CO2 species, whereas basic characteristic promotes the bent configurations (e.g., CO2– or HCO3– species). During the CO2– species (bent configuration) formation, electron transfer to the π antibonding orbitals of CO2 takes place.[13]

CO2–Ce–La–Cu–O(111) Interaction and Surface-Bound Species Geometry

According to the literature,[92] the CO2 activation depends on the Ov and the CT; the latter can be probed by the O–C–O angle. The most stable configurations of CO2 into contact with the relaxed structures of CeO2(111), Ce–La–O(111), and Ce–La–Cu–O(111) were studied, and the results are presented in Figures and 13. Adsorption of CO2 onto the surface leads to the formation of a CO2 complex, identified as carbonates; in the latter complex entity, two of the C–O bonds are elongated and the third one corresponds to the C (coming from CO2) and the O from the surface (Osurf) (see Tables , 5, and S19–S21).

Table 5

Activated CO2 Molecule on CeO2(111), Ce–La–O(111), and Ce–La–Cu–O(111); Bond Length (Å) and O–C–O Angle (°) in the Absence of Oxygen Vacancies (Ov) and at the 0, 5, and −5% Strain Level

strain (%)	5%	0%		–5%
oxide system	O–C–O (deg)	O=C (Å)	O–C–O (deg)	O=C (Å)	O–C–O (deg)	O=C (Å)
CeO2(111)	126.78	1.27/1.27	130.32	1.26/1.26	178.78	1.17/1.17
Ce–La–O(111)	119.96	1.28/1.28	125.93	1.26/1.27	123.64	1.26/1.25
Ce–La–Cu–O(111)	119.88	1.28/1.27	120.04	1.27/1.27	124.75	1.27/1.27

The O–C–O angle of the adsorbed CO2 molecule is in the range of 125–128°, while the two Osurf–C–O angles involving the surface oxygen are in the range of 114–121°. The deviation of the O–C–O angle from the linear geometry of the CO2 gas phase, 180°, demonstrates CO2 activation.[92] Moreover, the difference of the Osurf–C–O1 and Osurf–C–O2 angles denotes the different extent of bonding of the O1 and O2 of CO2 with the surface. This is in agreement with the Bader CT as discussed later. In the adsorbed state of CO2, the C–O bonds present distances between 1.26 and 1.28 Å with the C–Osurf bond length being in the 1.35–1.39 Å range. Compared to the gas phase (C–O ≈ 1.16 Å), it can be concluded that C–O bonds in the adsorbed CO2 state are elongated to a great extent with a significant degree of O–C–O bond bending, as discussed earlier. These findings corroborate for activated chemisorbed CO2 species. The significant reduction of the O–C–O angle upon adsorption is mirrored to the reduction of the adsorption energy (Eads).[93]

Eads and O–C–O Angle as Descriptors of Activation

Doping ceria and accounting for the presence of Ov (Figure ) (more realistic scenario) leads to a significant Eadsvalue reduction as we move from CeO2, to Ce–La–O and Ce–La–Cu–O surface (−1.40 eV). This is accompanied by a significant reduction of the O–C–O angle value (124.71°) for the Ce–La–Cu–O surface and a C–Osurf (Å) distance of 1.34 Å. This implies a stronger CO2-surface CT in this case. Doping effect in the absence of Ov (Figure ) causes similar impact on the Eads at zero, +5, and −5% level of the imposed strain. However, the impact is more significant (higher drop of Eads as we move from CeO2 to Ce–La–O oxide). The addition of the second dopant, Cu, has minor impact in this parameter.

Strain Effects

For the Ce–La–Cu–O case, high level of compressive strain does not favor the CO2–surface interaction, as it is supported by the Osurf–CCO distances reported in Tables S19–S21 (not bound species formed). On the other hand, the tensile strain leads to favorable CO2 interaction with the surface of interest. In the case of CeO2 and Ce–La–O surfaces, as we move from the compressive field to a tensile one, Eads drops corroborating for a stronger interaction as well.

CO2–Surface Electronic Interaction

Bader charge analysis was performed only in the case of Ce–La–Cu–O(111) surface, at site 2 (Cu–Ov–La) following two scenarios, namely, the presence or absence of Ov. The presence of Ov was found to favor the electron accumulation on the CO2-derived carbon (CCO) as well as the CO2-derived oxygen atoms (OCO), corroborating for the dual role of the Ce–La–Cu–O surface participating with both basic (O–C) and acidic (M–O) sites in its interaction with the CO2 molecule. Having a closer look on the effect of strain on the CO2–surface interaction, it can be seen that tensile strain leads to the same charge distribution for the CCO and the OCO atoms. However, it seems that compressive strain enhances the basic characteristic of the surface, as CCO is more electron enriched, allowing us to speculate for more participation of the surface basic sites in the interaction with the CO2 and thus for a possible surface acidity–basicity tuning upon strain imposed. The above are also supported from the data listed in Table , where both under compressive (−5%) and tensile (5%) strain level, the O–C–O angle and the C=O bond are distorted. The lower value of the O–C–O angle under tensile (119.88°) compared to the same angle value under compressive (124.75°) allows us to discuss the strength of the interaction (strength of acid/basic sites). This is supported by the DOS (Figure B) of Ce–La–Cu–O(111) reduced surface, where in both tensile and compressive fields, the formation of new gap states can be noticed, leading to an apparent narrowing of the band gap (VB–CB splitting). The latter has been reported to have an instrumental role in the strain tuning of surface basicity, when covalent bonding is involved.[87] The new gap states have different shapes and width implying different electron distributions on the surface upon different types of strain imposed on the surface. More focused studies using near-ambient pressure XPS and extended X-ray absorption fine structure can enlighten more the role of strain on the acid/base characteristic of the surface.[94,95]

Based on the present experimental findings of in situ DRIFTS studies, the bidentate carbonates were found as predominant species on the pristine (chemically strained) and DBM (chemically and mechanically strained) surfaces, justifying the dual-site interaction of CO2 with the surface.[80] Addition of mechanical strain led to an increase in the population of bidentate carbonate species. It has been reported that the free carbonate ion (D3 symmetry) has an IR band at 1415 cm–1. Upon adsorption, lowering of the symmetry gives rise to two ν(CO) bands on both sides of the band at 1415 cm–1. The separation between the two bands is known as Δν3 splitting,[96] which is used as a descriptor of the surface basic strength. Usually, a smaller splitting accounts for stronger basic sites; for the different carbonate species, different values are reported; unidentate (Δν3 = 100 cm–1), bidentate (Δν3 = 300 cm–1), and bridged species (Δν3 = 400 cm–1). The Δν3 splitting in Figure is 268 cm–1, characteristic of bidentate species formed in moderate-strength basic sites. Bidentate carbonates are favored over the medium-strength basic sites (see Figure ); the latter are present in the Ce–La–Cu–O surface and tuned upon mechanochemical processing (ball milling) (see Figure B). The pristine and the DBM solids is another example of isostructural oxides with a different behavior upon strain. The latter is demonstrated not only by the different population of the carbonates formed but also from their different thermal stabilities. The different thermal stability reflects the strength of interaction and is justified by the different Osurf–C–O1 and Osurf–C–O2 angles (deformation) in zero and 5% strain levels.

Case Study: DRM Reaction

In order to validate the effect of mechanochemistry concepts in catalysis, the DRM reaction was selected as a probe one. The supported Ni catalysts were prepared based on the procedure mentioned earlier (Section ), where the pristine, dry ball-milled, and wet ball-milled Ce–La–Cu–O oxides were used as supports.

Figure A and Table present the experimental CH4 and CO2 integral rate values after 0.5 and 12 h of DRM reaction, as well as the amount of carbon accumulated on the catalyst surface after 12 h of DRM. In the case of the pristine-supported Ni catalyst, the conversions were lower than the equilibrium values when DRM is only considered but also for the DRM/reverse water gas shift (RWGS) reaction network (see Table ). In our previous work,[27] it was found (use of transient isotopic experiments) that the decrease of carbon deposition in the pristine-supported Ni catalyst was due to carbon oxidation (or gasification) to CO by labile lattice oxygen, where CO2 reoxidizes the reduced ceria-based support. In the case of all the catalysts listed in Table , for short time-on-stream (TOS) (ca. 0.5 h), the H2/CO ratio adopts a value slightly higher than unity, signifying the simultaneous presence of some side reactions, such as the RWGS, CH4 decomposition, carbon oxidation to CO and CO2, and steam methane reforming. The RWGS explains the drop in the H2/CO gas ratio, the increase of XCO, and the drop in the H2-yield after 12 h TOS, as shown in Table . Overall, DBM and WBM have increased the CH4 and CO2 conversions for TOS of 0.5 and 12 h to a different extent. This is linked to the modification of the Ni–support interface upon the milling process, as demonstrated through H2-TPD experiments performed over the three supported Ni catalysts (pristine, dry ball milled, wet ball milled; Figure S16). In the H2-TPD profiles of the milled catalysts (both DBM and WBM), low-temperature desorption peaks appeared, which is not the case for the pristine-supported Ni catalyst (Figure S16 and Table S22), linked with the presence of smaller size Ni crystallites grown on the milled supports. Ball milling was also found to alter the redox properties of the catalysts along with the distribution of labile oxygen species present on the surface and in the subsurface region; the latter can be inferred by the shape of the peaks (Figure S17). Additionally, DBM seems to have an opposite effect on the catalyst reducibility compared to the DBM (Table S23).

Figure 16

(A) Integral rate of CH4 conversion (mmol g–1Ni s–1) and H2/CO gas product ratio obtained after 0.5 and 12 h of DRM (20 vol % CH4/ 20% CO2/He) at 750 °C over 5 wt % Ni/Ce–La–10Cu–O catalysts as a function of support pre-treatment (pristine, WBM, and DBM). (B) Transient response curves of CO2 formation rate (μmol g–1 s–1) obtained during TPO of carbon formed after 12 h of DRM (20 vol % CH4/20 vol % CO2/He) at 750 °C over 5 wt % Ni/Ce–La–10Cu–O catalysts: (a) pristine, (b) WBM, and (c) DBM.

Table 6

Conversions of CH4 and CO2, H2-Yield (%), and H2/CO Gas Product Ratio after DRM at 750 °C (20% CO2/20% CH4/He) for 0.5 and 12 h over the 5 wt % Ni Supported on Ce–La–10Cu–O Carriers (Pristine, WBM, and DBM)a

catalyst 5 wt % Ni/Ce–La–10Cu–O	time (h)	XCH4 (%)	XCO2 (%)	H2-yield (%)	H2/CO	mg C gcat–1
pristine	0.5	80.3 (86.6)b (83.6)c	76.4 (86.6)b (90.3)c	55.3	1.05
	12	80.8	81.4	42.7	0.8	1.5
WBM	0.5	91.5	93.5	71.7	1.1
	12	88.7	91.7	57.9	1	70.7
DBM	0.5	86.8	86.1	53.33	1.1
	12	85.1	85.2	53.9	1.1	7.7

The amount of carbon accumulated (mg C g–1cat) after 12 h of DRM is also presented.

Number in parentheses gives the equilibrium conversions of CH4 and CO2 and the H2/CO product gas ratio for the DRM reaction alone; feed gas composition (20% CO2/20% CH4/He); 750 °C.

Number in parentheses gives the equilibrium conversions of CH4 and CO2 and the H2/CO product gas ratio when both the DRM and RWGS reactions participate in the reaction network; feed gas composition (20% CO2/20% CH4/He); T = 750 °C.

The carbon accumulated on the catalyst surfaces was measured by temperature-programmed oxidation (TPO) (see Figure B). It is also noteworthy to mention the two prominent peaks that appear in the TPO-CO2 trace of Ni/Ce–La–10Cu–O catalyst, ca. ∼550 and 600 °C (Figure B), indicating the formation of two types of carbon. This is in agreement with our previous findings on Ni/Ce–La–Cu–O-supported catalysts (pristine),[27] where the La-doped catalyst resulted in an asymmetric (i.e., two peaks) TPO-CO2 trace at ∼550 °C, indicating the formation of two different types of carbon. These peaks became more prominent and well defined after WBM (curve b).

Based on the above results, it can be suggested that ball milling could benefit several catalytic reactions involving ceria, such as chemical looping water splitting (CLWS) coupled with the decomposition of glycerol,[97] aqueous-phase reforming (APR) of glycerol,[98] and chemical looping steam reforming of glycerol,[99] where ball milling could induce the tuning of CO2 (as product this time) interaction with the CeO2-related surface toward the preferred direction. In the study reported by Dou et al.,[97] CeO2 acts as a promoter, attributing to stronger metal–support interactions and enhancing the thermal stability of the Ni–CeO2/MCM-41 or SBA-15 catalysts. In the work of Wu et al.,[98] the Ni–Cu bimetallic supported on mesoporous CeO2 was used for APR of glycerol (biodiesel byproduct). The kinetic analysis conducted proved that after adding CaO onto the 1Ni2Cu/CeO2 catalyst, the apparent activation energy dropped, ca. 29.86 kJ mol–1. Here, CaO is integrated as an absorbent to facilitate the WGS reaction and reduce methanation reactions via in situ CO2 removal and capture. Last but not least, Lou et al. reported on Fe–Ce–Ni–O-based oxygen carriers (OCs) for CLWS coupled with glycerol decomposition in an attempt to simultaneously produce hydrogen and syngas.[99] A remarkable redox behavior was achieved by the prepared Fe–Ce–Ni–O-based OCs, leading to a significant catalytic function of partial oxidation and decomposition of glycerol at 750 °C. The best oxygen-transfer capability and highest hydrogen and syngas production were attained by OC with a 100:10:3 molar ratio of Fe/Ce/Ni.

Conclusions

In this work, we presented the first example of how the chemical (εC) and mechanical (εM) strain can be used to tailor the CO2 activation and adsorption onto a metal oxide surface by a combined experimental modeling approach. To demonstrate this, the Ce–La–Cu–O ternary oxide surface was prepared using a microwave coupled with sol–gel and then subjected into mechanochemical treatment (ball milling). Chemical strain is originated from the CeO2 lattice doping (La3+ or La3+/Cu2+), whereas mechanical strain is originated from post-synthetic ball milling treatment. This study reveals the impact of the mechanical strain (ball milling) on the mobility of lattice oxygen (OL) through state-of-the-art 16O/18O transient isotopic exchange experiments. Strained (tensile) Ce–La–Cu–O surfaces were found to facilitate the Ov formation and the CO2 adsorption. The interplay between the Ov entities and the CO2 activation and the role of mechanical strain were illustrated through in situ DRIFTS studies and Bader charge analysis. In particular, the DBM process (mechanical strain) was found to enhance by 6 times the population of the bidentate carbonates formed, though lowering their thermal stability. Geometric characteristics of the adsorbed CO2 species were used to evaluate the molecule deformation and the strength of adsorption, such as C–O and Osurf–C bond lengths (Å), as well as O–C–O and Osurf–C–O1 and Osurf–C–O2 angles, demonstrating how the mechanical strain can assist an already chemically strained surface in its interaction with the CO2. Based on the in situ DRIFTS studies along with the Osurf–C–O1 and Osurf–C–O2 angles and the Bader charge analysis, a dual role of the Ce–La–Cu–O surface is proposed, through which the surface utilizes both its acidic and basic sites; the relative contribution of the sites in the interaction with CO2 can be tuned by the mechanical strain applied. Having the possibility to tune the participation of the acidic or basic sites in the interaction with CO2, this leads to carbonates with different stabilities; the latter is of high importance in catalytic reactions, such as DRM and CO2 hydrogenation, where the CO2 activation and carbonate formation as active intermediates are important mechanistic steps. A proof of concept is herein demonstrated using the DRM reaction, where Ni-based catalysts supported on a ball-milled carrier exhibited different (smaller) Ni crystallite sizes and enhanced labile oxygen species features that led to higher CO2 and CH4 conversions and significantly lower carbon deposition.