Manipulating protons and oxygen vacancies in nickelate oxides via thermochemical dehydration†

Haowen Chen‡ ^ab, Zihan Xu‡ ^ab, Luhan Wei ^b, Mingdong Dong ^c, Yang Hu ^b, Ying Lu ^b, Nian Zhang ^d, Jie Wu ^c and Qiyang Lu *^be
aZhejiang University, Hangzhou 310027, China
^bSchool of Engineering, Westlake University, Hangzhou 310030, China. E-mail: luqiyang@westlake.edu.cn
^cKey Laboratory for Quantum Materials of Zhejiang Province, School of Science, Westlake University, Hangzhou 310030, China
^dShanghai Synchrotron Radiation Facility, Shanghai Advanced Research Institute, Chinese Academy of Sciences, Shanghai 201204, China
^eResearch Center for Industries of the Future, Westlake University, Hangzhou 310030, Zhejiang, China

First published on 13th August 2024

---

---

Introduction

Ionic defects are important building blocks for tuning the physical and chemical properties of nickelates with a perovskite structure (RNiO₃, R = rare earth or alkaline metals), including transport properties,^1–4 magnetism,^5–7 optical properties^8,9 and (electro)catalytic activity.^10,11 Mainly, two ionic defects can be incorporated into nickelates, i.e., protons (H⁺) and oxygen vacancies (V_O), and both defects show a profound impact on the crystal structure and electronic structure of perovskite nickelates.^12,13 Therefore, searching methods to precisely tune ionic defect concentration is highly desirable for designing the properties of nickelates and other related functional oxides, which is otherwise difficult to achieve through alternative strategies. For example, tuning the oxygen non-stoichiometry δ in perovskite nickelates RNiO_3−δ from δ = 0 (i.e., RNiO₃) to δ = 1 (i.e., RNiO₂) can effectively reduce the Ni valence state from Ni³⁺ to Ni¹⁺. The rarely obtained Ni¹⁺ valence state has eventually led to the discovery of unusual superconductivity in strontium-doped infinite-layer nickelates (Nd_0.8Sr_0.2NiO₂).¹⁴ In contrast, protonation of nickelates induces Ni valence state change from Ni³⁺ to Ni²⁺ (in fully protonated HNdNiO₃), which was shown to induce colossal chemical expansion up to 13% and insulator–metal–insulator transition.^15,16 Although rich physical properties have been discovered in perovskite nickelates induced by defect tuning (H_xRNiO_3−δ, 0 < δ < 1, 0 < x < 1), our understanding of the role of ionic defects in nickelates is far from complete. To the best of our knowledge, not much effort has been made to understand the differences and connections between protons and oxygen vacancies in nickelates as well as whether one type of ionic defect can be reversibly converted to the other. This lack of understanding is partly due to the issues with current chemical methods commonly used for ionic tuning. Almost all widely used methods to tune ionic defects can only be used for controlling one single type of ionic defect, either oxygen vacancies or protons. Moreover, the reaction environment of existing methods can be extreme, which can potentially degrade the crystal quality of nickelate thin film samples. We take the most widely used topotactical chemical method as an example. The commonly used metal hydrides are chemically very active and unstable, which largely impedes the use of in situ characterizations to understand the creation and annihilation of oxygen vacancies during the reaction process.¹⁷ Furthermore, during the ionic defect incorporation process, unfavorable side reactions introduce extra ions, which can complicate the correlation between the ionic defects and the physical properties. For example, using metal hydrides to tune the oxygen stoichiometry of perovskite nickelates can potentially introduce extra hydride ions (Nd_0.8Sr_0.2NiO_2−xH_x, where hydrogen is in the form of hydride H⁻) into the crystal lattice during the reaction process.^18,19 The introduced hydride ions further complicate the origin of superconductivity in nickelates.^18,20 Given these challenges, developing new methods to effectively and accurately tune oxygen vacancies in nickelates, preferably based on milder reaction conditions, is necessary to further understand the complexity of interactions between different types of ionic defects.

Dehydration of protonated nickelates (H_xNdNiO₃) is a potential alternative method to tune ionic defects that can address the challenges mentioned above. The dehydration process removes proton defects and creates oxygen vacancies due to the removal of a lattice oxygen along with two protons from the lattice. The creation and annihilation of those ionic defects can be expressed using the defect chemical reaction of (written in Kröger–Vink notations, represent protons (form bonding with oxide ions), oxygen vacancies and lattice oxygen, respectively).^21–23 For example, dehydration of protonated H_xSrCoO_2.5 has been reported to induce phase transitions to SrCoO₂, and the proton stoichiometry directly translates to oxygen deficiency.²³ The reaction condition required for dehydration is much milder compared to other methods, such as chemical treatment using metal hydrides. The reaction temperature of dehydration is relatively low (usually below 300 °C) so that the crystal quality of the oxides can still be nearly intact. Importantly, since there is a stochiometric translation from the proton concentration in the pristine sample to the resulting oxygen vacancy concentration, one can control the oxygen vacancy concentration in the sample after dehydration by adjusting the initial proton concentration in the pristine sample. Due to this direct conversion correlation between proton and oxygen vacancy concentration, if we introduce a spatial gradient of proton concentration into the pristine sample, then after dehydration, the sample will have a concentration gradient of oxygen vacancies. With the oxygen vacancy concentration gradient thin film, we can combine high throughput characterization techniques to construct a property-oxygen vacancy concentration phase diagram. By using this strategy, the correlation between properties and oxygen vacancy concentration can be understood with more precise quantitative results.

In this work, we show that thermochemical dehydration is an effective method for converting protons to oxygen vacancies and tuning the physical properties of nickelate thin films. We first induced phase transition from pristine NdNiO₃ (denoted as NNO in this paper) to fully protonated H_xNdNiO₃ (denoted as H-NNO throughout the paper) using electrochemical methods developed in our previous work.¹⁶ Subsequently, we used a relatively low temperature (only up to 300 °C) to dehydrate H-NNO, which converts the H-NNO phase transition to the oxygen-deficient NdNiO_3−δ phase (denoted as NNO-δ later in this paper). We used soft X-ray absorption spectroscopy (sXAS) to reveal the Ni valence state and oxygen stoichiometry of NNO-δ, which we show close to Ni²⁺ and NdNiO_2.5, respectively. Furthermore, we studied the evolution of crystal structures and electrical conductivity during the dehydration process. It was found that the dehydration of protonated H-NNO was triggered at a threshold temperature of 200 °C, while the dehydration process was completed once the temperature reached 300 °C. We also determined that both protonated H-NNO and NNO-δ were pure electronic conductors with negligible ionic conductivity. Moreover, we synthesized samples with an oxygen vacancy concentration gradient by dehydrating an NNO thin film sample with a proton concentration gradient. By using this method, we can have a continuous varying oxygen stoichiometry from NdNiO_2.67 (highest) to NdNiO_2.5 (lowest) in a single sample. Characterization tools with spatial resolutions, including X-ray diffraction (XRD), sXAS, transport measurement, and ultraviolet-visible (UV-Vis) spectroscopy, were employed to accurately understand the correlation between the physical properties and the oxygen vacancy concentration. Finally, we analyzed the crystal structure of the oxygen-deficient phase at the atomic level using a scanning transmission electron microscope (STEM). We found that the oxygen non-stoichiometry introduced by dehydration can readily lead to the formation of phases with oxygen vacancy ordering. We identified the formation of a 1/3-order phase, with a superstructure showing the periodicity of three unit cells. In summary, we have developed a soft chemistry method based on thermochemical dehydration to tune the oxygen vacancy concentration of nickelates, which is an alternative way compared with the widely used chemical treatment using the harsh chemical environment.

Results and discussion

Characterization of thermochemical dehydration-induced phase transition from H-NNO to NNO-δ

Twenty nm-thick stoichiometric NNO thin films were deposited on LaAlO₃ (LAO) single crystal substrates using pulsed laser deposition (PLD, see Method section). We then electrochemically induced phase transition in the sample from the pristine NNO phase to the fully protonated H-NNO phase (see Method section), following our previous work.¹⁶ The lattice constant of fully protonated H-NNO c_pc,H-NNO determined from XRD was 4.31 Å, consistent with our previous results.¹⁶ Then, we induced thermochemical dehydration from protonated H-NNO to oxygen-deficient NNO-δ by annealing the H-NNO sample in an N₂ atmosphere at 300 °C (see Method section). As shown in Fig. 1(b), the (002) diffraction peak of H-NNO (2θ ∼41.8°) disappeared, and a new (002) diffraction peak associated with the formed oxygen-deficient NNO-δ phase appeared at ∼46.1° (Fig. 1(b)). The out-of-plane lattice constant of oxygen deficient NNO-δ phase is 3.92 Å, with a chemical expansion of 2.9% compared with the pristine NNO (c_pc,NNO = 3.81 Å) and much smaller than that of the H-NNO phase. We analyzed the chemical composition of pristine NNO, protonated H-NNO, and oxygen deficient NNO-δ samples by using time-of-flight secondary ion mass spectrometry (ToF-SIMS). We observed that the proton content in the oxygen deficient NNO-δ is much smaller than that in the protonated H-NNO, which was close to the hydrogen concentration (from adventitious hydrocarbon contaminations) of the pristine NNO sample (Fig. 1(c)). The ToF-SIMS results indicate that the thermochemical dehydration process completely removed the protons in H-NNO. Importantly, the thickness of the thin film is much smaller than the critical length, which is usually of the order of several hundreds of nanometers.²⁴ Critical length l_c is defined as the ratio between the diffusivity D (with a unit of cm² s⁻¹) and the surface chemical reaction rate k (with a unit of cm s⁻¹), which is written as l_c = D/k. Therefore, a large critical length (larger than the thickness of our thin film samples) would indicate that the surface chemical reaction, rather than bulk diffusion, is the rate limiting step. The vertical diffusion of ionic defects within the thin film can be much faster than the surface reaction, which resulted in a uniform distribution of defect concentration in the bulk of the thin film. The narrow width of the rocking curve of oxygen deficient NNO-δ phase indicates that the crystal quality of oxygen deficient NNO-δ phase remains unchanged after the dehydration process (see Fig. S2 in ESI†). Moreover, we can restore the oxygen deficient NNO-δ phase back to the pristine NNO phase by annealing the sample at 300 °C in a pure oxygen atmosphere (Fig. S3(a)†). We also tried dehydrating protonated H-NNO in different reducing atmospheres at 300 °C, including the dry/humid N₂ as well as dry/humid H₂/N₂ mixed gas atmosphere (see Fig. S3(b)†). Interestingly, the lattice constants of oxygen deficient NNO-δ formed in different atmospheres were the same, which shows that the dehydration reaction on H-NNO was relatively insensitive to the partial pressure of H₂O and/or H₂ in the atmosphere. The independence of the dehydration reaction from the atmospheric partial pressure of H₂O and/or H₂ demonstrates that the final concentration of oxygen vacancies is solely determined by the initial concentration of proton defects. Finally, to further test the effectiveness of the dehydration method in manipulating oxygen vacancy concentrations, pristine NNO was annealed in a N₂ environment at 300 °C. The unchanged crystal structure of the annealed sample indicates that under such mild conditions, only the dehydration process can effectively manipulate the concentration of oxygen vacancies (Fig. S3(c)†).

Based on the stochiometric conversion from the proton concentration to the resulting oxygen vacancy concentration, dehydrating protonated H-NNO to oxygen-deficient NNO-δ does not alter the Ni valence state of the sample. We further investigated the Ni valence state of the protonated H-NNO phase and the oxygen-deficient NNO-δ phase. We conducted sXAS of Ni L_2,3-edge and O K-edge (see Method section) with data collected using pristine NNO samples (Ni³⁺) and NiO powder (Ni²⁺) as references. We observed two distinct feature peaks of the Ni L₃-edge in pristine NNO, which are denoted as feature A (centered at ∼854.5 eV) and feature B (centered at ∼856 eV). Similar peaks were observed in the Ni L₃-edge of NiO, where feature A shifts to lower energy and feature B is significantly suppressed. We can fit the Ni L₃-edge of NNO and NiO using two Gaussian peaks to extract the ratio between the peak areas of feature A and feature B (see Fig. S4 in ESI†).^25–27 Previous results show that the area ratio between feature A and feature B (denoted as area ratio_(B/A)) is directly correlated to the Ni valence state.⁴ We calculated the area ratio_(B/A) of pristine NNO and NiO to be 11.3 and 0.32, respectively (see Fig. S4†). We also calculated the area ratio_(B/A) of H-NNO and NNO-δ to be 0.52 and 0.37, respectively. Since the area ratio_(B/A) of H-NNO and NNO-δ is very close to that of NiO, the Ni valence state of both H-NNO and NNO-δ can be quantified to be close to Ni²⁺. These results are consistent with previous findings in protonated H_xSmNiO₃ (H-SNO) and oxygen-deficient SmNiO_3−δ (SNO-δ).⁴ Furthermore, a pre-edge peak (∼530 eV) in the O K-edge of pristine NNO was observed, as shown in Fig. 2(b), indicating the hybridization between Ni 3d and O 2p orbitals.^28–31 In both protonated H-NNO and oxygen-deficient NNO-δ, the formation of ionic defects (protons and oxygen vacancies, respectively) effectively induced electron doping, which suppressed the pre-edge peak of the O K-edge (Fig. 2(b)). Based on the sXAS results of Ni L_2,3-edge and O K-edge, we concluded that the Ni valence state in both protonated H-NNO and oxygen-deficient NNO-δ is Ni²⁺, which means that the dehydration process does not change the Ni valence state of nickelates. Furthermore, we analyzed the optical properties of pristine NNO, protonated H-NNO, and oxygen-deficient NNO-δ. The transmittance data collected on H-NNO and NNO-δ are comparable, which are both higher compared to pristine NNO in the ultraviolet-visible (UV) and infrared (IR) regions (Fig. 3(c)). This finding is consistent with the transport data that shows the insulating behavior of both H-NNO and NNO-δ, which will be discussed later in this paper.

Evolution of physical properties during the thermochemical dehydration process

To further understand the changes in crystal structures and lattice constants during the dehydration process that causes the formation of NNO-δ, we performed ex situ XRD 2θ–ω scans of H-NNO phase transition to NNO-δ at different temperatures (Fig. 3(a)). All the protonated H-NNO was heated in an N₂ atmosphere to a specific temperature and maintained at that temperature for 10 minutes. After heating, the samples were allowed to cool naturally to room temperature in the same atmosphere and then ex situ 2θ–ω scans were performed (see Method section). The (002) diffraction peak of H-NNO (2θ ∼41.8°) remains unchanged with the treatment temperature that is lower than 200 °C. Once the sample was heated to 200 °C, we observed a new (002) diffraction peak associated with the newly-formed oxygen deficient NNO-δ phase near the substrate peak. Meanwhile, we can observe that the width of the (002) diffraction peak of H-NNO increased and its position shifted to a higher 2θ angle, which indicates that the H-NNO phase in the sample had a lower volume fraction and the proton concentration in H-NNO also became lower, consistent with the findings in our previous work.¹⁶ With the annealing temperature further increased, the (002) diffraction peak of H-NNO continuously shifted to a higher angle. The two diffraction peaks of H-NNO and NNO-δ finally merged to a clear and sharp single diffraction peak after annealing at 300 °C, with a lattice constant of ∼3.92 Å. The XRD results suggest that the dehydration of protonated H-NNO to oxygen-deficient NNO-δ occurs in three stages. When the temperature is lower than 200 °C, the H-NNO phase is shown to be stable. Once the temperature is higher than 200 °C, the dehydration process happens, which induces the formation of oxygen vacancies. Finally, the dehydration process is complete at an annealing temperature of 300 °C, with the H-NNO phase completely gone.

Since the H-NNO and NNO-δ phases contain either protons or oxygen vacancies, it is important to understand if the ionic defects can contribute to ionic conductivity. We measured the change in total conductivity (i.e., both electronic and ionic conductivity) during the dehydration process at different temperatures in a humid N₂ atmosphere (Fig. 3(b)). The total conductivity was measured using electrochemical impedance spectroscopy (EIS, see Experimental section), plotted by Nyquist plots and fitted with an equivalent circuit (Fig. S5†). The change in the total conductivity during the dehydration process mirrors the crystal structure change shown by XRD data above, which shows a hysteresis in a warming-up and cooling-down cycle. Below 200 °C, the unchanged H-NNO phase remained stable, which resulted in a monotonically increasing total conductivity. As the temperature reached 200 °C, dehydration occurred, which caused a decrease in conductivity, as shown in Fig. 3(b). After cooling the oxygen-deficient NNO-δ from 300 °C down to room temperature, the total conductivity of the oxygen-deficient NNO-δ decreased monotonically, which was shown to be much lower than the conductivity of H-NNO, thus causing the hysteresis in the temperature-dependent conductivity data shown in Fig. 3(b).

Previous studies have reported high proton conductivity in protonated H-SNO in the temperature range from 300 °C to 500 °C.^32,33 However, in our case, we found negligible proton conductivity in both H-NNO and NNO-δ. The ionic conductivity of both H-NNO and NNO-δ thin films was measured by subtracting the electronic contribution quantified by DC measurement (using Au as ion-blocking electrodes) from the total conductivity measured by AC impedance spectroscopy.³⁴Fig. 3(c) and (d) summarize the total conductivity and electrical conductivity of H-NNO and NNO-δ at different temperatures (see details in the Experimental section). For both pure-phase H-NNO and NNO-δ, we show that the electrical conductivity roughly equals the total conductivity. Therefore, we conclude that both phases are pure electronic conductors with negligible ionic conductivity at temperatures below 300 °C. We attempted to measure the conductivity of NNO-δ at temperatures above 300 °C; however, the crystal structure of NNO-δ became unstable above 300 °C in our experiments (Fig. S3(c)†). Nevertheless, our results show that the ionic defects in H-NNO and NNO-δ, i.e., protons and oxygen vacancies, are immobile and do not contribute to appreciable ionic conductivity in the temperature range we measured, despite the quite large concentration of the ionic defects. This might be due to the strong defect–defect interactions that effectively serve as traps for mobile ions, which makes the mobility of ions quite low.³⁵ Future work is still needed to elucidate the physical origin of large differences between the negligible ionic conductivity of H-NNO/NNO-δ and the high conductivity of other oxide systems, e.g., HSrCoO_2.5.³⁴ It is speculated that the difference in the symmetry of crystal structure, which is correlated to the phonon dispersions, might account for the huge difference in ionic conductivity.³⁴

Construction of properties-oxygen vacancy phase diagram via fine-tuning of thermochemical dehydration

To better understand the correlation between the oxygen vacancy concentration and the physical properties of NNO-δ, we synthesized an NNO-δ thin film with an oxygen vacancy concentration gradient by dehydrating a sample with a proton concentration gradient. The advantage of this approach, compared with the alternative that involves multiple samples with different oxygen vacancy concentrations, is that we can effectively avoid the sample-to-sample variation caused by slight differences in sample growth or experimental conditions.¹⁶ We begin with an H-NNO sample with a proton concentration gradient by using the electrochemical method developed in our previous work¹⁶ (see Method section). Fig. 4(c) shows 2θ–ω XRD data at different spots of a proton concentration gradient H-NNO thin film. The proton concentration increases from the spot at 1 mm to the spot at 5 mm, as shown in Fig. 4(a). The lower-angle (002) diffraction peak at 2θ = 41.8° signifies the H-NNO phase, while the higher-angle (002) peak near the substrate indicates the NNO phase. We then annealed the proton concentration gradient sample in an N₂ atmosphere at 300 °C to induce thermochemical dehydration. Fig. 4(d) shows 2θ–ω XRD patterns at the same position as Fig. 4(c) after dehydration. The XRD results of the oxygen vacancy concentration gradient thin film indicate that the dehydration of the graded H-NNO thin film happened in two different stages. From 0 mm to 2.5 mm, dehydration of the NNO phase with low proton concentration induces the oxygen-deficient phase NNO-δ with a lattice constant of c_pc,NNO-δ = 3.87 Å (Fig. S6†). On the other hand, from 2.5 mm to 5 mm, dehydration of H-NNO with a higher proton concentration resulted in the oxygen-deficient phase NNO-δ with a changing lattice constant c_pc,NNO-δ between 3.87 Å and 3.91 Å (see also in Fig. S6 in ESI†). We further determined the Ni oxidation state with different defect concentrations using sXAS. Fig. 4(e) and (f) show the Ni L_2,3-edge of the samples with a proton concentration gradient and an oxygen vacancy concentration gradient, respectively. Again, we quantified the Ni valence state by calculating the area ratio_(B/A) between two peaks in the Ni L₃-edge (see Fig. S7 in ESI†). In the graded H-NNO sample, the Ni valence state decreases from Ni³⁺ to Ni²⁺ with increasing proton defect concentration (Fig. 4(g)). The change in proton content x, calculated based on the variation in the Ni valence state, ranges from 0 to 1. Detailed results are presented in Fig. S10(a).† The suppression of the pre-edge peak in the O K-edge of the proton concentration gradient thin film also confirms the change of the Ni valence state from Ni³⁺ to Ni²⁺ (Fig. S8†). However, Fig. 4(h) shows that the Ni valence state in the graded NNO-δ thin film was already around Ni^2.2+ at spots located from 0 mm to 2.5 mm, consistent with the XRD results, with a constant lattice constant shown in Fig. 4(d). The Ni valence state further decreased to Ni²⁺ at positions from 2.5 mm to 5 mm. Similarly, from 0 mm to 2.5 mm of the thin film, the pre-edge peak in the O K-edge of the graded NNO-δ sample remained nearly identical. From 2.5 mm to 5 mm, the pre-edge peak was suppressed (Fig. S8†). Based on the XRD and sXAS result analyses, we can determine the oxygen content change in the oxygen vacancy concentration gradient thin film. Throughout the entire length of the thin film, the Ni valence state changes from Ni^2.2+ to Ni²⁺, which indicates oxygen stoichiometry changing from NdNiO_2.67 to NdNiO_2.5. The detailed changes of oxygen non-stoichiometry δ at different positions of oxygen-graded thin film are shown in Fig. S10(b).† Importantly, within the 0 mm to 2.5 mm position of the thin film, the Ni valence state in the graded NNO-δ sample is lower than that in the graded H-NNO sample. This decrease in the Ni valence state could potentially be caused by the lateral diffusion of oxygen vacancies during the dehydration process in the graded thin film. Further investigation into the kinetics of this lateral diffusion is necessary.

We further attempted to anneal the proton concentration gradient H-NNO sample in a pure hydrogen atmosphere at 300 °C. Fig. S9† summarizes 2θ–ω XRD patterns of the proton concentration gradient sample and the same sample annealed in the pure hydrogen atmosphere. Compared with the previous sample annealed in N₂, the oxygen vacancy concentration in the hydrogen-annealed sample was more homogenous. Fig. S11† summarizes the Ni L_2,3-edge and O K-edge of the hydrogen annealed sample. Due to the more homogenous oxygen vacancy concentration, the Ni L_2,3-edge and O K-edge of the hydrogen-annealed sample remain consistent throughout the entire thin film. The enhanced homogeneity of oxygen vacancies in the hydrogen annealed sample is due to the stronger reducibility of the pure hydrogen atmosphere. Unlike the reaction in N₂, pure hydrogen at 300 °C can directly remove lattice oxygen from the film, creating oxygen vacancies, and the reaction equation is written as or written as . Previous studies have shown that placing pristine nickelates in a pure hydrogen environment creates oxygen vacancies.^36,37 This indicates that exposure of an H-NNO film with a proton concentration gradient to a pure hydrogen atmosphere not only facilitates thermochemical dehydration but also directly removes lattice oxygen to form oxygen vacancies. Therefore, the distribution of oxygen vacancies in the sample treated with pure hydrogen is more consistent. Importantly, in the pure N₂ atmosphere, due to the maximum proton concentration being a single proton per unit cell, dehydration of H-NNO can only generate a NdNiO_2.5 phase with oxygen non-stoichiometry δ = 0.5 and Ni²⁺ valence state. In the pure hydrogen atmosphere, the co-existence of dehydration and direct creation of oxygen vacancies in the graded H-NNO sample may further reduce the Ni valence, lowering it to below Ni²⁺. Further analysis of the sXAS results of the hydrogen-annealed sample confirms the point (Fig. S11†). We observed an extra peak labeled as A’ (∼854 eV) in the Ni L₃-edge of the hydrogen annealed sample. A similar extra peak was observed in nickelates with the infinite layer structure, such as NdNiO₂ and LaNiO₂.³⁸ Also, we observed a small pre-edge peak at 533 eV in the O K-edge of the hydrogen annealed sample. A similar pre-edge peak was also observed in nickelates with the infinite layer structure, such as Nd_0.8Sr_0.2NiO₂,³⁹ Nd₄Ni₃O₈ (ref. 40) and Nd₆Ni₅O₁₂.⁴¹ These results are the preliminary evidence that perhaps we can also fabricate nickelate samples with the infinite layer structure by using thermochemical dehydration with pure hydrogen. Nevertheless, further experiments are needed to evaluate if one can obtain the pure phase of the infinite layer structure.

The quantitative correlation between oxygen vacancy concentration and electrical conductivity has not yet been fully understood in nickelates.^42,43 Therefore, utilizing the continuously varying oxygen vacancy concentration in the graded NNO-δ thin film can provide insights into the effect of oxygen vacancies on transport properties. Electrical conductivity measurements at room temperature were performed at different spots of the thin film with a gradient of oxygen vacancy concentration (Fig. 5(a)). Our previous XRD and sXAS results confirm that from 0 mm to 2.5 mm of the graded thin film, the oxygen content mainly maintains at NdNiO_2.67. Similarly, the electrical conductivity at room temperature remains unchanged at ∼10 S cm⁻¹. On the contrary, from 2.5 mm to 5 mm NdNiO_2.67 transition to NdNiO_2.5, the electrical conductivity at room temperature drops by three orders of magnitude to approximately 10⁻² S cm⁻¹. The optical absorbance measurements also show that more visible light transmits through thin film with increasing oxygen vacancy concentration (Fig. 5(b)).

Crystal structure of oxygen-deficient NNO-δ at the atomic level

The formation of oxygen vacancies induces a complex oxygen-deficient NNO-δ phase diagram. A slight change in oxygen content in NNO-δ can impose a significant influence on its crystal structure and physical properties. For example, the charge ordering discovered in infinite layer structure NdNiO₂ and its correlation with the ordered oxygen atom superstructure is still under intense exploration.^42,44–46 Therefore, we further investigated the atomic-level crystal structure of the oxygen-deficient NNO-δ phase using electron microscopy. Our quantitative analysis above shows that certain parts of the oxygen vacancy gradient sample are in the stoichiometry of NdNiO_2.67. We performed STEM to analyze the atomic structure of the NdNiO_2.67 phase, which is summarized in Fig. 6. The Fourier transform (FFT) of high-angle annular dark field scanning transmission electron microscopy (HAADF-STEM, Fig. 6(b)) images show the existence of 1/3-order diffraction peaks, corresponding to the three-unit-cell periodicity along with the [110]_pc orientation of the perovskite lattice. The 1/3-order diffraction peaks in the FFT image are consistent with previous results, which show that the formation of oxygen vacancies at the apical position of the NiO₆ octahedron creates NiO₅ pyramidal chains.⁴² The HAADF-STEM images reveal the displacement of Nd atoms and the formation of a stripe phase indicated by a darker contrast (shown in white dashed lines in Fig. 6(c)). The orientation of the stripe phase is along with the [110]_pc, which also has the same three-unit-cell periodicity, presumably due to the ordered NiO₅ pyramidal chains.⁴⁷ The annular bright-field scanning transmission electron microscopy (ABF-STEM) images further support this interpretation. Fig. 6(d) shows the oxygen vacancy chain with the three-unit-cell lattice periodicity along with the [110]_pc orientation, indicated by the lower contrast in the ABF-STEM images. Overall, the STEM analysis of the NdNiO_2.67 phase reveals a superstructure with three-unit-cell lattice periodicity, which can be seen from both the HAADF images (showing mainly the Nd atoms) and the ABF images (showing the O atoms). Our results show that the oxygen vacancies created by dehydration can lead to complicated ordering of ionic defects, which has a strong impact on the ligand field surrounding the Ni cations. Although here we only show the atomic structure of NdNiO_2.67 as an example, further studies will be underway to correlate the oxygen anion ordering and oxygen stoichiometry once we overcome some experimental difficulties regarding STEM sample preparations. We expect a gradual change of oxygen vacancy ordering at different parts of the NdNiO_3−δ sample with an oxygen vacancy concentration gradient or even the co-existence of two or multiple phases with different oxygen vacancy ordering.

Conclusion

In conclusion, we studied the dehydration-induced phase transition from H-NNO to NNO-δ in detail. Our work shows that the dehydration of electrochemically protonated H-NNO is an effective method to tune the oxygen vacancy concentration in NNO. We identified the changes in physical properties, including the crystal structure, electrical conductivity and optical properties caused by the dehydration-induced phase transition. More importantly, we have developed a new method to quantitatively correlate oxygen vacancy concentration with the physical properties of NNO-δ. The device with spatially-varying oxygen vacancy concentration, coupled with spatially-resolved characterizations, can effectively avoid the sample-to-sample variation that can cause large uncertainties in traditional chemical treatment used for controlling oxygen non-stoichiometry. The detailed changes in Ni valence state, crystal structure, optical properties, and transport properties as a function of oxygen vacancy concentration have been studied in one single compositionally graded thin film. The STEM analysis reveals the three-unit-cell lattice periodicity superstructure of oxygen and Nd atoms in the oxygen-deficient phase NdNiO_2.67. Our study not only elucidates the impact of oxygen vacancies on the phase and properties of NNO as a model system for nickelate perovskites, but more importantly, our work provides a new pathway, based on thermochemical dehydration, for future research into ionic conductivity, chemical diffusivity, optical properties, magnetism, and other characteristics of other functional oxides with oxygen deficiency.

Methods

Thin film growth and characterization

In our work, perovskite oxide NdNiO₃ (NNO) was deposited on LaAlO₃ (LAO) using pulsed laser deposition (PLD). A KrF excimer laser with 248 nm wavelength was used, with a repetition rate of 5 Hz. The LAO substrate was heated to 650 °C, and oxygen partial pressure for growth was fixed at 100 mTorr. The laser fluence was at 1.7 J cm⁻². The single-phase NNO target used for PLD growth was purchased from Toshima Ltd (Saitama, Japan). After PLD growth, high-resolution X-ray diffraction (HRXRD, Bruker, D8 discover) was used to characterize the thin film to guarantee the desired crystal structure. The HR-XRD uses monochromatic Cu K_α1 radiation (λ = 1.5406 Å). The thickness of thin films is determined by X-ray reflectivity (XRR). The X-ray has a focused beam with a diameter of 0.6 mm.

High temperature experiments

All experiments that needed heating and gas flow were carried out on a probe station with four probes. The probe station has a circle-shaped ceramic heating plate with 1 cm diameter. The maximum heating temperature of the quartz heating plate is 450 °C. When the sample is put on the quartz heating plates, the heating rate is set at 20 °C min⁻¹. The purity of N₂, H₂/N₂ mixture is 99.99%. The pure hydrogen is obtained from a hydrogen gas generator. To measure the electrical conductivity at an elevated temperature, a Biologic SP-300 potentiostat was used. The rectangular geometry used for high-temperature electrical conductivity measurement is 5 mm in length and 3 mm in width. 100 nm thick Au electrode was sputtered on both sides of the thin film. The two-probe method was used to measure the electrical conductivity. The test voltage on the sample is 0.1 V.

Electrochemical impedance spectroscopy

EIS measurements were performed by using a Biologic SP-300 potentiostat. These measurements were carried out in a frequency range from 1 MHz to 1 Hz, with an amplitude of 1000 mV. The rectangular geometry used for transport measurement is 5 mm in length and 3 mm in width. A 100 nm thick gold layer was deposited on both sides of the sample for better contact.

Ultraviolet-visible light spectroscopy experiments

The UV-Vis transmittance spectra of different samples in Fig. 2(c) were collected by an ultraviolet-visible near-infrared spectrophotometer, UV3600Plus + UV2700 (Shimadzu Co., Ltd Japan). The used wavelength ranges from 400 to 2500 nm. The mapping of spatial-resolved UV-Vis transmittance spectra in Fig. 5(c) was conducted on a CRAIC 20/30 PV microspectrophotometer (CRAIC Technologies Inc, USA). At different spots of the thin film with oxygen vacancy concentration gradient, the diameter size of the spot is 50 × 50 μm². The distance between each spot is 100 μm. A total number of 50 spots were acquired. The wavelength for each spatial-resolved spectrum ranges from 300 to 800 nm. For the above UV-Vis experiments in transmission mode, thin films on transparent (double-side polished) LaAlO₃ single-crystal substrates were used. Reference and dark spectra were taken based on a pristine transparent LaAlO₃ single crystal before further UV-Vis spectra acquisition.

Electrochemical measurement in aqueous solution

All the electrochemical experiments are carried out in a standard three-electrode system using a Biologic SP-300 potentiostat. A Hg/HgO electrode and a Pt wire are used as the reference electrode and the counter electrode, respectively. Au thin films with a thickness of ∼100 nm were sputtered on both sides of NNO thin films as the contact. The electrochemical experiment was done in a 0.1 M KOH aqueous solution. Before every experiment, the Hg/HgO reference electrode is calibrated with a reverse hydrogen electrode in 0.1 M KOH aqueous solution. To avoid 0.1 M KOH solution contact with the Au electrode, a commercial electrolyte gating cell (Redox.me, Sweden) is used. To induce phase transition from pristine NNO to fully protonated H-NNO, a very reduced potential (0.02 V vs. RHE) has been applied for 5 min. To electrochemically synthesize the proton concentration gradient thin film, two potentials with different reducibility (0.02 V vs. RHE, 0.5 V vs. RHE) were applied on both sides of the NNO thin film. Potential differences introduce a spatially varying potential in the NNO sample that causes a gradient of the electrochemical driving force for protonation. The geometry of the NNO sample used for synthesizing the proton-graded thin film is 5 mm in length and 3 mm in width.

Time of flight-secondary ion mass spectrometry experiment

The ToF-SIMS used for chemical composition analysis is a PHI Nano ToF III system. The sputtering area was kept at 400 × 400 μm² and the detection area was kept at 50 × 50 μm². The Al element from the LaAlO₃ substrate and the thin film are used to identify the position of the heterointerface.

Soft X-ray absorption spectroscopy

Soft X-ray absorption spectroscopy (sXAS) was performed at BL02B02 at the Shanghai Synchrotron Radiation Facility (SSRF). The sXAS spectra of Ni L_2,3-edge and O K-edge were collected by using the total electron yield (TEY) mode. The bending magnet beamline delivers soft X-ray photons with photon flux around 1 × 10¹¹ photons per s @ E/ΔE = 3700 and a tightly focused beam spot size (∼150 × 50 μm²) at the sample.⁴⁸ The range of the photon energy used for Ni L_2,3-edge and O K-edge was 840–890 eV and 520–560 eV, respectively.

High throughput room temperature conductivity measurement

The rectangular geometry used for transport measurement is 5 mm in length and 3 mm in width. The rectangular geometry and gold electrode were fabricated using standard photolithography. We used a cryogenic system (Model RDK-101D, Sumitomo Heavy Industries. Ltd) with a designed switch where two different standard measuring units (Keithley 2450, Keithley DAQ 6510) were combined. A two-probe method was used to perform the high-throughput measurement. The distance between every two voltage-measuring probes is 0.4 mm. The test current on a sample with different oxygen vacancy concentrations is 100 μA and 200 nA, respectively.

Scanning transmission electron microscope

The FEI Helios Nano-lab 600i dual-beam, focused ion beam, field emission gun, and scanning electron microscope were used to prepare a cross-section for STEM imaging and analysis. The FEI TITAN THEMIS G2 transmission electron microscope was employed to acquire the STEM data. The transmission electron microscope was operated at 300 kV and 25 mrad probe convergence semi-angle. A series of 10 rapid-frame images were applied to perform high-precision structural measurements.

Data availability

All data supporting the findings of this study are available within the article and its ESI.†

Author contributions

Q. L. originated and supervised the research. H. C., Z. X. contributed equally to this work. H. C., Z. X. grew the sample and performed the electrochemical experiments, high temperature experiments, XRD, SIMS, STEM experiments. H. C., L. W. performed UV-visible light spectroscopy experiments. M. D. conducted the high throughput room temperature conductivity measurements and analyzed the data under the supervision of J. W. H. C., Z. X., Y. H., Y. L., L. W. conducted sXAS experiments. All the authors discussed the results and contributed to the writing of the manuscript.

Conflicts of interest

There are no conflicts of interest to declare.

Acknowledgements

This work was supported by funding from the Research Center for Industries of the Future, and the School of Engineering Dean Special Projects Fund (SOE-DSPF), Westlake University, and the National Natural Science Foundation of China (NSFC, Grant No. 52202148). This work used shared facilities at the Instrumentation and Service Centers for Molecular Science (ISCMS) and the Instrumentation and Service Centers for Physical Science (ISCPS) of Westlake University. Part of the work was performed at BL02B02 of the Shanghai Synchrotron Radiation Facility, which is supported by the ME2 project from the National Natural Science Foundation of China (Grant No. 11227902). The authors would like to thank Yinjuan Chen, staff scientist of ISCMS, Ou Yang, application scientist of ULVAC-PHI Instruments Co., Ltd, for their valuable advice on ToF-SIMS analysis.

References

1. X. Liao, V. Singh and H. Park, Phys. Rev. B, 2021, 103, 1–10 Search PubMed .
2. T. X. Nguyen, Y. C. Liao, C. C. Lin, Y. H. Su and J. M. Ting, Adv. Funct. Mater., 2021, 2101632, 1–10 Search PubMed .
3. Z. Zhang, D. Schwanz, B. Narayanan, M. Kotiuga, J. A. Dura, M. Cherukara, H. Zhou, J. W. Freeland, J. Li, R. Sutarto, F. He, C. Wu, J. Zhu, Y. Sun, K. Ramadoss, S. S. Nonnenmann, N. Yu, R. Comin, K. M. Rabe, S. K. R. S. Sankaranarayanan and S. Ramanathan, Nature, 2018, 553, 68–72 CrossRef CAS PubMed .
4. M. Kotiuga, Z. Zhang, J. Li, F. Rodolakis, H. Zhou, R. Sutarto, F. He, Q. Wang, Y. Sun, Y. Wang, N. A. Aghamiri, S. B. Hancock, L. P. Rokhinson, D. P. Landau, Y. Abate, J. W. Freeland, R. Comin, S. Ramanathan and K. M. Rabe, Proc. Natl. Acad. Sci. U. S. A., 2019, 116, 21992–21997 CrossRef CAS .
5. T. Moriga, O. Usaka, I. Nakabayashi, T. Kinouchi, S. Kikkawa and F. Kanamaru, Solid State Ionics, 1995, 79, 252–255 CrossRef CAS .
6. I. V. Nikulin, M. A. Novojilov, A. R. Kaul, S. N. Mudretsova and S. V. Kondrashov, Mater. Res. Bull., 2004, 39, 775–791 CrossRef CAS .
7. B. X. Wang, S. Rosenkranz, X. Rui, J. Zhang, F. Ye, H. Zheng, R. F. Klie, J. F. Mitchell and D. Phelan, Phys. Rev. Mater., 2018, 2, 1–7 CrossRef .
8. G. Niu, E. Hildebrandt, M. A. Schubert, F. Boscherini, M. H. Zoellner, L. Alff, D. Walczyk, P. Zaumseil, I. Costina, H. Wilkens and T. Schroeder, ACS Appl. Mater. Interfaces, 2014, 6, 17496–17505 CrossRef CAS .
9. Y. Sun, Q. Wang, T. J. Park, T. E. Gage, Z. Zhang, X. Wang, D. Zhang, X. Sun, J. He, H. Zhou, D. G. Lim, C. Huang, H. Yu, X. Chen, H. Wang, J. Mei, E. Deguns and S. Ramanathan, ACS Appl. Electron. Mater., 2021, 3, 1719–1731 CrossRef CAS .
10. C. Hu, X. Wang, T. Yao, T. Gao, J. Han, X. Zhang, Y. Zhang, P. Xu and B. Song, Adv. Funct. Mater., 2019, 29, 1–8 Search PubMed .
11. H. Guo, J. Huang, H. Zhou, F. Zuo, Y. Jiang, K. H. L. Zhang, X. Fu, Y. Bu, W. Cheng and Y. Sun, ACS Appl. Mater. Interfaces, 2021, 13, 24887–24895 CrossRef CAS .
12. C. Leighton, Nat. Mater., 2019, 18, 13–18 CrossRef CAS .
13. Y. Wu, D. Li, C. L. Wu, H. Y. Hwang and Y. Cui, Nat. Rev. Mater., 2023, 8, 41–53 CrossRef .
14. D. Li, K. Lee, B. Y. Wang, M. Osada, S. Crossley, H. R. Lee, Y. Cui, Y. Hikita and H. Y. Hwang, Nature, 2019, 572, 624–627 CrossRef CAS .
15. L. Gao, H. Wang, F. Meng, H. Peng, X. Lyu, M. Zhu, Y. Wang, C. Lu, J. Liu, T. Lin, A. Ji, Q. Zhang, L. Gu, P. Yu, S. Meng, Z. Cao and N. Lu, Adv. Mater., 2023, 35, 1–10 Search PubMed .
16. H. Chen, M. Dong, Y. Hu, T. Lin, Q. Zhang, E.-J. Guo, L. Gu, J. Wu and Q. Lu, Nano Lett., 2022, 22, 8983–8990 CrossRef CAS PubMed .
17. Y. Yu, Y. Lee, M. Gonzalez, K. Lee, B. Y. Wang, W. J. Kim, J. Fowlie and H. Hwang, Bull. Am. Phys. Soc., 2024 Search PubMed.
18. X. Ding, C. C. Tam, X. Sui, Y. Zhao, M. Xu, J. Choi, H. Leng, J. Zhang, M. Wu, H. Xiao, X. Zu, M. Garcia-Fernandez, S. Agrestini, X. Wu, Q. Wang, P. Gao, S. Li, B. Huang, K. J. Zhou and L. Qiao, Nature, 2023, 615, 50–55 CrossRef CAS PubMed .
19. P. Puphal, V. Pomjakushin, R. A. Ortiz, S. Hammoud, M. Isobe, B. Keimer and M. Hepting, Front. Phys., 2022, 10, 1–6 Search PubMed .
20. W. Wei, D. Vu, Z. Zhang, F. J. Walker and C. H. Ahn, Sci. Adv., 2023, 9, 1–7 Search PubMed .
21. K. Inaba, S. Suzuki, Y. Noguchi, M. Miyayama, K. Toda and M. Sato, Eur. J. Inorg. Chem., 2008, 5471–5475 CrossRef CAS .
22. J. R. Günter, J. Solid State Chem., 1972, 5, 354–359 CrossRef .
23. H. B. Li, S. Kobayashi, C. Zhong, M. Namba, Y. Cao, D. Kato, Y. Kotani, Q. Lin, M. Wu, W. H. Wang, M. Kobayashi, K. Fujita, C. Tassel, T. Terashima, A. Kuwabara, Y. Kobayashi, H. Takatsu and H. Kageyama, J. Am. Chem. Soc., 2021, 143, 17517–17525 CrossRef CAS .
24. M. Kubicek, Z. Cai, W. Ma, B. Yildiz, H. Hutter and J. Fleig, ACS Nano, 2013, 7, 3276–3286 CrossRef CAS .
25. R. J. Green, M. W. Haverkort and G. A. Sawatzky, Phys. Rev. B, 2016, 94, 1–5 CrossRef .
26. J. García, J. Blasco, M. G. Proietti and M. Benfatto, Phys. Rev. B: Condens. Matter Mater. Phys., 1995, 52, 15823–15828 CrossRef .
27. X. Gao, J. Liu, Y. Ji, L. Wei, W. Xiao, S. Hu, L. Li, K. Chen and Z. Liao, Adv. Electron. Mater., 2023, 9, 2300116 CrossRef CAS .
28. J. Liu, M. Kargarian, M. Kareev, B. Gray, P. J. Ryan, A. Cruz, N. Tahir, Y. De Chuang, J. Guo, J. M. Rondinelli, J. W. Freeland, G. A. Fiete and J. Chakhalian, Nat. Commun., 2013, 4, 2714 CrossRef PubMed .
29. D. Meyers, J. Liu, J. W. Freeland, S. Middey, M. Kareev, J. Kwon, J. M. Zuo, Y. De Chuang, J. W. Kim, P. J. Ryan and J. Chakhalian, Sci. Rep., 2016, 6, 1–8 CrossRef PubMed .
30. X. Lu, J. Liu, N. Zhang, B. Xie, S. Yang, W. Liu, Z. Jiang, Z. Huang, Y. Yang, J. Miao, W. Li, S. Cho, Z. Liu, Z. Liu and D. Shen, Adv. Sci., 2022, 9, 1–6 Search PubMed .
31. E. J. Guo, Y. Liu, C. Sohn, R. D. Desautels, A. Herklotz, Z. Liao, J. Nichols, J. W. Freeland, M. R. Fitzsimmons and H. N. Lee, Adv. Mater., 2018, 30, 1–8 Search PubMed .
32. Y. Zhou, X. Guan, H. Zhou, K. Ramadoss, S. Adam, H. Liu, S. Lee, J. Shi, M. Tsuchiya, D. D. Fong and S. Ramanathan, Nature, 2016, 534, 231–234 CrossRef CAS .
33. H. Li, Y. Li, B. Li, W. Huang and Y. Ding, RSC Adv., 2024, 14, 147–153 RSC .
34. N. Lu, Z. Zhang, Y. Wang, H. B. Li, S. Qiao, B. Zhao, Q. He, S. Lu, C. Li, Y. Wu, M. Zhu, X. Lyu, X. Chen, Z. Li, M. Wang, J. Zhang, S. C. Tsang, J. Guo, S. Yang, J. Zhang, K. Deng, D. Zhang, J. Ma, J. Ren, Y. Wu, J. Zhu, S. Zhou, Y. Tokura, C. W. Nan, J. Wu and P. Yu, Nat. Energy, 2022, 7, 1208–1216 CrossRef CAS .
35. Y. Yamazaki, F. Blanc, Y. Okuyama, L. Buannic, J. C. Lucio-Vega, C. P. Grey and S. M. Haile, Nat. Mater., 2013, 12, 647–651 CrossRef CAS .
36. T. Moriga, O. Usaka, I. Nakabayashi, Y. Hirashima, T. Kohno, S. Kikkawa and F. Kanamaru, Solid State Ionics, 1994, 74, 211–217 CrossRef CAS .
37. T. Moriga, O. Usaka, T. Imamura, I. Nakabayashi, I. Matsubara, T. Kinouchi, S. Kikkawa and F. Kanamaru, Bull. Chem. Soc. Jpn., 1994, 67, 687–693 CrossRef CAS .
38. M. Hepting, D. Li, C. J. Jia, H. Lu, E. Paris, Y. Tseng, X. Feng, M. Osada, E. Been, Y. Hikita, Y. D. Chuang, Z. Hussain, K. J. Zhou, A. Nag, M. Garcia-Fernandez, M. Rossi, H. Y. Huang, D. J. Huang, Z. X. Shen, T. Schmitt, H. Y. Hwang, B. Moritz, J. Zaanen, T. P. Devereaux and W. S. Lee, Nat. Mater., 2020, 19, 381–385 CrossRef CAS PubMed .
39. B. H. Goodge, D. Li, K. Lee, M. Osada, B. Y. Wang, G. A. Sawatzky, H. Y. Hwang and L. F. Kourkoutis, Proc. Natl. Acad. Sci. U. S. A., 2021, 118, 1–7 CrossRef .
40. D. Ferenc Segedin, B. H. Goodge, G. A. Pan, Q. Song, H. LaBollita, M. C. Jung, H. El-Sherif, S. Doyle, A. Turkiewicz, N. K. Taylor, J. A. Mason, A. T. N'Diaye, H. Paik, I. El Baggari, A. S. Botana, L. F. Kourkoutis, C. M. Brooks and J. A. Mundy, Nat. Commun., 2023, 14, 1–14 Search PubMed .
41. G. A. Pan, D. Ferenc Segedin, H. LaBollita, Q. Song, E. M. Nica, B. H. Goodge, A. T. Pierce, S. Doyle, S. Novakov, D. Córdova Carrizales, A. T. N'Diaye, P. Shafer, H. Paik, J. T. Heron, J. A. Mason, A. Yacoby, L. F. Kourkoutis, O. Erten, C. M. Brooks, A. S. Botana and J. A. Mundy, Nat. Mater., 2022, 21, 160–164 CrossRef CAS PubMed .
42. C. T. Parzyck, N. K. Gupta, Y. Wu, V. Anil, L. Bhatt, M. Bouliane, R. Gong, B. Z. Gregory, A. Luo, R. Sutarto, F. He, Y. Chuang, T. Zhou, G. Herranz, L. F. Kourkoutis, A. Singer and D. G. Schlom, Nat. Mater., 2024, 23, 486–491 CrossRef CAS .
43. K. Han, M. Xie, Y. Mei, R. Lin, L. Xu, P. Chen, P. Yin, S. Zeng, B. Ge, A. Ariando, D. Song, X. Renshaw Wang, W. Wu and Z. Huang, Appl. Phys. Lett., 2023, 123, 182601 CrossRef CAS .
44. M. Rossi, M. Osada, J. Choi, S. Agrestini, D. Jost, Y. Lee, H. Lu, B. Y. Wang, K. Lee, A. Nag, Y. De Chuang, C. T. Kuo, S. J. Lee, B. Moritz, T. P. Devereaux, Z. X. Shen, J. S. Lee, K. J. Zhou, H. Y. Hwang and W. S. Lee, Nat. Phys., 2022, 18, 869–873 Search PubMed .
45. C. C. Tam, J. Choi, X. Ding, S. Agrestini, A. Nag, M. Wu, B. Huang, H. Luo, P. Gao, M. García-Fernández, L. Qiao and K. J. Zhou, Nat. Mater., 2022, 21, 1116–1120 CrossRef CAS PubMed .
46. G. Krieger, L. Martinelli, S. Zeng, L. E. Chow, K. Kummer, R. Arpaia, M. Moretti Sala, N. B. Brookes, A. Ariando, N. Viart, M. Salluzzo, G. Ghiringhelli and D. Preziosi, Phys. Rev. Lett., 2022, 129, 27002 CrossRef CAS .
47. Y. M. Wu, P. Puphal, H. Lee, J. Nuss, M. Isobe, B. Keimer, M. Hepting, Y. E. Suyolcu and P. A. Van Aken, Phys. Rev. Mater., 2023, 7, 1–8 Search PubMed .
48. G. Ren, N. Zhang, X. Feng, H. Zhang, P. Yu, S. Zheng, D. Zhou, Z. Tian and X. Liu, Chin. Phys. B, 2020, 29, 16101 CrossRef CAS .

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ta03609c

‡ These authors contributed equally to this work.

This journal is © The Royal Society of Chemistry 2024