Giant Resistive Switching and Lattice Modulation at Full Temperature Range in a Sr‐Doped Nickelate Oxide Transistor

How to promote the resistance switching ratio, enlarge the operating temperature range and accelerate the switching speed is at the forefront of ionic gating electronics. Usually, most attention has been paid to materials with metal‐to‐insulator transition (MIT), e.g., VO2 and SmNiO3. Here, Sr‐doped nickelate (Nd0.8Sr0.2NiO3) films which do not exhibit MIT are used for electric field control of H‐doping and to detect the variation of resistance, lattice, and electronic structures. The experimental results directly show a giant resistive switching by more than 105 at the full temperature range (2 K–300 K) and lattice modulation by 3.4%. More importantly, much faster switching speeds can be achieved in Nd0.8Sr0.2NiO3 devices than in nondoped NdNiO3 ones. Such high switching performance is demonstrated to arise from strongly suppressed Ni–O hybridization after H doping. The results of this study provide a new material paradigm for developing energy‐efficient neuromorphic computing. Further, the doping effect on device performance also suggests a novel approach to develop correlated perovskite oxide transistors in order to fulfill practical applications.


Introduction
Perovskite rare-earth nickelates (RNiO 3 , R = rare-earth element) are correlated quantum materials with rich physical phases, such as metal-insulator transitions [1,2] and multiferroic effects. [3] The high ionic conductivity due to covalent ground state and low-energy phonon modes further make nickelates very promising for developing synaptic transistor, [4] ultrasensitive electric field sensor, [5] and reconfigurable electronics for artificial intelligence using electric field controlled phase Shedding light on the doping effect for nickelates, our findings suggest a novel approach to engineering the nickelate-based devices toward practical industrial application.

Mechanism of Electric Field Induced Hydrogen Doping and Growth of Thin Films
Electric field-controlled ionic liquid gating (ILG), which can induce proton insertion at a positive potential, has emerged as an attractive means to control physical properties of a material (Figure 1a). [12][13][14][15] A schematic proton diffusion process in NSNO film is shown in Figure 1b which includes: 1) proton rotational diffusion, 2) transferring to neighboring oxygen, 3) bending, and 4) elongation of the Ni-O bond. [16][17][18][19] Under positive potential, protons intercalate into the NSNO lattice accompanied by uptake of electrons released by oxidation at the counter electrode in order to remain electrically neutral. [20,21] With one more electron, Ni 3+ t e 2g 6 g 1 from metallic NSNO is converted into insulating Ni 2+ t e 2g 6 g 2 , owing to strong Coulomb repulsion between e g electrons of Ni 2+ which causes a large Mott-Hubbard splitting (Figure 1c). [8] In this study, high-quality epitaxial NSNO thin films were used for device fabrication. NSNO films were grown on SrTiO 3 (001) substrates using pulsed laser deposition (PLD) assisted by in situ reflection high-energy electron diffraction (RHEED). Continuous oscillation of RHEED intensity indicates that the films grew in a layer-by-layer growth fashion and the thickness of the film thus was precisely controlled (Figure 1d). Surface morphology characterized by atomic force microscopy (AFM) (inset in Figure 1d) demonstrates that the films were atomic smooth with visible terraces. High-resolution X-ray diffraction ( Figure 1e) and rocking curve (the value of full width at half-maximum is 0.02°; Figure 1e inset) measurements further confirm that the films are of high crystalline quality without impurity phases.

Determination of Hydrogen Profile across The Films
To illustrate how the ionic gating plays a role in hydrogen doping, we employed ex-situ secondary-ion mass spectrometry  (2), Proton transport by rotational diffusion within an octahedron (1) and transferring to a neighboring oxygen ion facilitated by the hydrogen bond (dashed red line) (2). (3) and (4), The bending (3) and the stretching (4) of the Ni-O bond promote processes (2) and (1), respectively. c) Schematic band diagram of Mott-Hubbard splitting induced by electron doping in NSNO films. d) RHEED intensity oscillation during the growth of NSNO films. Inset shows surface morphology of NSNO films observed by AFM. e) XRD symmetric theta-2theta scans of NSNO films, with film thickness of ≈36 nm. No visible impurity phases in the films. Inset shows the (002) rocking curve with a full width at half-maximum value of 0.02°, indicating a good quality of NSNO films.
(SIMS) to measure the depth profiles of hydrogen in NSNO films. As shown in Figure 2a, SIMS results unambiguously demonstrate that hydrogen migrated into NSNO films during ILG. Figure 2b shows the three-dimensional distribution of different elements in the as-gated sample. Finite concentrations of Nd, Sr, Ni, O, and H elements were observed in upper film, directly demonstrating the intercalation of hydrogen into NSNO. Intriguingly, for the underneath SrTiO 3 (STO) substrate, Sr, Ti, O, and H elements are detected, which indicates that the hydrogen doping during ILG is throughout the entire NSNO film. For the rest of this paper, the defaults are ex situ measurements, unless otherwise stated.

Colossal Resistance Switching and Electrochromic Effect
The impact of hydrogen doping on resistance is shown in Figure 3a by comparing the temperature-dependent resistance (R-T) measurements between pristine and hydrogenated NSNO films. The resistance of hydrogenated NSNO increases as the temperature decreases, indicating an insulating behavior in strong contrast to metallic behavior in pristine NSNO. The hydrogenated thin films can be reversed to the low-resistance state by applying a reverse bias (Figure 3a). Compared with pristine films, the resistance of de-hydrogenated NSNO films with a reversed voltage applied was slightly higher, similar as observed in the resistance regulation of SNO and NNO. [5,10] A possible reason is that a small amount of hydrogen ions remain in the films after the reversed voltage is applied, or defects are introduced during ionic liquid gating. The nondoped NNO films exhibit a characteristic first-order phase transition from high-temperature paramagnetic metal to low-temperature charge-disproportional antiferromagnetic insulator, [22] whereas NSNO films maintain the metallic behavior over a full temperature range (2 K-300 K) ( Figure S1a, Supporting Information). At a positive gating voltage, the NNO films changed from metallic to semiconducting behavior and the T MIT decreased ( Figure S1b and inset, Supporting Information), which is consistent with previously reported experimental results. [10] Since hydrogen doping will localize the electrons, their effect of resistance switching thus is severely suppressed in insulating phase. Taking the NNO at 2 K as an example, the hydrogenated NNO had a switching ratio r = [R(H + )/R 0 ] of only 20, while NSNO exhibited a ratio of 5300 at the same ionic gating condition ( Figure 3b). The insulating phase at low temperature in NNO leads to a reduced switching ratio. In strong contrast, the switching ratio increases with reducing temperature in NSNO, e.g., resistance switching at 2 K is nearly two orders of magnitude higher than that at 300 K. Enhanced resistance switching at low temperature is of great interest for research in the field of low-temperature devices. [23,24] The unique monotonic increasing switching ratio with decreasing temperature in NSNO can be well understood by the opposite behavior between insulating and metallic phases. Since resistance of insulating/ metallic phases will increase/decrease with decreasing temperature, respectively, their ratio (R I /R M ) thus will monotonically increase with reducing temperature.
The variation of NSNO films switching ratio with gating time under various gating voltages and its comparison with NNO films are summarized in Figure 3c. With a gating voltage of +3 V, the switching ratio of NSNO film (r = 5200) is 26 times larger than that of the NNO film (r = 200) when the gating time is 20 min. When the gating time is 30 min, switching ratio of NSNO film (r = 48 000) becomes 37 times larger than that of the NNO film (r = 1300). The higher switching ratio in NSNO films can be attributed to the higher Ni valence state introduced by Sr doping. Higher Ni valence is less stable and this chemical instability makes nickelates be easier changed to a lower valence state under electric field control. The NSNO film resistance varies by about five orders of magnitude at +3 V at room temperature during ILG. Such colossal resistance switching is also much larger than another well-known VO 2 -based ionic gating device. Resistance modulation of VO 2 by hydrogen doping during ionic gating process is found to be below three orders of magnitude. [25] As the gating time increases, the slope of the current-voltage (I-V) curve of the sample gradually decreases, indicating an increase in resistance ( Figure S2a  between the source and drain, the change in resistance of the film with the gating time can be in situ measured, showing the same behavior of colossal resistance switching ( Figure S2b, Supporting Information).
Magnetoresistance (MR) results of NSNO film are consistent with the hydrogen doping scenario. Figure 3d shows the MR [R(H)-R(0)]/R(0) measured in a magnetic field applied perpendicular to the NSNO thin films at a temperature of T = 2 K with +2 V gating voltage. Pristine NSNO films exhibit parabolic dispersion in MR, confirming that this is a conventional scattering transport regime. A significant enlarged amplitude is observed in hydrogenated NSNO films, which can be explained by the enhanced disorder scattering due to H intercalation. The enhanced disorder will lead to strong localization of carriers and thus contribute to negative MR.
To investigate the effect of gating voltage on the resistance switching, we compare the resistance at various voltages when the gating time was maintained at 20 min (Figure 3e). The NSNO film resistance was essentially unchanged at negative voltages. This is consistent with the results that the characteristic of the R-T curves of the NNO film was essentially unchanged at negative voltages and that the T MIT was only slightly reduced. [10,26] In the case that the gating voltage was +1 V, the NSNO film resistance increased slowly, which may come from the formation of oxygen vacancies since the gating voltage is less than the electrolysis voltage of water (≈1.23 V) and thus impossible to generate hydrogen ion. When the gating voltage was +3 V, a huge resistance switching was achieved, which should be mainly due to hydrogen doping as illustrated by Figure 2. To further confirm this assumption, we mixed the ionic liquid with 1 mol L −1 NaOH for regulation so that H + generated by electrolysis would be neutralized by OH − , thus inhibiting the insertion of H + into the NSNO film ( Figure S3a, Supporting Information). Without H + in the ionic liquid, it is found that the resistance changed by only an order of magnitude after regulation, which is thus much smaller than normal ionic liquids devices ( Figure S3b, Supporting Information). Therefore, it can be concluded that hydrogen doping dominates resistance switching of NSNO films, while oxygen nonstoichiometry may have a much less sensitive effect on its resistance regulation. It is worth noting that the resistance switching is closely related to the gating temperature, and the resistance switching increased significantly as the gating temperature increased ( Figure S4  with temperature under the same conditions. The switching ratio of NSNO at 2 K is nearly two orders of magnitude higher than that at 300 K, while the switching ratio of NNO at low temperature is lower than that at 300 K. c) Variation of switching ratio with gating time at different gating voltages. Compared with NSNO, the regulated NNO switching ratio is more than an order of magnitude smaller than NSNO under the same conditions. d) Magnetoresistance variation of the pristine and hydrogenated NSNO. e) Variation of resistance with gating voltage at a gating time of 20 min. Only a small change in resistance occurs below the electrolytic voltage of water. f) Optical transmittance spectra of the pristine and hydrogenated NSNO. Inset shows the difference in their photographs as seen by eye.
gating temperature of 90 °C, the as-gated film resistance was more than two orders of magnitude higher than that at 30 °C, which may attribute that the elevated temperature promotes the electrolytic production of hydrogen ions and the diffusion of hydrogen ions in the film. Figure 3f shows optical transmittance measurements in the visible and near-infrared regions. These measurements demonstrate a significant modulation of optical transmittance ≈25%-30% at visible wavelengths between the pristine and hydrogenated NSNO films. This process is accompanied by the transformation of the highly reflective metallic phase into the highly transmissive insulating phase due to a significant increase in the band gap. [12,27] This modulation can also be directly observed by eye (Figure 3f inset), with the words barely visible underneath the pristine NSNO sample (left) and clearly seen under the hydrogenated sample (right). Such strong electrochromic effect can be applied in the smart windows design. [28][29][30]

Pronounced Lattice Expansion
To reveal the crystal structure evolution during ILG, highresolution X-ray diffraction (XRD) was used to characterize the evolution of (002) diffraction peaks of NSNO films with gating time when the gating voltage was +2 V. Figure 4a depicts that the NSNO film peak position gradually shifts to the left as the gating time increases, indicating a gradual expansion of lattice. Details of the lattice structures are further illustrated by reciprocal space mapping (RSM) of (-103) peak as shown in Figure 4b. The in-plane Q x of the NSNO films share the same value with that of STO while the Q z of the NSNO decreases and moves toward STO as the gating time increasing. This result reveals that the NSNO films were always coherently strained to the underlying STO substrates with identical in-plane lattices, whereas a significant expansion of out-of-plane lattices was induced in NSNO films due to hydrogen doping. The RSM result is consistent with the experimental results of lattice expansion from XRD theta-2theta scan. The value of c-axis lattice constant determined from XRD is shown in Figure 4c. The c-axis lattice constant increases from 3.78 Å in the pristine film to 3.91 Å after ILG. Therefore, a pronounced lattice expansion by 3.4% was realized. The full-width half-maximum (FWHM) of the rocking curve after gating for 10 min and 20 min are both 0.03°, indicating that the NSNO films maintained high crystalline quality throughout ILG ( Figure S5a,b, Supporting Information). With a gating voltage of +1 V, the film peak slightly shifts to the left. The lattice modulation rate is significantly faster at a gating voltage of +3 V than at +2 V ( Figure S5c,d, Supporting Information). Upon applying a gating voltage of +2 V (Figure 4d), the X-ray reflectivity (XRR) results show a noticeable decrease in their oscillation period from a substantial expansion of film thickness, being consistent with the observed increase in the c-axis lattice constant. The significant expansion of film thickness after hydrogenation also proves that the NSNO films were not obviously degraded during ionic liquid gating. Besides, in the regulation of the SNO films, no Adv. Electron. Mater. 2023, 9,2300116  obvious corrosion was found on the surface of the SNO films according to cross-sectional conductive AFM. [5]

Electronic Structure Changes and Orbital Polarization Suppression
To investigate the electric structure change of Ni-sites during ILG, we carried out X-ray absorption spectroscopy (XAS) at the Ni L 2,3 edge and O K-edge. Figure 5a shows the Ni L 2,3 edge XAS of NSNO films as a function of the degree of ILG. A distinct shoulder at ≈853 eV is observed in pristine NSNO films, which corresponds to the electron transition from the Ni corelevel 2p to 3d 8 L state (L is ligand hole), beside the main sharp peak at ≈852 eV corresponding to the electron transition from the core-level 2p to 3d 7 state. [31] The shoulder peak intensity gradually decreases with the increase in the degree of ILG, consistent with the decrease in Ni valence caused by hydrogen doping. Interestingly, a straightforward comparison with the reference NiO spectrum confirms that after a gating at +3 V for Figure 5. Electronic structure characterization. a) XAS of the Ni L 2,3 -edge of pristine and hydrogenated NSNO films. As the degree of hydrogenation increases, the main sharp peak moves to the left, and the intensity of the shoulder peak gradually decreases. XAS spectrum of NiO standard sample (gray dashed line) is also listed for comparison. b) Split peak fit result for the Ni L 3 -edge of pristine NSNO films. c) Relative content of "α", "b" and "g" in the NSNO films as a function of hydrogenation degree. d) XAS of pristine and hydrogenated NSNO films for the Ni L 3 -edge measured by the quasi-out-of-plane (E//c) and in-plane (E//ab) linearly polarized X-ray beams. Left insets show that the c-axis is clearly stretched during ILG. Right inset shows the schematic diagram of the scattering geometry for XAS and XLD measurements, with the X-ray beam detecting in-plane and quasi-outof-plane orbits. e) XAS curves of the O K-edge of pristine and hydrogenated NSNO. The oxygen K-edge results show a full suppression of hybridization between oxygen 2p orbitals and the 3d orbital of nickel, and the appearance of features associated with hydroxyl bonds in hydrogenated NSNO. f) Relationship between the resistance of the hydrogenated NSNO films and the area of O K pre-peak ε. 30 min, the valence state of Ni in the regulated NSNO films is dominated by Ni 2+ , which indicates a high degree of hydrogenation in this situation, consistent with the result of a colossal increase in resistance. Three chemical contributions of Ni-sites could be seen after the spectra deconvolution: the low-energy features, which are labeled as "α", arise from the 3d 8 state; the middle-energy peaks, which are labeled as "b", originate from the 3d 8 L state and the high-energy peaks labeled "g" are from the form 3d 8 L 2 state (Figure 5b and Figure S6, Supporting Information). [32][33][34][35] It is clear that the feature "α" increases monotonically, and the feature "b" and "g" both monotonically decrease as the degree of hydrogenation increases (Figure 5c). These facts prove that the valence state of Ni-sites in NSNO films gradually decreases with hydrogenation.
X-ray linear dichroism (XLD) was measured to better understand the changes in Ni-sites orbital configuration during ILG and impact of lattice expansion on electronic structure. The XLD was performed by calculating the intensity difference (I E//c − I E//ab ) of the photon polarization parallel to the quasi-outof-plane (E//c) and in-plane (E//ab) with incident X-ray beam angles of 20° and 90°, respectively (Figure 5d. right inset). [36] For the pristine NSNO films, the c-axis shrinkage revealed by XRD (c/a = 0.97) clearly leads to an increase of the d z r − 3 2 2 orbital energy (Figure 5d. left inset). [37] The absorption for polarization perpendicular to the ab plane is shifted ≈0.1 eV higher in energy than the in-plane polarization absorption (Figure 5d). The integrated XLD intensity is positive for the tensile strained NSNO films, indicating preferential occupation of the d x y − 2 2 orbital ( Figure S7, Supporting Information). [38,39] After hydrogenation, the lattice (c/a = 1) of NSNO films expanded and became close to a cubic structure. As the result, the orbital polarization was severely suppressed (Figure 5d).
A prominent pre-peak at ≈528 eV is observed near the O K-edge XAS in pristine NSNO films, which is attributed to the presence of a ligand hole in oxygen, as shown in Figure 5e. [31,40,41] In perovskite nickelates, the charge transfer energy from the oxygen p to the nickel d band is negative and electrons are spontaneously transferred from the oxygen ligand to the Ni cation, and holes remain on the oxygen side even without chemical doping. [31] During ILG, hydrogen will combine with apical oxygen of NSNO octahedron, thereby weakening the hybridization between Ni-O, which is consistent with the gradual weakening and disappearance of the prominent pre-peak intensity. We also note that distinct XAS features around 540 eV were observed, which are known to indicate the presence of hydroxyl groups. With the increase of the hydrogenation degree, the area of the pre-peak gradually decreases, indicating that a less Ni-O hybridization, and thus the electron is more localize (Figure 5f). This observed reduced pre-peak intensity thus further explains the origin of dramatic increase of resistance with hydrogenation.

Conclusion
In summary, we have reported that full temperature range (2 K-300 K) colossal resistance switching and pronounced lattice expansion can be achieved in NSNO films by ILG. Over five orders of magnitude in the resistance modulation and lattice expansion by 3.4% were demonstrated, much larger than nondoped NNO films. Compared with NNO films, NSNO films can exhibit faster resistance switching behavior and colossal resistance modulation over the entire temperature range, reflecting the significant modification of their material properties by A-site doping. Our SIMS unambiguously illustrated that the hydrogen doping was throughout the entire NSNO films. What is more, XAS results further reveal the reduced valence and smaller Ni-O hybridization in hydrogenated NSNO, explaining the origin of colossal resistance switching in NSNO. Concomitant with enhanced c/a ratio from 0.97:1 of NSNO toward cubic perovskite of 1:1, the orbital polarization is also severely suppressed. Our results suggest a route to engineer device performance by A-site doping and pave a path toward practical application of nickelate perovskite oxide electronics.

Experimental Section
Synthesis of Nd 0.8 Sr 0.2 NiO 3 and NdNiO 3 : The atomically flat TiO 2terminated SrTiO 3 (001) substrates were obtained by buffered hydrofluoric acid chemical etching and subsequent annealing at 950 °C for 90 min. The Nd 0.8 Sr 0.2 NiO 3 films were grown at a growth temperature of 650 °C and an oxygen pressure of 0.1 mbar. The energy density of the krypton fluoride excimer laser (λ = 248 nm) was fixed at 2.3 J cm −2 of the target with a repetition rate of 4 Hz. The samples were annealed for 10 min at 650 °C and 0.1 mbar oxygen after deposition, and then cooled down to room temperature at a rate of 10 °C min −1 in 200 mbar oxygen. The NdNiO 3 films were grown at T g = 620 °C and p O 2 = 0.2 mbar, using 2 J cm −2 laser fluence, with a repetition rate of 4 Hz. Both Nd 0.8 Sr 0.2 NiO 3 and NdNiO 3 films thicknesses were controlled ≈36 nm by RHEED monitoring during the growth, and confirmed with X-ray reflectometry measurements.
Ionic Liquid Gating: Both sides of the film were painted with a gold conductive adhesive prior to ionic liquid gating. Subsequently, conductive gold wires on probe were fixed on the conductive electrode by adjusting the movement of the three-dimensional transfer stage, and a screwed platinum wire was used as the gate electrode. The ionic liquid DEME-TFSI was heat treated in a vacuum oven at 120 °C for 48 hours before use. Then, a small drop of ionic liquid was added to cover both the film surface and the screwed platinum wire. The gating voltage was set to 0 V at the start, and ramped to the desired value over ≈1 min. After gating, the films were rinsed several times with acetone and isopropanol to remove the ionic liquid residue.
Characterization: The AFM images were acquired in tapping mode. X-ray diffraction (XRD) measurements were performed by using a high-resolution diffractometer using monochromatic Cu K α1 radiation (λ = 1.5406 Å). Optical transmittance spectra were taken in air at roomtemperature with spectrophotometers, which cover the visible and nearinfrared range with wavelengths between 400 nm and 2200 nm. Optical photos are taken by the camera equipped on the microscope. The wire connection for the electrical transport measurement was made by Al ultrasonic wire bonding. The electrical transport measurements was performed using a quantum design physical property measurement system. The current versus voltage (I-V) curve and in situ resistancegating time curve of the sample are measured by the source measure unit (Keithley 2400). During the secondary-ion mass spectrometry (SIMS) measurements, the argon-ion beam (3 KV) sputtering area was ≈400 × 400 µm 2 ; data were collected only in an area of 100 × 100 µm 2 within that region, in order to avoid disturbance from the crater edge. The X-ray absorption spectroscopy (XAS) and X-ray linear dichroism (XLD) measurements were performed using linearly polarized X-ray at XMCD beamline of National Synchrotron Radiation Laboratory (NSRL), using a total electron yield (TEY) detection method. By rotating the sample, the XLD shown in the main text was performed by calculating the intensity difference (I E//c − I E//ab ) of the photon polarization parallel to the quasi-out-of-plane (E//c) and in-plane (E//ab) with incident X-ray beam angles of 20° and 90°, respectively.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.