Nd1-xSrxMnO3 is a well-known manganite due to close connection among structure, transport, magnetism, and chemistry. Thus, it would be an ideal system to study the modification of physical properties by external stimuli including control of stoichiometry in growth. In this work, we show that an abrupt change of electronic and magnetic properties can be achieved by a subtle change of oxygen partial pressure in pulsed laser deposition. Interestingly, the pressure indeed modulates cation stoichiometry. We clearly observed that the films grown at 140 mTorr and higher showed clear insulator to metal transition and stronger magnetism, commonly found in less hole doping, while the films grown at 130 mTorr and lower showed insulating behavior and weak magnetism. From soft x-ray spectroscopic methods, we clearly observed the compositional difference in those thin films. This result is further supported by scattering of lighter elements in high oxygen partial pressure but not by anion deficiency in growth.
Recent advancement in pulsed laser deposition (PLD) has opened a new horizon in material research such as the stabilization of metastable materials,1,2 formation of atomic scale heterostructures,3–7 and elucidation of anisotropic properties.8–10 Advantages of PLD are its fastness in growth, non-equilibrium growth, capability in atomic-level film synthesis, and stoichiometric transfer of target materials in many cases. However, in some cases, it has been reported that non-stoichiometric transfer of target materials can happen during the PLD process, since the PLD process is easily complicated by gas pressure, laser fluence, substrate temperature, angle between plume and substrate, etc.11–15 Specifically, it is known that the gas pressure in a certain range can lead to multiple scattering of ablated lighter elements in multicomponent targets. However, this non-stoichiometric transfer may be another mean to tune materials' properties, if we can deliberately control such non-stoichometry.15–19
In this regard, hole-doped manganites would be an ideal model system, since they have received great attention for decades due to close links among transport, magnetism, structure, and chemistry. Among them, we chose Nd0.5Sr0.5MnO3 as our model material,20 since multiple electric and magnetic phases such as paramagnetic insulating (PM-I), ferromagnetic metallic (FM-M), and antiferromagnetic charge-ordered insulating (AFM-COI) phases can coexist depending on hole doping, temperature, and magnetic fields. Even though the stabilization of the AFM-COI phase by pulsed laser deposition has been reported,21,22 stabilization of Nd0.5Sr0.5MnO3 is a formidable task, since a slight change of hole doping or strain can tip toward completely different ground states.23 For the case of strain, it was reported that strain can change transport properties from the study of Nd0.5Sr0.5MnO3 films on (001) SrTiO3 and (011) SrTiO3.24 In the previous report, the differences in epitaxial strain drive the differences in transport properties from those of the single crystals. They claimed that the films on (011) SrTiO3 have a more degree of freedom to keep their original crystal structure.24 In terms of hole doping, Yoshizawa et al.25 reported that the phase with less hole doping is ferromagnetic metallic, while the AFM-COI phase is stabilized in the more hole-doped case. However, the attempt to modulate hole doping is rare in growth. Taking advantage of phase competition in Nd1-xSrxMnO3 (NSMO) would provide a chance to tailor materials' properties in thin film techniques, since the cation composition can be modulated by fine control of the oxygen partial pressure (PO2).26–31
In this work, we show the modulation and possible control of cation stoichiometry via PLD. Using detailed electronic and magnetic characterization and two soft x-ray spectroscopic methods, we unambiguously demonstrate that one can tune material properties through the modulation of A-site cation stoichiometry.
About 20-nm-thick epitaxial NSMO thin films were grown by PLD (Q-switched pulsed Nd:YAG laser with λ = 355 nm) with a laser fluence of 0.9 J/cm2. We used a nominal Nr0.5Sr0.5MnO3 target from TOSHIMA, Japan and (001) (LaAlO3)0.3-(SrAl0.5Ta0.5O3)0.7 (LSAT) substrates for the synthesis. Note that the lattice mismatch between Nd0.5Sr0.5MnO3 and LSAT is 0.27%. To control cation stoichiometry, we used different PO2 in growth. PO2 was changed from 10 mTorr to 200 mTorr. After the growth, we applied 300 Torr of PO2 to the growth chamber and then cooled down the samples to remove possible oxygen vacancies. Structural properties of all the thin films were characterized by using a high resolution x-ray diffractometer (HR-XRD, SmartLab, Rigaku) and synchrotron-based XRD (λ = 1.23983 Å) in the 3D beamline of Pohang Accelerator Laboratory (PAL). After checking the structural properties of our NSMO thin films, we measured temperature dependent resistivity. A conventional four probe method with the delta scheme was used in a closed cycle refrigerator with the constant ramping rate (2 K/min). We fixed the current to 100 nA due to possible current and/or electric field effects.9,32 To compare magnetic properties, we measured temperature dependent magnetization and field dependent magnetization using a SQUID magnetometer (MPMS 3, Quantum Design). To elucidate the possible origin of the different electric and magnetic properties, we adopted x-ray absorption spectroscopy (XAS) in the 2A beamline of PAL to measure chemical states of Mn and O from the Mn L-edge and O K-edge with normal incidence by the total electron yield (TEY) mode. In addition, we used x-ray photoelectron spectroscopy (K-Alpha+, Thermo scientific) to compare the Sr/(Nd + Sr) ratio of each NSMO thin film. We observed the Sr 3d edge and Nd 3d edge of NSMO and compared the A-site cation ratio by integrating each peak.
To find optimal growth condition for NSMO thin films, we grew NSMO thin films in a wide range of temperatures and oxygen partial pressures. Note that we fixed other growth parameters such as laser fluence, repetition rate, and the number of laser shots. In all the range of pressure and temperature (see Fig. 1), we observed the formation of single-phased epitaxial NSMO thin films from XRD. However, interestingly the electronic properties are very different between the films grown at 100 and 200 mTorr. To find the critical oxygen pressure to separate the electronic properties, we performed the growth condition optimization by varying PO2 between 100 and 200 mTorr, while the temperature was fixed as 600 °C. We chose 600 °C as growth temperature, since the crystallinity of the films is not much varied in this growth pressure regime. By transport and x-ray diffraction, we observed that the NSMO thin films grown at 130 mTorr and below are insulating, while the films grown at 140 mTorr and above undergo insulator to metal transition (IMT). Note that the NSMO film grown at 140 mTorr undergoes IMT at different temperatures (TC = 175 K), while the IMT transition temperatures of NSMO films grown from 150 mTorr to 200 mTorr are similar to each other. The film grown at 140 mTorr is likely to be in the proximity of both regions (supplementary material). So, we focused to understand how the oxygen pressure affects the creation of electronically different phases, and we chose two thin films such as NSMO films grown at 100 mTorr and 160 mTorr from different regimes for the in-depth studies on structural, magnetic, electronic, and chemical properties.
Growth phase diagram of epitaxial NSMO thin films. In all the attempted temperatures and PO2, single phased thin films were formed. Red dots indicate that the film shows insulating behavior, while blue dots show that the film undergoes insulator-metal transition.
Growth phase diagram of epitaxial NSMO thin films. In all the attempted temperatures and PO2, single phased thin films were formed. Red dots indicate that the film shows insulating behavior, while blue dots show that the film undergoes insulator-metal transition.
Since the physical properties of manganite are easily affected by mechanical stimuli,33,34 we performed x-ray diffraction scans on the films. To observe the structural difference of NSMO films by the PO2 difference, we measured diffraction patterns around (002) LSAT at the 3D beamline in PAL. Figure 2 shows the diffraction peak of the (002) NSMO films grown at 100 mTorr and 160 mTorr. (002) diffraction peaks of NSMO are observed at 37.95° and 37.91°, respectively. The lattice constants are 3.81(3) Å for the film grown at 100 mTorr and 3.81(7) Å for the film grown at 160 mTorr. Since the bulk lattice constant of NSMO is 3.8399 Å in pseudocubic notation,20 the reduced lattice constants result from the effect of substrate-induced tensile strain. The insets of Fig. 2 show reciprocal space maps (RSMs) of the films at the 103 reflections of the LSAT substrates. From the RSMs, it is clearly seen that the in-plan lattice constant is fully locked to that of the substrate. We could not see any difference in the structural properties. In other words, the effect of PO2 in this range is not significant enough to disturb the structural properties of NSMO thin films. Furthermore, clear Kiessig fringes in x-ray diffraction patterns show that roughness of both NSMO films is very low. Note that thickness of both films is 18 nm, determined from Kiessig fringes. Then, crystallinity of the two films was determined by rocking curve measurements. The FWHM of (002) NSMO thin films grown at 100 mTorr and 160 mTorr is 0.052° and 0.054°, respectively. These results show that structural defects and dislocations are not causing the changes in electronic and magnetic properties.
θ-2θ x-ray diffraction patterns of NSMO thin films grown at (a) 100 mTorr and (b) 160 mTorr of PO2. Clear Kiessig fringes indicate that atomically flat surfaces are formed. Each inset shows reciprocal space mapping at around (103) LSAT substrate. It shows that each film is fully strained.
θ-2θ x-ray diffraction patterns of NSMO thin films grown at (a) 100 mTorr and (b) 160 mTorr of PO2. Clear Kiessig fringes indicate that atomically flat surfaces are formed. Each inset shows reciprocal space mapping at around (103) LSAT substrate. It shows that each film is fully strained.
To check the change in physical properties due to a slight difference in PO2, we measured temperature dependent resistivity of the NSMO thin films. Both NSMO thin films show similar resistivity at room temperature as seen in Fig. 3(a). However, at lower temperature, the NSMO thin films grown at 100 mTorr show insulating behavior down to 11 K, which is similar to the case of NSMO thin films on (001) SrTiO3 from the previous report.24 From this reason, the insulating behavior shown in NSMO grown at 100 mTorr is likely due to the effect of biaxial strain in Nd0.5Sr0.5MnO3. However, the NSMO thin film grown on 160 mTorr undergoes an IMT at around 210 K. Since it happens by a simple change of PO2, we rather focused on a possible change of stoichiometry as an origin of the difference. There would be two possible origins: change of cation stoichiometry and/or change of anion stoichiometry. However, we can easily rule out anion stoichiometry by scrutinizing the phase diagram of bulk Nd1-xSrxMnO3 (NSMO),35 since the NSMO with less hole doping from half-doped NSMO only undergoes an IMT transition. In this regard, the first scenario requires that our NSMO film grown at higher PO2 should have a lower Sr concentration than 0.5, while the second requires that the same film has more oxygen vacancies. Since it is unlikely to have the higher level of oxygen vacancies from the film growth at the higher PO2, we ruled out our second scenario. In addition, this can be further supported by our quenching method at an oxygen pressure of 300 Torr. Thus, we hypothesized that this change of transport behavior is caused by the change of cation stoichiometry in the film grown at higher PO2. Note that the film showed inflection points, where d2log[ρ]/dT2 is zero, at around 210, 160, and 50 K. The origin of each inflection point will be explained later. To check how the magnetic properties are affected by PO2, Figure 3(b) shows temperature dependent magnetization [M(T)] curves of NSMO thin films. To see the phase transition temperature correctly, we used 100 Oe. The magnetization of the NSMO thin film grown at 100 mTorr starts to increase below 160 K. This result implies that the FM-M phase is formed at around 160 K, while the inflection point at 210 K appears possibly due to the creation of the AFM-COI phase, as seen in the case of NSMO on (001) SrTiO3.24 However, ferromagnetism in the NSMO film grown at 160 mTorr arises below 220 K. Note that this magnetic phase transition was also found as an inflection point in temperature dependent transport measurements. From both magnetic and transport results, the IMT behavior from the NSMO thin film grown at 140 mTorr and higher could be explained by double exchange interaction.36–38 The IMT is triggered by the creation of the ferromagnetic-metallic (FM-M) phase and its percolation by merging the phases. At 220 K, FM-M phases are generated, so that magnetization starts to increase, but the lack of the FM-M phase shows that NSMO is insulating. Below 210 K, the FM-M fraction is enough to merge the percolation paths, so that NSMO becomes metallic. However, the behavior of the NSMO thin film grown at 130 mTorr and below can be explained by a stabilization of more non-ferromagnetic phases. The formation of the AFM-COI phase at low temperature and insufficient amount of ferromagnetic phase in the thin films are likely the reason for insulating behavior down to low temperature. The lower magnetic moment and higher coercivity from magnetic hysteresis curves [see Fig. 3(c)] and inflection point at around 50 K [see Fig. 3(a)] are possible signature of the phase coexistence. If we assume total magnetic moment of Nd0.5Sr0.5MnO3 is from electron spins, 3.5 μB/Mn is expected. However, saturation magnetic moment of NSMO grown at 100 mTorr is 0.9 μB/Mn. This infers that the FM-M phase and other insulating phases coexist. From this assumption, we expect 26% of FM-M phase with other insulating phases including the AFM-COI phase. Furthermore, in Fig. 3(c), saturated magnetic moment of NSMO grown at 160 mTorr is 1.7 μB/Mn. It is about 80% larger than NSMO grown at 100 mTorr whose saturated magnetic moment is 0.9 μB/Mn. This difference of saturated magnetic moment is confirmed in bulk NSMO.39,40 A change of small cation stoichiometry makes an increase or a decrease of large level saturated magnetic moment. It is worth noting that we also observe the possible effect of biaxial strain. When we compared NSMO thin films on (001) LSAT and on (001) SrTiO3,24 magnetic moment of Nd0.5Sr0.5MnO3 on (001) LSAT is larger than that of the Nd0.5Sr0.5MnO3 thin film on (001) STO at 10 K and 0.5 T. This infers that the tensile strain may reduce the stabilization of the FM-M phase.
(a) Resistivity vs temperature curves of NSMO thin films grown at different partial pressures. (b) Magnetization vs temperature curves measured in 100 Oe. (c) Magnetic hysteresis curves at 10 K. The result clearly shows a huge difference in coercive field and magnetization values.
(a) Resistivity vs temperature curves of NSMO thin films grown at different partial pressures. (b) Magnetization vs temperature curves measured in 100 Oe. (c) Magnetic hysteresis curves at 10 K. The result clearly shows a huge difference in coercive field and magnetization values.
To elucidate the origin of the changes in the physical properties, we adopted two spectroscopic tools. First, to see the valence states of Mn and hybridization between Mn and O, XAS was used in the total electron yield (TEY) mode. XAS is a powerful method to see chemical information of materials due to its element-specific excitations.41,42 Mn L2 and L3-edge spectra of NSMO thin films imply the absorption energy of electrons, when the core electrons located in Mn 2p orbitals are excited to outer empty orbitals such as 3d orbitals. Mn L-edge spectra are shown in Fig. 4(a). When we compared the peak positions of the Mn L3-edge of the NSMO thin films, it is clearly seen that the peak position of the NSMO thin film grown at 100 mTorr is shifted about 0.2 eV to the higher energy compared to that of the NSMO thin film grown at 160 mTorr. This implies that the valence state of Mn in the NSMO thin film grown at 100 mTorr is higher than that of the NSMO thin film grown at 160 mTorr.43,44 In a similar system such as La1-xSrxMnO3, when x increases to 0.9, the maximum of the peak shift is 0.9 eV.42 Therefore, the Mn valency in the NSMO thin film grown at 160 mTorr seems to be lower by 0.2 from that of the NSMO thin film grown at 100 mTorr. Another signature of the higher valency of Mn in the NSMO film grown at 100 mTorr is the pronounced shoulder peak at around 641 eV. It indicates that the Mn valence state shifts toward a higher valence state.42–44 These facts are indirect evidence of our hypothesis about lower divalent cation content in the thin film grown at the higher PO2. After confirming the change of valence states in Mn, the O K-edge spectra of NSMO thin films are measured in Fig. 4(b). Energy ranges from 527 to 532 eV and from 535 to 540 eV are related to hybridization between Mn3d-O2p and hybridization between Nd5d-O2p and Sr4d-O2p.45,46 From the peak between 527 and 532 eV, we observed that the peak position of the NSMO thin film grown at 100 mTorr is lower than that of the NSMO thin film grown at 160 mTorr, while the peak intensity is higher. These results are similar to the previous results in La1-xSrxMnO3 reported by Abbate et al.42 According to these results, we unambiguously conclude that the NSMO thin film grown at 100 mTorr has a higher valence state.42 It is worth emphasizing that the higher valency in the film grown at lower PO2 is possibly caused by modulation of the A-site cation ratio at the higher PO2. If the oxygen content affects the Mn valence state, the valence state of Mn in the oxide grown at higher PO2 must be higher.
(a) Mn L-edge spectra and (b) O K-edge spectra measured from XAS. The inset shows the magnified view of the Mn L3-edge from 642 to 645 eV. The maximum of the peak position of the NSMO film grown at 100 mTorr of PO2 is located at the higher energy than that of the NSMO film grown at 160 mTorr.
(a) Mn L-edge spectra and (b) O K-edge spectra measured from XAS. The inset shows the magnified view of the Mn L3-edge from 642 to 645 eV. The maximum of the peak position of the NSMO film grown at 100 mTorr of PO2 is located at the higher energy than that of the NSMO film grown at 160 mTorr.
To directly compare the difference in A-site cation stoichiometry, we used x-ray photoelectron spectroscopy, which can penetrate about 10 nm. Before XPS measurements, we etched the surface of the films about 1 nm by Ar etching to eliminate contaminated surface layers and reduce the contribution of carbon on XPS data. Note that the etching depth was calibrated from a prior experiment on SiO2. Figure 5 shows XPS results of the NSMO thin films. Figure 5(a) shows the Sr3d edge spectra of NSMO thin films, which were measured at 130∼137 eV. Each spectrum consists of SrO and SrCO3 peaks. Note that the SrO spectrum is from the lattice of the NSMO thin film,47 while the SrCO3 spectrum is due to possible re-contamination after etching of NSMO thin films.48 Etched Sr and O combine with carbon by getting enough energy to combine during surface etching. In order to compare the Sr content of each NSMO thin film, we integrated the underneath area of the SrO peaks. Figure 5(b) shows the Nd3d edge spectra of NSMO thin films measured at 970∼990 eV. Each spectrum shows the Nd2O3 and O KLL contributions. The Nd2O3 spectrum is from the lattice of NSMO thin films,49 while the O KLL spectrum came from oxygen Auger election.50 In order to compare contents of Sr and Nd of each NSMO thin film, we integrated the area underneath SrO and Nd2O3 peaks and calculated the Sr/(Nd + Sr) ratio using each area. From XPS, the change in the Sr content is likely the reason for the change in the Mn valency. However, it is also possible that cation vacancies, which cannot be clearly observable, would contribute to the changes in Mn valence states. Figure 5(c) shows that NSMO thin films grown at higher PO2 have relatively lower strontium contents. This XPS result clearly supports the explanation that indeed more scattering of the lighter element like Sr in the higher PO2 leads to stabilization of the less hole doped manganite thin films.
(a) Sr 3d spectra and Nd 3d spectra from XPS of the both NSMO films. Circles indicate measurement data, while solid lines are from fitting. Magenta-, green-, violet-, and cyan-colored lines correspond to spectra of SrO, SrCO3, Nd2O3, and O KLL, which come from the oxygen Auger electron, respectively. (c) Ratio between integrated intensities of SrO and Nd2O3 + SrO of both the NSMO films. It is clearly seen that Sr/(Nd + Sr) of the film grown at 160 mTorr is lower than that of the film grown at 100 mTorr. It supports a lower Mn valence in the film grown at 160 mTorr.
(a) Sr 3d spectra and Nd 3d spectra from XPS of the both NSMO films. Circles indicate measurement data, while solid lines are from fitting. Magenta-, green-, violet-, and cyan-colored lines correspond to spectra of SrO, SrCO3, Nd2O3, and O KLL, which come from the oxygen Auger electron, respectively. (c) Ratio between integrated intensities of SrO and Nd2O3 + SrO of both the NSMO films. It is clearly seen that Sr/(Nd + Sr) of the film grown at 160 mTorr is lower than that of the film grown at 100 mTorr. It supports a lower Mn valence in the film grown at 160 mTorr.
In conclusion, we could stabilize electronically and magnetically distinct epitaxial manganite thin films. We take advantage of a subtle change of PO2 to modulate the cation stoichiometry and instability of electronic phases in hole doping. The differences in electronic and magnetic phase transitions were confirmed by various bulk measurements. In addition, from various spectroscopic methods, we could elucidate that the relative Sr content can be lowered, when PO2 is higher than 140 mTorr. From this result, we believe that we have developed another tuning knob to change the physical properties of manganite thin films.
See supplementary material for transition temperature depending on growth PO2.
This work was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (NRF-2015R1D1A1A02062175). AH and HNL were supported by the U.S. Department of Energy, Office of Science, Basic Energy Sciences, Materials Sciences and Engineering Division (magnetic characterization).
