Spatial control of the conductivity in SrTiO3-based heterointerfaces using inkjet printing

Interfaces between complex oxides host a plethora of functional properties including enhanced ionic conductivity, gate-tunable superconductivity and exotic magnetic states. The enhanced electronic, ionic and magnetic properties along the oxide interfaces are generally exploited in functional devices by spatial confinement of ions and electrons. Different patterning methods have been used to spatially control the conductivity at the interface, but a key limitation is the multiple steps needed to fabricate functional devices. In this investigation, inkjet printing of thermally stable oxides is introduced as an alternative pathway for spatially controlling the interface conductivity. We inkjet print yttrium-stabilized zirconia and TiO2 with various shapes and use these as physical masks to confine the electronic conductivity in SrTiO3-based heterostructures. By performing in-situ transport measurements of the electrical conductivity as LaAlO3 and γ-Al2O3 are deposited on SrTiO3, we witness the birth of the interface conductivity and find a consistent transient behavior as conductivity emerges in patterned and non-patterned heterostructures. We find that conductivity appears after the first laser pulse in the pulsed laser deposition corresponding to the film covering only a few percent of the substrate. We attribute the emergence of conductivity to oxygen vacancies formed by a combination of plasma bombardment and oxygen transfer across the interface during growth. In this vein, inkjet patterned hard masks protects the SrTiO3 substrate, effectively confining the conductivity. The study paves a scalable way for realizing energy devices with spatially controlled electronic and ionic interface conductivity.


Introduction
Oxide interfaces host a variety of interesting properties for energy and information technologies including enhanced ionic transport, gate-tunable superconductivity and exotic magnetic states.The properties of oxide interfaces play a crucial role in developing the next generation of energy-saving electronics based on spintronics or neuromorphic computations [1,2] as well as achieving high chemical stability and fast ionic conduction in energy materials [3].A key example is the numerous properties observed at the conducting interface between LaAlO 3 and SrTiO 3 , many of which are absent in the parent materials.Devices based on the LaAlO 3 /SrTiO 3 heterostructure feature, e.g.electrostatic on/off switching of superconductivity [4], electron pairing without coherent superconductivity [5,6] and electrically controllable ferromagnetism at room temperature [7].
Two strategies are typically employed when forming devices with spatially confined electronic or ionic conductivity at oxide interfaces: (a) Patterns in resist are defined using, e.g.UV or e-beam lithography and used to spatially control etching, thin film deposition or Ar + -ion irradiation [8][9][10].For instance, Trier et al used e-beam lithography followed by selective wet chemical etching and pulsed laser deposition (PLD) to produce hall-bar devices with electron mobilities of 8700 cm 2 Vs −1 at 2 K [11].(b) Local probes such as a conducting atomic force tip have been successfully used to induce conductivity or insulating regions.For conductive atomic force microscopy, lines of conductivity with a width down to 3 nm have been induced by scanning a positively charged tip on the LaAlO 3 or γ-Al 2 O 3 surface and reversibly erased using a negative charge [12][13][14].In both cases, however, multiple steps are needed.In the case of lithography, after spinning, exposure, baking and development, the patterned resist is incompatible with the typical growth temperatures for oxide films and deposition of a hard mask is generally needed [11,15,16].When using resist to spatially define the removal of material, special care should be taken to avoid modifying the properties of the remaining oxides [10].C-AFM offers a versatile platform for patterning, but a canvas defined with lithography is needed and the induced conductivity is unstable at room temperature [17] and can only be formed if the initial resistance of the interface is high prior to the c-AFM writing [12,14].
Inkjet printing has attracted a lot of attention in multiple areas as a versatile, and scalable method for forming patterned structures.The method is fast, automatized, cheap and can be used to coat a large range of materials [18][19][20].Inkjet printing is used in diverse areas ranging from biology with printed mammalian cells [21] and SERS substrates for molecular and chemical analysis [22], to energy research where it has been used to fabricate thin-film anodes in battery technology [23] and functional electronic materials [24].Inkjet printing has also been used to print various oxides including SrTiO 3 buffer layers [25], thick BaTiO 3 layers [26], as well as LaNiO 3 electrodes [27].
Here, we demonstrate how inkjet printing can be used as a fast method to pattern the conductivity at the heterointerfaces formed by depositing either LaAlO 3 or γ-Al 2 O 3 on SrTiO 3 .We inkjet print a thermally stable oxide hard mask preventing LaAlO 3 or γ-Al 2 O 3 to be deposited directly on SrTiO 3 in certain areas and use this to confine the metallic conductivity in the areas without inkjet printing.We further investigate the electronic properties in the patterned regions in-situ as they emerge during the PLD.

Inkjet patterning
Inkjet printing, as illustrated in figure 1, was performed by jetting droplets of ink with either dissolved TiO 2 or yttrium-stabilized zirconia (YSZ) onto a SrTiO 3 substrate in a predefined pattern.The printing is followed by a calcination process to obtain solid oxide structures.By controlling the jetting of the oxide ink spatially, the inkjet printing can produce various geometric structures as visualized in figure 2 with a total printing time on the order of a minute for a 5 × 5 mm 2 substrate.During the oxide inkjet printing, we used a 21.5 µm wide printer nozzle to jet TiO 2 droplets with a diameter of 35 µm (see inset in figure 1), which land on the SrTiO 3 substrate and form 40 µm solid dots after calcination.Variation of the nozzle size, the surface tension of the ink and substrate as well as the printer settings can yield different droplet sizes, which determines the resolution of the printed patterns [28].This represents an optimization process that ultimately could lead to nanometer-sized resolution if carefully perfected.Figure 2(a) depicts these dots formed by jetting single droplets with an inter-droplet distance of 200 µm.In figure 2(b), TiO 2 lines with a width of approximately 30 µm are printed onto a SrTiO 3 substrate by jetting interconnected droplets.The fairly straight lines show the potential for the ink droplets to form well-defined patterns.The result of jetting droplets to form large areas surrounding a Hall-bar shape without ink is shown in figure 2(c).In the case of TiO 2 , the printed area appear with several white points arising from the ink coalescing into an inhomogeneous, nanostructured layer rather than forming a homogeneous film during the solvent evaporation in the calcination process (see figure 2(d)).These imperfections in the printed layer can be avoided by optimizing the composition of the TiO 2 ink [28][29][30] or, as shown in figures 2(e) and (f), by printing YSZ [31].Here, the YSZ ink forms a homogeneous oxide film without micro-or nanometric cracks.

Confining conductivity
We further evaluated the capabilities of the ink to spatially confine the interface conductivity in SrTiO 3 heterostructures.The oxide ink was first spatially patterned on the SrTiO 3 surface such that ink covers the surface except for a region shaped as a hall-bar.The hall-bar was electronically connected with aluminum wires allowing in-situ measurements of the transport properties while depositing either LaAlO 3 or γ-Al 2 O 3 thin films on the patterned substrate at room temperature (see figure 3(a)).The room temperature condition results in the deposition of amorphous LaAlO 3 [32], while the γ-Al 2 O 3 layer becomes crystalline [33,34].For the deposition of LaAlO 3 displayed in figure 3(b), the resistance measured on the inkjet patterned Hall-bar is closely matching that observed when depositing directly on an unpatterned TiO 2 -terminated SrTiO 3 substrate.For the γ-Al 2 O 3 deposition, the overall trend is similar, but some variations are observed such as the inkjet patterned sample exhibiting a lower resistance of 4.4 kΩ/sq as displayed in figure 3(c).However, by comparing with the results of three nominally identical, unpatterned samples, we attribute the observed discrepancy in figure 3(c) to sample-to-sample variations in γ-Al 2 O 3 /SrTiO 3 .This is consistent with the sample-to-sample variations reported previously for the γ-Al 2 O 3 /SrTiO 3 heterostructure [33] where the lower carrier density at these interfaces may render more prone to sample-to-sample variations.
In both the patterned and unpatterned cases, the samples are insulating prior to the deposition with a sheet resistance higher than the measurement limit.High resistance is also found below the inkjet printed layer.However, in areas where the SrTiO 3 surface is exposed, the resistivity drops several orders of magnitude after the first few laser pulses, corresponding to a LaAlO 3 or γ-Al 2 O 3 coverage of only a few percent.The resistivity continues to drop after the first few laser pulses until it reaches a stable value on the order of 10 4 Ω/sq for LaAlO 3 /SrTiO 3 and, after an upturn, 10 5 Ω/sq for the γ-Al 2 O 3 /SrTiO 3 .In both cases, the saturated resistance is obtained after deposition of approximately 1 nm oxide layer on SrTiO 3 .The sudden drop in resistivity after only a few laser pulses is consistent with previous in-situ measurements [33], which stands in sharp contrast to ex-situ measurements where conductivity is only found above a critical thickness of more than 1 nm at room temperature [32,35,36].Insertion of oxygen into the growth chamber after three laser pulses was previously found to produce insulating LaAlO 3 /SrTiO 3 interfaces, whereas similar flushing of nitrogen was not degrading the conductivity [33].Therefore, we link the formation of conductivity to the emergence of oxygen vacancies in SrTiO 3 , which acts as an electron donor in SrTiO 3 -based heterostructures as proposed in other studies as well [14,[37][38][39].In-situ measurements further revealed a different temporal behavior of the conductivity in the case where LaSr 1/8 Mn 7/8 O 3 was deposited on SrTiO 3 under similar deposition conditions [33].Here, an initial drop of sheet resistance was followed by a large increase into a  highly resistive state after a 30% substrate coverage.The initial drop in resistance when depositing both LaAlO 3 , γ-Al 2 O 3 and LaSr 1/8 Mn 7/8 O 3 on SrTiO 3 is therefore likely a result of the kinetic energy of the bombarding particles that impact the substrate with tens of eV during growth [40] and facilitate removal of oxygen ions from the surface (figure 4(a)).Due to the clear difference in the resulting resistance after depositing γ-Al 2 O 3 , LaAlO 3 and LaSr 1/8 Mn 7/8 O 3 , the composition of the oxide thin film material plays a key role as well.In particular, the high oxygen affinities of the deposited γ-Al 2 O 3 and LaAlO 3 films facilitate the transfer of oxygen atoms from the SrTiO 3 substrate to the oxide thin film in contrast to the low oxygen affinity of LaSr 1/8 Mn 7/8 O 3 containing multivalent Mn ions that more easily can accommodate oxygen vacancies (figure 4(b)) [32,41,42].
The interface conductivity thus appears to be formed from a mixture of bombardment and an interface redox reaction, which are both countered by an oxidation process where oxygen-if present in the environment-fills up oxygen vacancies (figure 4(c)).The oxidation process takes place primarily below the ex-situ critical thickness where the SrTiO 3 surface is less protected by the thin film, explaining the difference between ex-situ and in-situ measurements [33].As apparent from both previous studies [14,32,33,43] as well as the saturation of the electronic properties after deposition of approximately 1 nm (see figure 3), these processes take place close to the SrTiO 3 surface.Hence, conductivity is formed on the exposed areas inside the hall-bar as well as on the unpatterned substrate whereas the inkjet printed film acts as a protective layer in regions outside the hall-bar, forming a high-resistive state underneath.Using two-terminal measurements on devices patterned with TiO 2 film, a resistance on the order of 40 MΩ was, however, detected when measuring at room temperature from inside the hall-bar to different places outside the hall-bar.For most purposes, this resistance is negligible compared to the typical two-terminal resistance of 4 kΩ at room temperature measured inside the hall-bar with a comparable distance between the contacts.On the other hand, in the case of the dense YSZ film, the two-terminal resistance was above measurement limit, and we thus attribute the leak resistance in the former case to the inhomogeneous TiO 2 film (figure 2(d)), which may form a high-resistive state through a percolating network of conducting lines.The depositions presented here were done at room temperature to simplify the in-situ transport measurements, but the YSZ ink also blocks the conductivity and yields insulating interfaces when depositing γ-Al 2 O 3 at 650 • C without any measurable leak resistance.
We further tested the thermal stability of the ink by heating a calcinated YSZ film to 1000 • C for 1 h with a ramp rate of 100 • C h −1 in an oxygen environment.The ink was thermally stable (see inset in figure 2(f)) and after a subsequent deposition of γ-Al 2 O 3 at 650 • C, the two-point resistance was above our measurement limit in regions covered by the annealed YSZ layer.This further supports the usability of the patterning approach in a broad range of temperatures typically used in epitaxial growth of oxide heterostructures.

Conclusion
Our study introduces inkjet patterning as a viable method for patterning oxide interfaces.It is found to provide a flexible, fast and cheap way to spatially confine conductivity on the micro-scale in SrTiO 3 -based heterostructures with a thermally stable hard mask approach that does not involve lift-off, etching or several PLDs.Even in the case of an imperfect TiO 2 ink that did not result in a fully dense layer after calcination, a high-resistive state was still formed underneath the ink with further potential for realizing 1D quantum physics, study percolating networks or constructing wearable strain sensors from metallic networks formed by a percolating growth template [44,45].Advances in stabilizing inkjet inks [28][29][30], scale-down of characteristic feature sizes as well as the formation of 3D structures through 3D printing further enhances the potential of using the present hard mask approach for forming patterned electronic and ionic conductivity with various 2D and 3D shapes [46,47].This allows for inkjet printing to be used for, e.g.control of optoelectronic properties via printing of repetitive conductive patterns, formation of quantum confinement, extraordinary magnetoresistive devices and various ionic or ionotronic devices [3,48].

Experimental
SrTiO 3 substrates were TiO 2 -terminated by ultrasonicating for 20 min first in milli-Q water then in acidic solution of 16:1:3 H 2 O:HNO 3 :HCl at 70 • C. The substrates were then washed in milli-Q water and annealed for 1 h at 1000 • C with a heating and cooling rate of 100 • C per hour.The TiO 2 ink was prepared by mixing titanium(IV) isopropoxide, MDEA, ethanol and water under argon fumes to prevent chemical reactions between the air humidity and the metal precursor as described by Gadea et al [28].The ink was inkjet printed using a Pixdro LP50 printer in various shapes leaving Hall-bars and other patterns exposed.The printed structures were calcined for 1 h at 400 • C in ambient atmosphere.The YSZ ink was made by mixing the precursor zirconium(IV) propoxide mixed with zirconium(III) nitrate hexahydrate dopant, MDEA, water and ethanol as described by Gadea et al [31].After inkjet printing of the YSZ layers, the samples were annealed in ambient atmosphere at a slow ramp rate (15 • C h −1 ) up to 500 • C with 1 h holds at 90 • C, 120 • C and 500 • C to burn off organic material and calcinate the YSZ.Afterward, either amorphous LaAlO 3 or crystalline γ-Al 2 O 3 thin film oxides were deposited onto the TiO 2 -terminated SrTiO 3 substrates containing either no ink or prepatterned TiO 2 or YSZ ink.During the deposition, in-situ measurements of the resistance were performed.This was done in a Hall bar geometry for the inkjet patterned samples and using the 4-probe van der Pauw method for the unpatterned substrates.In both cases, the samples were placed in a chip carrier holder with wedge-bonded aluminum wires creating the electrical contact to the sample.The depositions were done using PLD at room temperature at an oxygen background pressure of 3 × 10 −6 mbar.The ablation of the single-crystalline LaAlO 3 and Al 2 O 3 targets was done with a KrF laser (λ = 248 nm) using a repetition rate of 0.5 Hz, a laser fluence of 2.5 J cm −2 and a fixed target-substrate distance of 40 mm.The samples were analyzed by Zeiss Merlin FEGSEM microscope and a JEOL 7800F SEM.

Figure 1 .
Figure 1.Schematic of the inkjet patterning process.The nozzle (grey) moves in the horizontal plane jetting ink onto the substrate below it with spatial selectivity.The inset shows an image of a ink droplet jetted from the nozzle and falling towards the sample surface.

Figure 2 .
Figure 2. Scanning electron micrographs of inkjet printed oxides on a SrTiO3 substrate showing (a) a matrix of TiO2 dots, (b) interconnected dots forming straight lines of TiO2, (c) an ink layer surrounding a hall-bar pattern, (d) high-magnification view of a TiO2 inkjet printed film, (e) high-magnetification view of an YSZ inkjet printed film and (f) the border region of an YSZ inkjet film printed on SrTiO3.The inset shows a border region after annealing at 1000 • C for 1 h in oxygen.

Figure 3 .
Figure 3. (a) Schematic of the in-situ measurement setup in the pulsed laser deposition chamber, including an overview of the experimental process for using inkjet printing for confining electronic conductivity in SrTiO3-based heterostructures.In-situ sheet resistance measurements of patterned and unpatterned (b) LaAlO3/SrTiO3 and (c) γ-Al2O3/SrTiO3 heterostructures during pulsed laser deposition.For the γ-Al2O3/SrTiO3 heterostructure, results of three nominally identical unpatterned samples are shown to display the sample-to-sample variation.The insets show the contact setups used for the measurements.

Figure 4 .
Figure 4. Schematic of three processes influencing the final conductivity during the pulsed laser deposition of oxide thin films on SrTiO3: (a) bombardment of the SrTiO3 surface with plasma species to form oxygen vacancies.(b) When thermodynamically favourable, an oxygen deficient deposited film reduces the substrate and generates oxygen vacancies in SrTiO3.(c) Molecular oxygen diffuses to the interface through defects in the deposited film and annihilates the oxygen vacancies in SrTiO3.