High-energy resolution X-ray absorption and emission spectroscopy reveals insight into unique selectivity of La-based nanoparticles for CO₂

Significance

CO₂ has become a challenge for our society and we have to develop new materials for its photo-/electrocatalysis, chemoresistive sensing, and storage. Particularly, for the variety of electrochemical applications the selective interaction of CO₂ and charge transfer with solids is in the foreground, but their origins are poorly understood. Our story will undoubtedly showcase how to access the key information, which is relevant for electrochemical application from in situ X-ray absorption spectroscopy/X-ray emission spectroscopy studies.

Abstract

The lanthanum-based materials, due to their layered structure and f-electron configuration, are relevant for electrochemical application. Particularly, La₂O₂CO₃ shows a prominent chemoresistive response to CO₂. However, surprisingly less is known about its atomic and electronic structure and electrochemically significant sites and therefore, its structure–functions relationships have yet to be established. Here we determine the position of the different constituents within the unit cell of monoclinic La₂O₂CO₃ and use this information to interpret in situ high-energy resolution fluorescence-detected (HERFD) X-ray absorption near-edge structure (XANES) and valence-to-core X-ray emission spectroscopy (vtc XES). Compared with La(OH)₃ or previously known hexagonal La₂O₂CO₃ structures, La in the monoclinic unit cell has a much lower number of neighboring oxygen atoms, which is manifested in the whiteline broadening in XANES spectra. Such a superior sensitivity to subtle changes is given by HERFD method, which is essential for in situ studying of the interaction with CO₂. Here, we study La₂O₂CO₃-based sensors in real operando conditions at 250 °C in the presence of oxygen and water vapors. We identify that the distribution of unoccupied La d-states and occupied O p- and La d-states changes during CO₂ chemoresistive sensing of La₂O₂CO₃. The correlation between these spectroscopic findings with electrical resistance measurements leads to a more comprehensive understanding of the selective adsorption at La site and may enable the design of new materials for CO₂ electrochemical applications.

CO₂ has become a challenge for our society and we have to develop new materials for its photo/electrocatalysis, chemoresistive sensing, and storage (1–8). Particularly, for the variety of electrochemical applications the selective interaction of CO₂ and charge transfer with solids is in the foreground. At the same time, the interaction of CO₂ with solids in the electrochemical cell or sensing device is rather complex, thus it remains challenging to experimentally identify the key elements determining their selectivity and efficiency. X-ray absorption spectroscopy (XAS) and X-ray emission spectroscopy (XES) provide complementary information on the electronic structure of materials (9, 10) and on the orbitals participating in the interaction with absorbing molecules (11). High-energy resolution fluorescence-detected (HERFD) XAS probes unoccupied states with a spectral resolution higher than regular XAS. Furthermore, with the same experimental setup XES can be measured, which allows one to probe the occupied states within the valence band (12). In situ HERFD XAS or XES experiments have been previously carried out to study the catalytic reaction at the surface of noble metals (11, 13–16), zeolites (17), and metal organic frameworks (18). Thus far, no such in situ experiments have been performed to directly track the changes of the electronic structure of a solid and its electrochemical activity toward CO₂. The rare-earth–based materials like perovskites and oxycarbonates, owing to their unique f-electron configuration of Ln (Ln = rare earth) and layered crystal structure, emerge as the most interesting for future photo- and electrochemical applications (3–8). Among rare-earth oxycarbonates (19, 20), particularly lanthanum strongly responds to CO₂ and shows up to 16-fold conductivity changes, not seen before for any metal oxides (21). This is very surprising because a direct injection of an electron into CO₂ molecule requires the activation energy of nearly 2 eV (22). To assess the origins of the unique CO₂ sensitivity of rare-earth oxycarbonate, it is essential to study in situ the interplay between the changes of the electronic structure of La-based nanoparticles upon CO₂ adsorption and changes of the macroscopic conductivity of a device.

Here, to elucidate the underlying mechanism we first determine the structure and atomic positions of the lanthanum oxycarbonate. Using HERFD XAS and valence-to-core (vtc) XES results, we gain information about the electronic structure and band gap. Moreover, we combine in situ HERFD XAS and XES measurements with sensing performance tests to obtain the structure–function relationship. Finally, with all of the obtained information we discuss a mechanism of CO₂ adsorption on the La₂O₂CO₃ surface.

Results and Discussion

The synthesis and characterization of La₂O₂CO₃ nanoparticles are described in SI Experimental Methods, Synthesis and Characterization, including Figs. S1–S3. The crystallite size is between 11 and 14 nm as shown in the high resolution transmission electron microscopy (HRTEM) images in Fig. 1 A and B and Fig. S1A. Three different polymorphs of rare-earth oxycarbonates are known: a tetragonal type I, the monoclinic type Ia, and the hexagonal type II. All these types share a layered structure with alternating Ln₂O₂²⁺ and CO₃²⁻ layers (23, 24). In the hexagonal type this layer consists of distorted LnO₈ coordinated rhombohedra (19). Whereas the local coordination around the Ln atom of type II polymorph is known (25), the atomic positions within the type Ia unit cell are not resolved yet. Here, we assign the powder X-ray diffraction (PXRD) peaks solely to the Ia polymorph of La₂O₂CO₃ as shown in Fig. 1C. The La₂O₂CO₃ nanoparticles crystallize in the space group P2₁/c with the refined lattice parameters a = 4.0756 Å, b = 13.4890 Å, c = 5.8034 Å, α = γ= 90°, β = 135.37°, as deduced from the Rietveld analysis of the PXRD (26–35). The lattice parameters and the atomic positions of La, C, and O in the La₂O₂CO₃ lattice and a detailed refinement procedure are given in SI Experimental Methods, Structure Determination, and Table S1. The analysis of the atomic positions within the structure helps us to identify local structural commonalities and differences between the monoclinic and the hexagonal polymorph. The Ln₂O₂²⁺ layer in the hexagonal type II consists of distorted LnO₈ coordinated rhombohedra (19), whereas the same layer in the type Ia polymorph contains pyramidal-oriented oxygen with a Ln atom in the middle as shown in Fig. 1D. The pyramids are pointing in positive or negative b direction and alternate along the c direction (Fig. S4). Within a distance (R) of 3 Å the lanthanum atom has six oxygen neighbors in type Ia, whereas La has eight oxygen neighbors in type II. Both crystal structure and the coordination of the metal cations and the oxygen anions are in general known to have strong impact on the chemical reactivity of rare-earth–based nanoparticles (36). However, reports correlating the structure of La₂O₂CO₃ with chemical reactivity are relatively sparse. For example, the crystal structure of intermediate La₂O₂CO₃ determines the efficiency of methane reforming to produce hydrogen (37, 38). Additionally, it was previously shown that a La₂O₂CO₃-based sensor shows a higher sensor signal than Nd₂O₂CO₃ (19, 21). No further information was given about the origin of these differences, but remarkably Nd oxycarbonate crystallizes in the hexagonal type II, whereas La oxycarbonate crystallizes in monoclinic type Ia structure. The formation of Ln₂O₂CO₃ from Ln(OH)₃ is a sequential process involving compositional and structural modifications. The heating rate, temperature, and CO₂ concentration determine the free energy of transformations and thus the occurrence of the individual polymorphs.

Fig. S1.

Additional PXRD and ATR-IR measurements of La₂O₂CO₃ and La(OH)₃ nanoparticles. (A) The PXRD pattern of La₂O₂CO₃ after 2 h of heat treatment at 500 °C in air. The red vertical lines represent the reference pattern for monoclinic type Ia La₂O₂CO₃, the blue and green lines show the reference pattern for the hexagonal (type II, ICSD: 37–0804) and tetragonal (type I, ICSD: 23–0320) polymorph. The formation of Ln₂O₂CO₃ from Ln(OH)₃ is a sequential process involving compositional and structural modifications. The heating rate, final temperature, and CO₂ concentration determine the free energy of transformations and thus the occurrence of the individual polymorphs. (B) The PXRD patterns of the as-synthesized La(OH)₃ nanoparticles. (C) The ATR-IR spectrum of La₂O₂CO₃ with the characteristic absorption bands.

Fig. S3.

Cross-section view of a La₂O₂CO₃ layer.

Fig. 1.

Structure and morphology of the La₂O₂CO₃ nanoparticles. (A and B) HRTEM images of La₂O₂CO₃ nanoparticles at different magnifications. (B, Inset) Fourier transform of the region highlighted in red including the zonal axis and indexed reflections. (C) Recorded PXRD pattern of monoclinic La₂O₂CO₃ (black line), the calculated (red line), and the difference (blue line) resulting from the Rietveld analysis. Reference patterns of the three polymorphs are shown [monoclinic, hexagonal (Inorganic Crystal Structure Database [ICSD]: 37–0804), tetragonal (ICSD: 23–0320)]. (D) A model of the refined La₂O₂CO₃. The lanthanum atoms are cyan, oxygen red, and carbon yellow. The polyhedrons denote the local La surrounded by pyramidal-oriented O atoms.

Table S1.

Lattice, atomic, and refinement parameters of La₂O₂CO₃

Lattice parameters, Å	Atomic parameters					Refinement parameters
	Atom	x/a	y/b	z/c	Occ
a = 4.0755(5)	C	0	0	0	1	Rwp = 6.039
b = 13.4890(2)	La	0.9872(20)	0.8370(3)	0.4762(14)	1	Rexp = 3.186
c = 5.8033(8)	O1	0	0	0.5	1	χ2 = 3.593
β = 135.374(5)	O2	0.5310(13)	0.7376(33)	0.9750(79)	1
V = 224.115(6)	O3	0.9200(13)	0.9161(31)	0.9435(82)	1

The procedure to obtain the values involved a transformation from a nonstandard to a standard space group, structure determination, and Rietveld refinement.

Fig. S4.

Local structure of La atoms in the La₂O₂CO₃. The La atoms are surrounded in a pyramidal configuration. The pyramids point in positive or negative b direction and this pointing direction changes alternately in the c direction.

We measured HERFD XAS (Fig. 2A) to obtain information about the unoccupied states and vtc XES to retrieve information about the occupied states (Fig. 2B) of La₂O₂CO₃. The large lifetime broadening at the 2p level [∼3.4 eV (39)] renders the XAS technique at the La L₃ edge rather insensitive to the electronic structure. Especially, the preedge structure of the La L₃ XAS spectrum is very difficult to resolve with conventional XAS experiment. During the HERFD measurements, the X-ray emission spectrometer is tuned to the maximum of the Lα₁ (3d_5/2–2p_3/2) transition and the absorption is recorded by monitoring the maximum of the Lα₁ intensity as a function of the incident energy. The advantage of such a setup is that the width of the spectral features is no longer limited by the 2p_3/2 core-hole lifetime but by the sharper 3d_5/2 core-hole width in the final state ∼0.7 eV (40), which is on the same order of magnitude as the experimental broadening. To test the sensitivity of the HERFD XAS method for the oxygen coordination of La, we compare the spectrum of lanthanum oxycarbonates (six oxygen neighbors) with lanthanum hydroxide (nine oxygen neighbors), as shown in Fig. 2A. The experimental data show the maximum absorption intensity located at the same energy, which corresponds to a La³⁺ ion as it is expected from the stoichiometric formula. There are two important discrepancies between the spectra of these two compounds; La₂O₂CO₃ exhibits a wider whiteline, whereas La(OH)₃ exhibits a higher whiteline intensity. These observations imply that the charge distribution is more localized around the La ion in the hydroxide than in oxycarbonate, whereas the formal charge stays three plus. The spectra at energies above the absorption edge reveal more structural features for the hydroxide than for the oxycarbonate. To understand the differences in the experimental data, we calculate XAS spectra with the FEFF program package as shown in Fig. 2A, dashed lines. The shape of the whiteline as well as the preedge structure are reproduced for both compounds, whereas the postedge features are not well modeled in the calculations. Furthermore, the calculation reproduces the differences in the whiteline width and intensity well. The atomic orbital angular momentum projected density of states (DOS) of both compounds reveal the electronic states that give rise to the preedge and the whiteline excitations and are plotted in Fig. 2 C and D for La(OH)₃ and La₂O₂CO₃, respectively. The width of the whiteline and therefore the width of the unoccupied d-states reflects the number of oxygen neighbors within the first shell around the La absorber (41, 42). The higher the coordination number of La, the narrower is the whiteline. Also the crystal structure may induce similar changes of the spectral features. Fig. 2 E and F shows atoms within a sphere of 3 Å for La(OH)₃ and La₂O₂CO₃, respectively. We count nine nearest oxygen neighbors for La(OH)₃ whereas La in La₂O₂CO₃ has only six nearest oxygen neighbors. Band structure calculations of the electronic states of La₂O₃ with its seven La–O pairs within the sphere of 3 Å show also a wide distribution of empty d-states (43, 44). In the spectra of La(OH)₃ and La₂O₂CO₃ we observe a preedge feature at 6.7 eV below the maximum of the La L₃ absorption edge, which is generally attributed to the mixed dipole and quadrupole transitions (40, 45, 46). Only systems with inversion symmetry at the absorbing atom can have a preedge structure of pure quadrupole contribution, but if the inversion symmetry is broken the quadrupole and dipole transitions contribute to the preedge intensity (47). The compounds are not having an inversion center at La atom and thus we observe mixed d- and f-states in the preedge structure. The analysis of the preedge DOS reveals a small contribution of dipole transitions of d-state character and strong contribution of quadrupole transition of f-state character. However, the intensity of quadrupole transitions is considerably smaller than the intensity of dipole transitions, thus the measured preedge feature is very small (45).

Fig. 2.

Ex situ HERFD XAS and vtc XES experiments along with FEFF calculations and coordination of La atom for La(OH)₃ and La₂O₂CO₃. (A) A comparison of the experimental HERFD XAS spectra (solid lines) with the FEFF-calculated spectra (dashed lines): La(OH)₃ (black) and La₂O₂CO₃ (red). (B) Comparison of measured vtc XES spectra of La(OH)₃ and La₂O₂CO₃. We observe differences at the feature at −10 eV and at the features at −50 eV and −60 eV, which are magnified in the inset. We assume the first feature is caused by La d- and O p-states, the second feature by La p- and O s-states, and the feature at −40 eV by La s-states. (C) La(OH)₃ and (D) La₂O₂CO₃ DOS of the La absorber atom. The width of d-DOS width and therefore the whiteline width relates to the amount of oxygen neighbors of the La atom, which are shown in E and F as cluster of LaO_x bonds in lanthanum hydroxide and oxycarbonate. The cutoff length of the sphere around the La atom (R) was set to 3 Å. The lower coordinated La atom in La₂O₂CO₃ exhibits a broader distribution of unoccupied d-states than the ninefold coordinated La(OH)₃.

To obtain information about the occupied states we perform resonantly excited XES measurements. We therefore tune the incident photon energy to the maximum of the absorption edge and detect emission spectra with energy differences from 0 to −70 eV with respect to the incident energy and probe therefore the highest occupied states. Fig. 2B shows the experimental XES spectra of La(OH)₃ and La₂O₂CO₃. The main features for both compounds are very similar and located at energies at −10 eV and at −40 eV. Interestingly, even at lower transfer energies, 50 and 60 eV below the whiteline maximum some features are detectable, which show some small differences in relative intensity between the hydroxide and the oxycarbonate. The most pronounced differences are visible in the features which correspond to the highest occupied levels. Furthermore, the shape of the structure at −10 eV also differs. For spectral features at energies close to the Fermi level at −10 eV we assume, based on earlier calculations of La(OH)₃ and La₂O₃, a mixing of La d-states and O p-states (43, 44, 48–50). We observe a feature at −25 eV, which we can assign to a mixing of La p- and O s-states on the basis of calculations on La₂Ti₂O₇ (51). The feature at −40 eV was previously observed in X-ray photoelectron spectroscopy spectra of La(OH)₃ and La₂O₃, and was assigned to La 5s states. This feature is showing an additional shoulder, which is assigned to a difference in charge transfer from the ligand in La(OH)₃ and La₂O₃. In XES spectra of La(OH)₃ and La₂O₂CO₃ these differences are also clearly visible, as shown in Fig. 2B (Inset). Within La(OH)₃ and La₂O₂CO₃ the formal valence of lanthanum ions is La³⁺ and the d-states are empty in the ground state. In both cases occupied La d-states exist and they form, together with O p-states, the valence band. The La d-states are only partly occupied and the empty fraction of the La d-states forms the conduction band. The valence and conduction band are separated by an electronic band gap E_g. Combining XAS and XES provides a gap in the d-DOS integrated over all directions of momentum transfer and over a range of modulus over momentum transfer that is given by the experimental geometry (52–54). Here we determine electronic d-DOS band gap energies of 5.1 and 3.7 eV for La(OH)₃ and La₂O₂CO₃ nanoparticles, respectively. These values are lower with respect to the optical band gap values derived from UV-Vis spectroscopy with the assumption of an indirect band gap; 5.4–5.6 eV and 4.35 eV for micrometer-large La(OH)₃ and La₂O₂CO₃ particles, respectively (55, 56). However, the direct comparison is not straightforward because UV-Vis probes the DOS that is projected according to selection rules for the optical transition. At this wavelength the momentum of a photon is very small compared with the wave vector of the electrons and thus most of the UV-Vis signal arises from direct transitions between valence and conduction band, whereas the indirect transitions are less pronounced.

The chemoresistive sensing principle involves interaction of CO₂ molecules with sensing materials, which then leads to conductivity changes. In real applications, the interaction is not an isolated CO₂ reaction but it is altered by the presence of oxygen and humidity. For example, in situ diffuse reflectance infrared Fourier-transformed spectroscopy studies revealed for Nd₂O₂CO₃-based sensor a correlation between the amount of surface carbonates groups and adsorbed hydroxyl groups; the higher the amount of adsorbed carbonates the lower the amount of water-related species on the surface. However, it was suggested that the two gases CO₂ and H₂O do not compete for the same adsorption sites (19). A similar reaction mechanism has been postulated for La₂O₂CO₃-based sensors, but so far the sites for CO₂ adsorption have not been identified (21). Therefore, we are particularly interested in determining the reaction which induces the charge transfer between La₂O₂CO₃ and CO₂ in humid air, and as a result decreases the overall conductivity of the La₂O₂CO₃-based sensor. Before we discuss the in situ spectroscopic data we briefly summarize the sensing characteristic of La₂O₂CO₃-based sensors and the impact the incident X-rays thereof. The resistance of the La₂O₂CO₃-based sensor decreases, and upon exposure to CO₂ or CO increases as shown in Fig. 3A and Fig. S5 A and B. Moreover, we observe the positive cross-sensitivity to CO₂ and humidity: The baseline resistance of La₂O₂CO₃-based sensor at 250 °C is decreasing with increasing humidity, and exposure of CO₂ in humid conditions results in higher relative resistance changes than in dry air, as shown in Fig. S5C. We define the sensor signal according to the definition for p-type semiconductors as the resistance of the sensor upon exposure to reducing gas (R_CO or R_CO2) over the baseline resistance of the sensor before exposure (R₀) (57).

Fig. 3.

Overview on the CO₂ sensing performance of La₂O₂CO₃-based sensor with and without X-ray irradiation. (A) Sensing performance upon exposure to CO₂ from 250 to 10,000 ppm in 50% rh at 250 °C without X-ray irradiation. (B) Comparison of the sensor signal upon exposure to 10,000 ppm of CO₂ without (blue) and with X-ray irradiation (black). The sensor signal is defined as R_CO2/R_air, where R_air and R_CO2 are the resistance of the sensor in air and under CO₂ exposure, respectively. Resistance under X-ray irradiation is energy dependent, thus we use as R_air the initial value before the X-ray irradiation. During 0–3 h under air, we first measure two sets of 15 X-ray absorption near-edge structure (XANES) scans each. Between the individual scans the fast X-rays shutter is closed (X-rays off) and open (X-rays on), which is visible as a spike in the sensor signal. Finally, we measure only the resistance for 20 min. We repeat the same data acquisition protocol for the measurements under CO₂. (C) Comparison of the simultaneously recorded XANES spectra (red) and the sensor signal (black) illustrates the dependence of resistance of the sensor on the incident energy. XANES and the sensor signal measurements were measured with different temporal resolution. One measurement point of the sensor signal corresponds to 17 points at the energy scale of XANES scan (1.7 eV). (D) The simplified model for the resistance of the sensor at different experimental conditions shown in C. We assume that at 250 °C the Debye length (L_D) is smaller than the radius (r) of La₂O₂CO₃ nanoparticles. Gases adsorbing at the surface create a depletion/accumulation layer with the thickness of L_D. The resistance of this layer (R_s) is different from the resistance of bulk of nanoparticles (R_b). The charge carriers have to overcome an energy barrier (eV_s) to reach the neighboring nanoparticle. Dependent on the conditions the height of the barrier is changing as schematically shown (57); the escape depth of photoelectrons emitted from the absorbing atom is larger than the size of nanoparticles (58), which means that X-rays equally change the resistance at the surface as well as in the bulk of La₂O₂CO₃ nanoparticles. The colors represent the measured resistance.

Fig. S5.

Dependence of the sensor resistance toward different gases. The resistance (A) and the sensor signal (B) changes upon CO exposure give additional information about the sensing behavior. The experiment was conducted at 250 °C and 50% rh. The increase of resistance upon CO exposure shows the p-type semiconductor behavior of the La₂O₂CO₃ layer. (C) The resistance changes with the rh in air. The lower the rh the higher the baseline resistance rises. Additionally, the sensor signal to CO₂ decreases with decreasing humidity.

For the in situ XAS and XES experiments we used conditions resulting in the highest sensor signal, which is a pulse of 10,000 ppm of CO₂ in synthetic air with 50% relative humidity (rh) and at operating temperature of 250 °C. The corresponding changes in the sensor signal during in situ HERFD XAS measurements under air and CO₂ pulse are shown in Fig. 3 B and C. At a given condition two sets of 15 spectra each were recorded. The labels “X-rays on/off” indicate that the shutter was open or closed, respectively. The incident beam even at energies below the edge causes a rapid decrease of the sensor’s resistance of about two orders of magnitude, which results in sensor signal of more than 150 as shown in Fig. 3C. During an XAS scan the resistance shows a dependence on the incident X-rays, which indicates that electrons are escaping from the La₂O₂CO₃ layer as a result of absorption of an X-ray photon. At an energy corresponding to the maximum of whiteline intensity the sensor signal increases even further to almost 400. The X-ray–induced changes of resistance have been previously used to record the total electron yield-XAS spectra (58). However, for the wide-band semiconductors and insulating materials, the charging of the sample has to be experimentally or mathematically compensated by taking into account the effective penetration depths of secondary electrons as a function of incident X-ray energy (58–61), and thus we will not discuss it any further. Instead, we analyze the impact of the incident X-rays on the reactivity of La₂O₂CO₃ toward CO₂. We observe that between the XAS scans, when the X-ray shutter is shortly closed, the resistance/sensor signal only partially recovers as shown in Fig. 3 B and C; but if enough time is allowed, it fully recovers to the initial value as shown in Fig. S6. In summary, the resistance of the sensor increases 16-fold upon exposure to 10,000 ppm of CO₂; the resistance of the sensor decreases almost 400-fold upon irradiation with X-rays and depends on the energy of incident beam as shown in Fig. 3 B and C. The overall resistance of the sensor exposed to CO₂ under X-ray irradiation is clearly higher than in air, as shown in Fig. S6. The direct quantitative comparison of sensor signal toward CO₂ with and without X-ray irradiation is not possible. We assume that the mechanism underlying the chemical reactivity of La₂O₂CO₃ toward CO₂ is not affected by incident X-rays but only the transduction of the chemical reaction into the electrical signal, and thus we can qualitatively compare resistivity changes as a function of gas environment under exposure to X-rays. Additionally, the simplified model for the resistance of the sensor at different experimental conditions is given in Fig. 3D. The XAS and XES spectra of films at 250 °C show identical features as spectra of pellets at room temperature as shown in Figs. 2 A and B and 4, respectively. Even though the sensor’s resistance upon exposure from dry air to 50% rh decreases 50-fold, the XAS spectra of the sensor measured in dry and 50% rh air are identical as shown in Fig. S7A. Thus, we conclude that the La is not an adsorption site for water. Upon exposure to 10,000 ppm of CO₂ in 50% rh, we notice in the XAS spectra an increase of the intensity and sharpening of the whiteline, whereas other pre- and postedge features do not change, as shown in Fig. S7B. These, by analogy to previous HERFD XAS studies on Ce (62), point to the oxidative character of CO₂ adsorption on La₂O₂CO₃, namely CO₂ molecules are acting as electron acceptors. We ascribe the changes of the whiteline to the presence of additional oxygen in the direct vicinity of La. The in situ vtc XES measurements further confirm that CO₂ interacts directly with La sites and the electronic structure is rearranged because of an additional ligand, Fig. S7C; We observe an increase of the states at −40 eV and a decrease of the feature at −10 eV, which we assume to consist of La d- and O p-states. We conclude that CO₂ adsorbed at La₂O₂CO₃ as surface carbonates, as schematically depicted in Fig. 4. Remarkably, the XAS studies reveal an oxidative character, whereas the resistance measurements point at reducing character of CO₂ adsorption at La₂O₂CO₃ in humid conditions. We note that XAS selectively probes the adsorption at La site, whereas the resistance of the sensor is not selective to particular reaction or site, but instead probes the net charge transfer between molecules (O₂, CO₂, H₂O) and surface of a solid (57). This further underlines the advantage of in situ XAS/XES studies for investigating the reactions mechanism in real conditions.

Fig. S6.

Resistance measurement of the La₂O₂CO₃-based sensor during acquisition of series of in situ XAS and XES scans. For the sake of completeness the concentration of CO₂, scan type, recalibration of the beam, and refill are indicated.

Fig. 4.

In situ studies of structure–functions relationship. (A) The sensor signal upon exposure for 3 h to 10,000 ppm of CO₂. (B) The schematic of unit cell of La₂O₂CO₃ illustrating the layered structure and the sixfold coordination of La atoms. La, O, and C atoms are labeled cyan, red, and yellow, respectively. (C) In situ XAS and XES spectra before (red) and after exposure to 10,000 ppm of CO₂ (yellow) in 50% rh at 250 °C. To underline the changes the area under the experimental curves was colored. We observe a whiteline sharpening, a decrease of the feature at −10 eV, and at the same time an increase at the feature at −50 eV. This points at the oxidative character of CO₂ and additional oxygen in the La neighborhood. (Inset) Schematic of possible carbonate species forming at the surface of La₂O₂CO₃.

Fig. S7.

In situ spectra of La₂O₂CO₃ under various conditions. (A) Dependence of the HERFD XAS signal to the rh. Each spectrum is an average of nine spectra. (Inset) Whiteline maximum. The rh changes from 0% (green line) to over 20% (red line) to 50% (black line) and no difference is observable. (B) A comparison of the HERFD XAS spectra in air with 50% rh (black line) and additional 10,000 ppm CO₂ (red line). (C) vtc XES spectra in air with 50% rh (black line) and additional 10,000 ppm CO₂ (red line).

SI Experimental Methods

Synthesis and Characterization.

Chemicals.

All chemicals were used without further purification. Lanthanum isopropoxide [La(OiPr)] (99%) was purchased from Strem Chemicals and acetophenone from Fluka Analytical.

Synthesis.

The synthesis was prepared in an oxygen- and water-free atmosphere in a glovebox (O₂ and H₂O < 0.1 ppm). The synthesis of La(OH)₃ nanoparticles is adopted from a previously published synthesis route (19, 21). La(OiPr) (158.08 mg) was added to acetophenone (4.68 mL) in a reaction vessel containing a stirring magnet. The vessel was sealed with a Teflon cap, taken out of the glovebox, and heated at 200 °C for 20 min in a CEM Discover microwave reactor with high stirring rate. The synthesized particles were separated from the reaction liquid by centrifugation and later washed with ethanol and acetone. The powder was finally dried in air at 60 °C. Lanthanum oxycarbonate was prepared by heating the synthesized sample for 2 h at 500 °C in a furnace under air with a heating ramp of 5 °C/min. The final product was cooled down to room temperature at ambient conditions. Both compounds were characterized by PXRD and HRTEM; additionally, La₂O₂CO₃ was characterized with ATR-IR, as shown in Figs. S1 and S2. Information about the characterization measurements can be found below.

Fig. S2.

TEM images of the different nanoparticles. (A) Overview and (B) HRTEM images of the as-prepared La(OH)₃ nanoparticles. The average crystallite size is 8 nm. (C) Overview and (D) HRTEM images of La₂O₂CO₃ nanoparticles, which were annealed for 60 h at 250 °C to perform Rietveld refinement.

Discussion on the occurrence of different polymorphs for La- and Nd oxycarbonates.

Although the starting material is in both cases the corresponding hydroxide, the heat treatment is crucially different. Nd(OH)₃ is heated in air to form an amorphous compound, which is later on transformed to hexagonal Nd₂O₂CO₃ through exposure of 5,000 ppm CO₂ in humid air at elevated temperatures (19). Here, we anneal La(OH)₃ for 2 h at 500 °C in air, which contains about 400 ppm of CO₂, and we obtain phase-pure type Ia La₂O₂CO₃, which is considered as metastable (24). The heating rate, final temperature, and CO₂ concentration determine the free energy of transformations and thus the occurrence of the individual polymorphs.

PXRD.

The PXRD data for structure refinement were collected with a Panalytical X’Pert-Pro diffractometer equipped with an X’Celerator detector using Cu Kα radiation. The step size was set to 0.0334° and the counting time was 280 s per point. The powder sample was mounted in an AP HTK-1200 oven and the incident antiscatter and divergence slit were set to 1/2° and 1/4°, respectively; on the diffracted side a 5-mm antiscatter slit was used together with a β-Ni filter. On both sides 0.04-rad Soller slits were installed.

ATR-IR.

IR analysis was performed with a Bruker ALPHA-P FTIR spectrometer with a resolution of 4 cm⁻¹ and averaged over 128 scans.

TEM.

TEM measurements were conducted on a Philips Tecnai F30 operated at 300 kV. A Gatan 1k CCD chip was used for image acquisition via the software Digital Micrograph.

Sensor fabrication.

About 10 mg of the washed, but still wet La(OH)₃ powder was dispersed in 4 mL of ethanol and dip-coated on an alumina substrate. The substrate was equipped with the interdigitated platinum electrodes for the resistance readout on the front side and on the back side with the platinum heaters. This substrate was dip-coated into the prepared dispersion 60 times, with a holding time of 5 s within the dispersion, 20 s for drying in air, and a moving velocity of 300 mm/min. The sensor was afterward heated under the above-mentioned conditions to obtain highly porous La₂O₂CO₃ films as shown in Fig. S3.

Structure Determination.

Structure solution.

The PXRD pattern is indexed on a primitive monoclinic unit cell (a = 4.0803 Å, b = 13.5090 Å, c = 4.0720 Å, β = 90.97°, ICSD: 48–1113), using the program TREOR (26) implemented in the software CMPR (27). A careful examination of the diffraction pattern indicates that h0l reflections with h + l = 2n + 1, and 0k0 reflections with k = 2n + 1 are systematically absent, so the most probable space group is expected to be P2₁/n. As P2₁/n is not a standard space group, we thus convert it to the corresponding standard space group P2₁/c (a = 4.0803 Å, b = 13.5090 Å, c = 5.8131 Å, β = 135.54°) using cctbx (35). We perform structure solution using modified direct method implemented in the software package EXPO2013 (28). After subtraction of the background using polynomial functions in selected angular intervals, the intensity of each individual reflection is extracted by using Pearson profile functions. Several initial structure models are generated, but all of them exhibit some structural similarities. Therefore, we select the best solution proposed by the program for further analysis.

Rietveld refinement.

We use an initial structure model as a starting point for Rietveld refinement (29) using the program GSAS (30, 31). We use the shifted Chebyshev function to fit the background, and the peak shape function to fit the patterns is the pseudo-Voigt function (32). We apply the surface roughness correction suggested by Suortti to minimize the intensity error introduced in Bragg–Brentano geometry (33). The preferred orientation correction was performed using Spherical Harmonics (34). The structure model was finally converged with R_wp = 6.039 (R_exp = 3.186).

The results are presented in Table S1.

Gas-sensing measurements.

The gas-sensing measurements were performed in a test chamber made out of Teflon, equipped with the electrodes for resistance measurements and heating of sensing layer. The atmosphere within the chamber was adjusted by a gas-mixing system. Different CO₂ pulses ranging from 250 to 10,000 ppm were added to humid air (50% rh) and the resistance of the active material was measured with a Keithley model EMM 617 electrometer.

X-ray spectroscopy.

The HERFD XAS and vtc XES experiments were carried out at the ESRF in Grenoble, France, at the beamline ID 26. The incident energy was selected using the (311) reflection of a double Si crystal monochromator. The incident beam had a flux on the order of 10¹³ photons per second at the sample position. The emission lines were selected with an X-ray emission spectrometer (65, 66) and detected with an avalanche photodiode. In the HERFD XAS experiments the energy of the incident beam was scanned over the La L₃ edge from 5,475.0 to 5,575.0 eV with a step size of 0.1 eV and the emission spectrometer was tuned to the maximum of La Lα₁ fluorescence line at 4,649.9 eV selected by the (400) reflection of one spherically bent Ge crystal analyzer. In vtc XES the incident beam energy was fixed to 5,490.8 eV and the emission was recorded between 5,415.0 and 5,492.0 eV with a step size of 0.4 eV with five spherically bent Ge crystal analyzers using the (422) Bragg reflection.

Ex situ experiments.

During the ex situ experiments pellets pressed out of pure La(OH)₃ and La₂O₂CO₃ nanoparticles were measured by means of HERFD XAS and vtc XES.

In situ experiments.

The sensor was placed in a gas-sensing chamber equipped with a pair of contacts for the electrical resistance measurements, a pair of contacts for applying voltages to the heater, Kapton window for XAS measurements in fluorescence mode, gas inlet, and gas outlet; further technical details are described elsewhere (16). In a typical HERFD XAS experiment we acquired 30 scans for 3 h at a given condition. The sensor temperature was kept at 250 °C.

FEFF simulations.

For the calculation of the HERFD XAS and vtc XES spectra the full multiple scattering calculation program package FEFF9 was used (67, 68). The absorption and emission spectra were calculated using the COREHOLE card, and dipole and quadrupole transitions were included in the calculations with the MULTIPOLE card. The core-hole lifetime broadening was reduced by 1 eV with the EXCHANGE card and the Hedin–Lundquist potential was used. In the La₂O₂CO₃ calculations the Fermi energy was upshifted by 1 eV. The UNFREEZEF card was activated to include the f-electrons into the self-consistent potential calculations. A cluster radius of 4 Å for the self-consistent potential was used for both materials. The full multiple scattering sphere was set to 4 Å and to 4.5 Å for La(OH)₃ and La₂O₂CO₃, respectively.

Conclusions

Here, we present complementary X-ray diffraction, absorption, and emission studies of lanthanum oxycarbonate nanoparticles. To calculate the accurate XAS spectra with FEFF code a precise determination of atomic positions is required. Using the PXRD pattern we determine the atomic structure of monoclinic type Ia La₂O₂CO₃. Eventually, this information is the foundation for the FEFF calculations, which reproduce the measured XAS spectrum very well. Moreover, we experimentally verify that the La d-states are partially occupied and the empty fraction forms the conduction band, the occupied fraction forms together with O p-states the valence bands, and the electronic d-DOS band gap between them is 3.7 eV. To further elucidate the role of the oxycarbonate ligands an advanced density functional theory calculation of XES spectrum is needed (12). Herein the combination of HERFD XAS and vtc XES techniques allows us to in situ visualize the charge transfer between relatively inert carbon dioxide and La₂O₂CO₃. It reveals changes of both occupied (La d and O p) and unoccupied states (La d) upon the interaction with CO₂; this information is not accessible with other techniques. The CO₂ sensing mechanism in humid air is highly complex; nevertheless, our in situ results show that La is the adsorption site for carbon dioxide but not for water. In the future, the soft-range XAS/XES studies at O and C edges could possibly answer the question concerning sites for water and oxygen adsorption (63, 64), however such an experiment would be very challenging because both elements are present in La₂O₂CO₃ as well as in the ambient atmosphere.

Experimental Methods

Synthesis.

All reagents were used without further purification. La(OH)₃ nanoparticles were synthesized in a microwave reactor through heating lanthanum isopropoxide in acetophenone at 200 °C for 20 min. Afterward the samples were washed and dried in air. La₂O₂CO₃ was prepared by heating this sample at 500 °C for 2 h under air.

Structure Determination.

The PXRD of type Ia La₂O₂CO₃ was indexed using TREOR (26) implemented in CMPR (27). The space group was adjusted (35) and the structure solved with EXPO2013 software (28). The lattice parameters were refined with GSAS (29–34).

X-Ray Spectroscopy.

HERFD XAS and vtc XES experiments were carried out at ID26 at the European Synchrotron Radiation Facility (ESRF), which is equipped with an X-ray emission spectrometer (65, 66). For the HERFD XAS experiments the incident energy was scanned over the La L₃ edge and the La Lα₁ fluorescence line was recorded. In the vtc XES experiments the incident energy was fixed to the maximum of the whiteline energy and the spectrometer recorded emission just below this incident energy. Powders were pressed in pellets or directly deposited on the alumina substrate, which was equipped with Pt electrodes. The sensors were measured under working conditions in humid air at 250 °C with pulses of 10,000 ppm CO₂ in an in situ cell (16).

FEFF Calculations.

The spectra were calculated with the FEFF9 program package (67, 68). The calculations for the self-consisting potential were performed including atoms within a sphere of 4 Å. For the full multiple scattering calculations, atoms within a sphere of 4 Å and 4.5 Å for La(OH)₃ and for La₂O₂CO₃, respectively, were included.

Characterization Techniques.

Phase composition and phase purity were investigated with PXRD on a Panalytical diffractometer and FTIR on a Bruker attenuated total reflection infrared (ATR-IR) spectrometer. HRTEM micrographs were recorded with a Phillips Tecnai F30 electron microscope.

Detailed information about the synthesis, gas-sensing conditions, the X-ray spectroscopy measurements, the FEFF calculations, and sample characterization techniques are described in SI Experimental Methods.