betaFeedback

WorksPhysics and Astronomy

Leveraging high fluence and low pressure for pulsed laser deposition of high-mobility γγ-Al2_2O3_3/SrTiO3_3 heterostructure growth

Abstract

High-mobility oxide heterostructures could be applied for high-frequency devices, transparent conductors, and spin-orbit logic devices. SrTiO3_3 is one of the most studied oxide substrate materials for heterostructures. To date, the highest SrTiO3-based charge carrier mobility at 2 K was measured in the interfacial 2-dimensional electron gas (2DEG) of γγ-Al2_2O3_3/SrTiO3_3. The formation mechanism and origin of the high electron mobility are not yet fully understood. This investigation presents a successful growth protocol to synthesise high mobility γγ-Al2_2O3_3/SrTiO3_3 interfaces, and a description of the underlying growth optimisation. Furthermore, indicative features of high-mobility γγ-Al2_2O3_3/SrTiO3_3, including the room-temperature sheet resistance, are presented. Signs of epitaxial and crystalline growth are found in a high-mobility sample (μ10K=1.6×104cm2/Vsμ^{10K} = 1.6 \times 10^4 \mathrm{cm}^2/\mathrm{Vs}). Outlining the growth mechanisms and comparing 40 samples, indicates that high-fluence (F>3J/cm2F > 3\mathrm{J}/\mathrm{cm}^2) and low pressure (P≈1×10−6mbarP \approx 1 \times 10^{-6} \mathrm{mbar}) are essential growth parameters for high-mobility γγ-Al2_2O3_3/SrTiO3_3 interfaces. γγ-Al2_2O3_3 having single-element cations allows higher laser fluences during growth, compared to thin films with multi-element cations such as LaAlO3_3, without causing stoichiometric imbalances.

Introduction

For two decades, oxide-based two-dimensional electron gasses (2DEGs) have been intensively studied[1] for their diverse features, such as superconductivity,[2] ferromagnetism,[3] glassy dynamics,[4] tunable Rashba spin-orbit coupling,[5] high-mobility transport,[6, 7] extreme magnetoresistance,[8] and spin-charge interconversion.[9, 10, 11] In recent years, several steps have been taken toward implementing such oxide interfaces. This includes the generation and detection of spin-currents in ferromagnet-free oxide nanodevices[9] and fabrication of free-standing oxide interface membranes,[12, 13, 14] paving the way for cointegrating oxide and silicon electronics and spintronics in everyday devices. While most studies focus on the LaAlO3_3/SrTiO3_3 interface, the highest charge carrier mobility, μ=140 000\mu= 140\,000 cm2^2/Vs at 2 K, is recorded for γ\gamma-Al2_2O3_3/SrTiO3_3,[6] presumably due to a band-inversion that renders the dxz/yzd_{xz/yz} band dominant in electronic transport.[15, 16] This attractive property allows for mean free path dependent quantum and spintronics measurements,[17] with potential in applications including high-frequency devices,[18] transparent conductors,[18] and spin-orbit logic devices.[19] At room temperature, longitudinal optical (LO) phonons and electron-electron scattering dominate the carrier mobility in SrTiO3_3.[7, 20, 21] These two scattering phenomena dominate at different carrier density ranges. One attempt to counteract the LO phonon scattering, leveraged the non-polar, low effective electron mass, oxide BaSnO3_3, instead of SrTiO3_3, to reach room temperature mobilities on 60 cm2^2/Vs, an order of magnitude higher than SrTiO3_3-based 2DEGs.[22] The different mobilities of the two materials are attributed to the orbitals near the Fermi Energy. In SrTiO3_3, these orbitals are the dd-band orbitals of Ti, with degenerate t2gt_{2g} symmetry, while it is the isotropic ss-band orbitals in BaSnO3_3. From the less dispersive dd-band orbitals, the carriers have a higher rate of inter-orbital scattering in SrTiO3_3 compared to ss-band orbital BaSnO3_3.[22] At low temperatures, phonon-scattering vanishes, and the main mobility-reducing mechanism in SrTiO3_3 2DEGs is impurity scattering. This complicates achieving high mobility, as one of the major mechanisms forming the 2DEGs uses oxygen vacancies. Oxygen vacancies act as both dopants and scattering sites, motivating several studies into the spatial separation of oxygen vacancies and the 2DEG that arises from them.[16, 23, 24, 25] Here, we present an optimisation of the pulsed laser deposition (PLD) parameters of γ\gamma-Al2_2O3_3 grown on SrTiO3_3 towards forming a high-mobility 2DEG. First, the deposition chamber used in growth, along with four different protocols is presented. Subsequently, relations between low-temperature mobility and residual resistance ratio (RRRRRR) and room-temperature sheet resistance (RSRTR_S^{RT}), respectively, are analysed. Then the interface structure of a high-mobility sample is addressed, including epitaxial strain. We then discuss the mechanisms responsible for the high-mobility interfaces throughout the growth process, from as-received substrate SrTiO3_3 to high-mobility γ\gamma-Al2_2O3_3/SrTiO3_3 (001) 2DEG. Based on this discussion, we compare the growth protocols and highlight indications of a strong pressure-dependence of the 2DEG formation.

Experiment

Commercial SrTiO3_3(100) substrates[26] were cleaned by ultrasonication in acetone (480 s), isopropanol (IPA,480 s), and acetone:IPA 50:50 (480 s), then annealed in a 1 bar O2_2 atmosphere with a 60 min dwell time at 1000∘^\circC and an up/down ramp rate of 100 ∘^\circC/h (≈\approx 25 h in total with passive cooling). Before γ\gamma-Al2_2O3_3 deposition, the substrates underwent a second round of ultrasonication in acetone (480 s), IPA (480 s), and acetone:IPA 50:50 (480 s), respectively. The depositions, carried out by PLD, used a single-crystalline α\alpha-Al2_2O3_3 target. The ablated plasma species forms the γ\gamma-Al2_2O3_3 structure on SrTiO3_3, as displayed in Figure 1a). A 248 nm KrF excimer laser with a repetition rate of 1 Hz was used. We find that the detailed configuration of the growth reactor plays an important role in the optimisation process. The PLD chamber, illustrated in Figure 1b), is directly connected to three manually controlled valves. One of the valves, V1, is connected to a mass flow controller for oxygen injection, while two other valves, V2 and V3, connect the chamber to a turbomolecular pump. This pump is further connected to a backing pump through another manual valve (not shown). V2 and V3 controlled the specific pressure when working below 10−410^{-4} mbar. V2 is a manual gate valve controlling the range of pressure, while V3 is an angle valve, allowing for fine-tuning of the pressure. V1 was open during O2_2 flushing and automatised proportional–integral–derivative (PID) pressure control. The thermocouple, measuring the sample temperature, resides in the heater centre, and the substrate-target distance was 4.5 ±\pm 0.1 cm.

Schematic of the $\gamma$-Al$_2$O$_3$/SrTiO$_3$ crystal structure and pressure-control protocols

Figure 1. Pressure-control for increased reproducibility and crystal structure. a) Schematic of the γ\gamma-Al2_2O3_3/SrTiO3_3 crystal structure grown with the PLD setup illustrated in b). c-f) Examples of temperature (brown) and pressure (purple) profiles on Protocols 1-4, respectively.

Pressure and temperature data of four protocols are displayed in Figure 1c-Figure 1f) and the different growth parameters are listed in Table I. Protocol 1 was the first attempted protocol that yielded high-mobility samples, the other three protocols were attempts to automate and further optimise the growth process concerning reproducibility and mobility.

Protocol 1, Figure 1c), started with flushing the chamber with Pflush=7×10−2P_{\mathrm{flush}} = 7 \times10^{-2} mbar O2_2 to ensure that the dominant gas in the chamber was O2_2. Then the chamber was evacuated to Ppre=1P_{\mathrm{pre}} = 1-2×10−62 \times10^{-6} mbar and the target preablated for tpredep=10t_{\mathrm{pre}}^{\mathrm{dep}} = 10 min with a shutter shielding the SrTiO3_3 substrate. Using Pheat=Pcool=PpreP_{\mathrm{heat}} = P_{\mathrm{cool}} = P_{\mathrm{pre}}, the heater, onto which the substrate was thermally and electrically attached using conducting silver paint, was heated to the deposition temperature of Tdep=650T_{\mathrm{dep}} = 650 ∘^\circC with a ramp rate of Γramp=15\Gamma_{\mathrm{ramp}} = 15 ∘^\circC/min. The sample was pre-annealed for tpre=11t_{\mathrm{pre}} = 11 min, except where else is stated, at deposition pressure and temperature before 4 min of ablation began. After ablation, the sample was post-annealed for tpost=5t_{\mathrm{post}} = 5 min at deposition pressure and temperature before the onset of the cooling. The cooling was, like the heating, set to a ramp rate of Γramp=15\Gamma_{\mathrm{ramp}} = 15 ∘^\circC/min, below 300 ∘^\circC the ramp rate decreased since the sole cooling power came from the local, low-pressure, environment. All samples grown with protocol 1 were attached to the heater through a metal plate. For the subsequent protocols, the omicron plate was removed to obtain a more precise representation of the sample temperature. Supplementary Material Sec. IV describes the growth of a sample using Protocol 1 without the metallic plate. This resulted in a bulk conducting sample, due to the additional thermal energy during heating, growth and cooldown.

ParametersAbbreviationProtocol 1Protocol 2Protocol 3Protocol 4
Flushing Pressure (mbar)PflushP_{\mathrm{flush}}7×10−27 \times10^{-2}NA7×10−27 \times10^{-2}7×10−27 \times10^{-2}
Deposition pressure (mbar)PdepP_{\mathrm{dep}}11-2×10−62 \times10^{-6}1×10−61 \times10^{-6}1×10−61 \times10^{-6}1×10−61 \times10^{-6}
Preablation Pressure (mbar)PpreP_{\mathrm{pre}}PdepP_{\mathrm{dep}}PdepP_{\mathrm{dep}}PdepP_{\mathrm{dep}}PdepP_{\mathrm{dep}}
Heating pressure (mbar)PheatP_{\mathrm{heat}}1×10−61 \times10^{-6}1×10−41 \times10^{-4}Base PPBase PP
Cooling pressure (mbar)PcoolP_{\mathrm{cool}}1×10−61 \times10^{-6}1×10−41 \times10^{-4}Base PP1×10−41 \times10^{-4}
Ramp rate (°C/min)Γramp\Gamma_{\mathrm{ramp}}151510,1510,15
Deposition temperature (°C)TdepT_{\mathrm{dep}}650550, 650550550, 660, 650
Preablation time (min)tpret_{\mathrm{pre}}10101010
Predeposition annealing time (min)tpreannt_{\mathrm{pre}}^{\mathrm{ann}}11,16,516-511111
Postdeposition annealing time (min)tpostt_{\mathrm{post}}5555
Laser fluence (J/cm²)FF3.1-9.63.7-9.27.1-9.36.4-8.0

Table 1. Overview of growth parameters.

Protocol 2, as displayed in Figure 1d), differed from Protocol 1 mainly by Pheat=Pcool=1×10−4P_{\mathrm{heat}} = P_{\mathrm{cool}} = 1 \times10^{-4} mbar O2_2. For this protocol, different tpreannt_{\mathrm{pre}}^{\mathrm{ann}} durations were tested and initial flushing with O2_2 was excluded.

In Protocol 3, shown in Figure 1e), Tdep=550T_{\mathrm{dep}} = 550 °C and V2 and V3 left completely open without any O2_2 in-flow during the whole process except for the initial O2_2 flush. This decreases PheatP_{\mathrm{heat}} and PcoolP_{\mathrm{cool}} to base pressure, which depend on the heating and pumping.

Protocol 4 was identical to Protocol 3 with the exception that Pcool=10−4P_{\mathrm{cool}} = 10^{-4} mbar, as displayed in Figure 1f). The pressure peak initializing the cooldown corresponds to the PID fluctuating to find the correct value. It should be noted that, for all depositions, the pressure increases during ablation, since the ablated species from the target affect the pressure.

The γ\gamma-Al2_2O3_3 growth thicknesses were measured by X-ray reflectivity (XRR) using a film grown by Protocol 1 (dγ-Al2O3≈8.5d_{\gamma\text{-Al}_2\text{O}_3} \approx8.5 nm). The growth time (4 min) was the same for all samples. X-ray diffraction (XRD), XRR, and atomic force microscopy (AFM), were used to assess the crystal structure.

Electrical connections were made by bonding with Al ultrasonic wire-bonding, in van der Pauw geometry. Electronic transport below room temperature was measured using a Cryogen-Free Measurement System equipped with a 16 T magnet.

Indicative room temperature parameters

For the samples of this study, explicit low-temperature mobility measurements need cooling and testing in vdP and Hall configurations with magnetic field sweeps. Indicative room temperature parameters allow quicker mobility optimisation, as reported previously[7]. As displayed in Figure 2a) such indicative features are extractable from the sheet resistance temperature-dependency. Both high (4.2 × 104^{4} cm2^{2}/Vs at 2 K) and low (6 × 103^{3} cm2^{2}/Vs at 2 K) mobility samples here show metallic behaviour but differ in the absolute resistance and residual resistance ratio (RRR(10 K)=Rs(300 K)/Rs(10 K)RRR(10\ \mathrm{K}) = R_s(300\ \mathrm{K})/R_s(10\ \mathrm{K})). The mobility as a function of carrier density is shown in Figure 2b). A peak in mobility was reported[7] at ns≈4×1014 cm−2n_s \approx4 \times10^{14}\ \mathrm{cm}^{-2}, within a range of carrier densities not accessed in this investigation. Plotting the mobility as a function of room temperature sheet resistance (RsRTR_s^{RT}), displayed in Figure 2c), reveals a qualitative mirror image of Figure 2b). The inset indicates the log-log linear dependency between the carrier density and RsRTR_s^{RT}, explaining these similarities. The dependency between mobility and RsRTR_s^{RT}, through the carrier density, reveals RsRTR_s^{RT} as a relevant indicative value to estimate the low-temperature mobility.

Indicative values for high mobility prediction

Figure 2. Indicative values for high mobility prediction. a) Sheet resistance, RsR_{\mathrm{s}}, as a function of temperature (TT) for a high mobility (brown) and a low mobility (purple) γ\gamma-Al2_2O3_3/SrTiO3_3 sample. The electron mobility at 10 K (μ10 K\mu^{10\,K}) as a function of carrier concentration at 10 K (ns10 Kn_{\mathrm{s}}^{10\,K}) in b), as a function of sheet resistance at room temperature (RsRTR_{\mathrm{s}}^{RT}) in c), and as a function of residual resistance ratio (RRR(10 K)RRR(10\,K)) in d). Inset in c) displays the ns10 Kn_{\mathrm{s}}^{10\,K}-dependency of RsRTR_{\mathrm{s}}^{RT}. The brown points are data and the purple dashed line a log-log linear fit. Carrier densities and mobilities are extracted through a single electronic band model as displayed in Supplementary Material Fig. S1.

One can define a relevant upper threshold for high mobility in RsRTR_s^{RT} at 10 kΩ\Omega/sq. A lower threshold is not determinable from Figure 2, although some level of conductance between the top and bottom of a sample can provide an alternative, as 3D conductance is undesirable for 2D features. A threshold 3D conductance requires additional investigations.

The total carrier density is often temperature independent, hence RRRRRR can be understood as an indirect measurement of the thermal contribution to μ\mu, which will be high for high μ(10 K)\mu(10\ \mathrm{K}). Figure 2d) exhibits the mobility at 10 K as a function of RRR(10 K)RRR(10\ \mathrm{K}). RRR(10 K)RRR(10\ \mathrm{K}) increases with increasing mobility, yielding a range for high-mobility (μ>104\mu> 10^4) samples at RRR(10 K)>3×103RRR(10\ \mathrm{K}) > 3 \times10^3. This can be observed because both RsRTR_s^{RT} and μ10K\mu^{10\mathrm{K}} are proportional to the approximately temperature-independent nsn_s.

Structural Analysis

To gain insights into the relations between the crystalline quality and the mobility, sample 9, grown with protocol 1, (μ10K=1.6×104 cm2/Vs\mu^{10\mathrm{K}} = 1.6 \times10^4\ \mathrm{cm}^2/\mathrm{Vs}) was measured using AFM, XRR, and XRD. The visible terraces in AFM after deposition, as displayed in Figure 3a), indicate crystalline and epitaxial growth of the γ\gamma-Al2_2O3_3 thin film. It should be noted that the surface termination of the substrates before deposition did not show a clear single-termination. However, the high-mobility carriers in the dxz/yzd_{xz/yz}-band should reside deeper in the SrTiO3_3 substrate than those in the dxyd_{xy}-band,[16] why interface-related scattering from e.g. surface roughness should contribute less in dxz/yzd_{xz/yz}-dominated γ\gamma-Al2_2O3_3/SrTiO3_3 compared to similar dxyd_{xy}-dominated heterostructures. Since reflection high-energy electron diffraction was not installed in this chamber, the thin film thicknesses and epitaxial crystalline growth were confirmed by XRD and XRR measurements. In Figure 3b), the XRR data is displayed along with a fit determining the γ\gamma-Al2_2O3_3 thickness to dγ-Al2O3=8.44±0.04d_{\gamma\text{-Al}_2\text{O}_3} = 8.44 \pm0.04 nm. Figure 3c) shows XRD 2θ/ω2\theta/\omega scans of the γ\gamma-Al2_2O3_3 sample 9 grown on SrTiO3_3 (001). γ\gamma-Al2_2O3_3 grows with a positive out-of-plane strain of 0.721(1)%. Calculated by using the c-lattice parameter of bulk γ\gamma-Al2_2O3_3 (7.911 Å).[27]

Structural data of $\gamma$-$\mathrm{Al}_2\mathrm{O}_3$/SrTiO$_3$ Sample 9

Figure 3. Structural data of γ\gamma-Al2O3\mathrm{Al}_2\mathrm{O}_3/SrTiO3_3 Sample 9. a) AFM after γ\gamma-Al2O3\mathrm{Al}_2\mathrm{O}_3 deposition on SrTiO3_3. b) X-ray reflectivity data and fit (solid brown line) yielding a γ\gamma-Al2O3\mathrm{Al}_2\mathrm{O}_3 thickness of dγ-Al2O3=8.44±0.4d_{\gamma\text{-}\mathrm{Al}_2\mathrm{O}_3} = 8.44 \pm0.4 nm. c) X-ray diffraction showing the SrTiO3_3 (002) and γ\gamma-Al2O3\mathrm{Al}_2\mathrm{O}_3 (004) peaks. Sample 9 was grown using protocol 1. The y-axes in b) and c) is logarithmic intensity.

2DEG Formation Mechanisms

A thorough understanding of the 2DEG formation mechanism is essential to optimise the mobility in γ\gamma-Al2_2O3_3/SrTiO3_3 further. As previously described in ref.[28], annealing in oxygen causes a transition from Rs(RT)=310 Ω/sqR_s(\mathrm{RT}) = 310\ \Omega/\mathrm{sq} to Rs(RT)>950 MΩ/sqR_s(\mathrm{RT}) > 950\ \mathrm{M}\Omega/\mathrm{sq}. This points towards oxygen vacancies being the primary source of the 2DEG formation.

Figure 4a) illustrates how the samples go through six regimes in the growth process. Regime I involves annealing at atmospheric pressure and high temperatures. The sample is then loaded into the PLD chamber, initiating Regime II during heating under low pressure. Regime III occurs as the SrTiO3_3 surface remains exposed to low pressure at the growth temperature. Deposition begins in Regime IV and ends before Regime V, which includes cooldown. Regime VI covers subsequent sample storage. Regime I is illustrated in Figure 4b), where the arrow indicates the direction and magnitude of the oxygen atom diffusion to restore thermodynamic equilibrium between the SrTiO3_3 lattice and the environment. Since the temperature reaches 1000 °C in O2_2 environment, the oxygen atoms become highly mobile and vacancies are filled. This creates a reproducible starting point regarding oxygen distribution. During the subsequent cooling, the atomic mobility decreases, making the oxygen occupation in the sample reflect the environment, functioning as a pressure memory of the furnace equilibrium.

Oxygen dynamics through the growth protocols

Figure 4. Oxygen dynamics through the growth protocols. a) Illustration of the full timespan of the growth process separated into six different regimes. b) SrTiO3_3 surrounded by oxygen molecules, as is the case during initial annealing. c) SrTiO3_3 in a low-pressure oxygen environment as the pulsed laser deposition chamber. d) same as c) but with an oxygen vacancy in the topmost layer. The arrows in b-d) indicates the direction and magnitude of the oxygen atoms diffusion to restore equilibrium, not drawn to scale. e) Bombardment of SrTiO3_3 in the initial part of the pulsed laser deposition. Here v⃗\vec{v} is the velocity vector of the plasma species. f) Oxygen scavenging by the Al atoms of γ\gamma-Al2_2O3_3 that have higher oxygen affinity than Ti and Sr in SrTiO3_3. g) Oxygen recombination from the atmosphere. The arrows in f) and g) do not indicate forces but movements.

In regimes II and III, illustrated in Figure 4c), the substrate is loaded in the PLD chamber, where the pressure is adjusted to ≤10−4\leq10^{-4} mbar. Now, the pressure balance is reversed. With the low-pressure environment and the memory of the furnace pressure in the lattice, the oxygen atoms will diffuse out of the lattice to restore thermodynamic equilibrium. The topmost atomic layer will experience the fastest reduction since this layer is directly exposed to the environment. Once a vacancy is generated in the topmost layer, the system has moved slightly towards equilibrium. This decreases the oxygen diffusion, as illustrated in Figure 4d) and allows the oxygen vacancies to move further into the bulk SrTiO3_3. When the substrate reaches deposition temperature, a laser pulse is sent into the target, entering regime IV. The laser pulse is ablated into a high-energy plasma. When the plasma reaches the substrate a bombardment effect occurs, as illustrated in Figure 4e). The plasma adds energy on the scale of tens of eV[29, 30] to the top layers of the lattice, enabling the escape of light elements. In SrTiO3_3, the light element is oxygen. Because γ\gamma-Al2_2O3_3 only possess a single-element cation, one can increase the fluence without considering the effect on the cationic stoichiometry. Hence, one can utilise the fluence to access a larger range of energetic minima associated with crystal growth of γ\gamma-Al2O3\mathrm{Al}_2\mathrm{O}_3 compared to e.g. LaAlO3_3 or LaTiO3_3.

As the γ\gamma-Al2O3\mathrm{Al}_2\mathrm{O}_3 thin film is grown in the oxygen-deficient environment, it further reduces SrTiO3_3 since the oxygen affinity of Al is higher than that of Ti and Sr. The combination of the high temperature and the kinetic energy from the plasma ensures high atomic mobility.[30, 31, 32] This corresponds to regime IV and V and is illustrated in Figure 4f).

At last, during sample cooldown (regime V), for Protocol 2 and 4, the pressure is increased from deposition pressure to 10−410^{-4} mbar. This may cause oxygen recombination, as displayed in Figure 4g), where an atmospheric oxygen molecule is split up and added to the lattice, annihilating two vacancies.[30, 32, 33] This effect may also occur during subsequent storage corresponding to regime VI.

Protocol Comparison

Samples 1, 2, 3, 9, and 10 grown by Protocol 1, as already displayed in Figure 2, were confirmed to host high-mobility 2DEGs. Protocol 1, however, has some weaknesses. Stabilising a constant low pressure during the temperature profile is challenging, often failing to reach the intended low pressure during heating, probably due to outgassing from the silver paint and heater. This is likely the reason for the large fluctuation in RsRTR_s^{RT}, as shown in Supplementary Material Table S1. Additionally, manual pressure control is time-consuming (>5>5 hours) and less reproducible than automated control. To increase the reproducibility and time efficiency of the growth, Protocol 2, with Pheat=Pcool=1×10−4P_{\mathrm{heat}} = P_{\mathrm{cool}} = 1 \times10^{-4} mbar, was attempted to create samples 19-33. In contrast to Protocol 1, however, Protocol 2 shows measurable conductance R2G(RT)R^{2G}(\mathrm{RT}) at room temperature between the chip carrier back plane and any of the four corners on

the upper surface of the sample. Thus, Protocol 2 yields bulk conductance before high mobility 2DEGs, when decreasing the sheet resistance by extending tpreannt_{\mathrm{pre}}^{\mathrm{ann}}.

The tendency to create bulk conductance before high-mobility 2DEGs, could be explained by the balance between the effects of temperature and pressure. When the substrate is uncapped by the thin film, in regimes II and III, the pressure primarily affects the surface of the sample, where the pressure gradient is highest. In contrast, the temperature acts primarily on the back side of the substrate, where it is thermally anchored and then decreasingly through the bulk and up to the front surface. The pressure gradient at the surface effectively pulls the oxygen atoms out of the lattice, while the temperature facilitates energy to allow atomic movement. Thus, both effects are needed, but the low pressure drives the surface oxygen vacancy formation, while the temperature drives a homogenous distribution of oxygen vacancies. Therefore, a dominating low-pressure before deposition is preferential for high-mobility 2DEGs.

Attempting to leverage this mechanism for the formation of surface oxygen vacancies, Protocol 3 was applied for growing samples 34-36, as shown in Table S1. This protocol also yields 3D conductance before high mobility 2DEG formation. In this case, however, the 3D conductance likely develops during post-annealing and cooldown when the pressure is at its lowest, the temperature is still significant, and the potential 2DEG area is shielded by the γ\gamma-Al2O3\mathrm{Al}_2\mathrm{O}_3, acting as a capping layer that keeps the oxygen from escaping from the 2DEG area.

To counteract both the lack of surface oxygen vacancies from Protocol 2 and the increase in bulk oxygen vacancies in Protocol 3, Protocol 4 was developed as displayed by samples 37-40. Here, the bulk conductance does not reach measurable ranges, but the 2DEG resistance tends to increase with increasing growth temperature, rendering the protocol unsuited for high-mobility 2DEG growth. This can be attributed to oxygen recombination while the pressure is 10−410^{-4} mbar and during the peak in pressure while the pressure PID fluctuates towards the correct value.

carrier mobility. This purpose was initially reached by Protocol 1, displaying a growth window of such epitaxial, crystalline, high mobility γ\gamma-Al2_2O3_3/SrTiO3_3 2DEGs. Protocol 1 however, is highly dependent on manual work, decreasing the reproducibility. Hence, the subsequent protocols attempted to stay within the high mobility growth window while automating the work. To optimise that process, the residual resistance ratio (RRRRRR) and the room-temperature sheet resistance (RsR_{\mathrm{s}}) were identified as indicative parameters for estimating the mobility at 10 K without time-consuming low-temperature measure-

Conclusion

Within this investigation, four different growth protocols attempting to grow γ\gamma-Al2_2O3_3 on SrTiO3_3 are presented alongside small variations within each of them. The purpose was to create a 2DEG with high carrier mobility. This purpose was initially reached by Protocol 1, displaying a growth window of such epitaxial, crystalline, high mobility γ\gamma-Al2_2O3_3/SrTiO3_3 2DEGs. Protocol 1 however, is highly dependent on manual work, decreasing the reproducibility. Hence, the subsequent protocols attempted to stay within the high mobility growth window while automating the work. To optimise that process, the residual resistance ratio (RRRRRR) and the room-temperature sheet resistance (RsR_{\mathrm{s}}) were identified as indicative parameters for estimating the mobility at 10 K without time-consuming low-temperature measurements between every growth. For both RRR\mathrm{RRR} and RsR_s, the reason for the indicative implications is identified as the approximate temperature independence of the carrier concentration of the γ\gamma-Al2O3\mathrm{Al}_2\mathrm{O}_3/SrTiO3\mathrm{SrTiO}_3 2DEGs. Four different mechanisms for oxygen-vacancy-driven 2DEG formation were discussed. These were SrTiO3\mathrm{SrTiO}_3-environment equilibrium, bombardment, reduction of SrTiO3\mathrm{SrTiO}_3 by γ\gamma-Al2O3\mathrm{Al}_2\mathrm{O}_3, and oxygen recombination. γ\gamma-Al2O3\mathrm{Al}_2\mathrm{O}_3 having a single-element cation allows high laser fluences during growth without affecting stoichiometry, providing an increased bombardment effect and available energy for the crystal growth. Considering the growth mechanisms and comparing 40 samples grown with the four different protocols, pointed toward a need for low pressure to induce oxygen vacancies in the surface layers of SrTiO3\mathrm{SrTiO}_3 before deposition.

Supplementary Material

The supplementary material includes an overview of the 40 samples described in this investigation, linear fitting of the Hall resistance used to extract the carrier densities and mobilities in Figure 3. Furthermore, it includes a detailed electronic transport analysis of one high-mobility sample (sample 1), including a 2-band and anomalous hall effect fitting.

Acknowledgements

We thank R. T. Dahm, C. N. Lobato, V. Rosendal, and C. E. N. Petersen for helpful discussions. T.H.O., A. C. and F.T. acknowledge support by research grant 37338 (SANSIT) from Villum Fonden. T.S.J. acknowledges support from the Novo Nordic Foundation Challenge Program, grant no. NNF21OC0066526 (BioMag).

References

  1. [1]A. Ohtomo and H. Y. Hwang, “A high-mobility electron gas at the laao3/srtio3 heterointerface,” Nature 427, 423–426 (2004).DOI
  2. [2]N. Reyren, S. Thiel, A. D. Caviglia, L. F. Kourkoutis, G. Hammel, C. Richter, C. W. Schneider, T. Kopp, A.-S. Rüetschi, D. Jaccard, M. Gabay, D. A. Muller, J.-M. Triscone, and J. Mannhart, “Superconducting interfaces between insulating oxides,” Science 317, 1196–1199 (2007).pubmed.ncbi.nlm.nih.gov/17673621
  3. [3]A. Brinkman, M. Huijben, M. van Zalk, J. Huijben, U. Zeitler, J. C. Maan, W. G. van der Wiel, G. Rijnders, D. H. A. Blank, and H. Hilgenkamp, “Magnetic effects at the interface between non-magnetic oxides,” Nature Materials 6, 493–496 (2007).pubmed.ncbi.nlm.nih.gov/17546035
  4. [4]J. Maity, S. K. Ojha, P. Mandal, M. Beniwal, N. Bhattacharya, A. Gloskowskii, C. Schlue ter, and S. Mid dey, “Glassy dynamics of srtio3-based two-dimensional electron gas,” Communications Materials 6, 84 (2025).DOI
  5. [5]A. D. Caviglia, M. Gabay, S. Gariglio, N. Reyren, C. Cancel lieri, and J.-M. Triscone, “Tunable rashba spin-orbit interaction at oxide interfaces,” Physical Review Letters 104 (2010), 10.1103/physrevlett.104.126803.pubmed.ncbi.nlm.nih.gov/20366557
  6. [6]Y. Z. Chen, N. Bovet, F. Trier, D. V. Christensen, F. M. Qu, N. H. Andersen, T. Kasama, W. Zhang, R. Giraud, J. Dufouleur, T. S. Jespersen, J. R. Sun, A. Smith, J. Nygård, L. Lu, B. Büchner, B. G. Shen, S. Linderoth, and N. Pryds, “A high-mobility two-dimensional electron gas at the spinel/perovskite interface of γ-al2o3/srtio3,” Nature Communications 4, 1371 (2013).
  7. [7]D. V. Christensen, Y. . Frenkel, P. Schutz, F. Trier, S. Wissberg, R. Claessen, B. Kalisky, A. Smith, Y. Z. Chen, and N. Pryds, “Electron mobility in γ-al2o3/srtio3,” Physical Review Applied 9 (2018), 10.1103/PhysRevApplied.9.054004.DOI
  8. [8]D. V. Christensen, T. S. Steegemann, T. D. Pomar, Y. Z. Chen, A. Smith, W. N. Strocov, B. Kalisky, and N. Pryds, “Extreme magnetoresistance at high-mobility oxide heterointerfaces with dynamic defect tunability,” Nature Communications 15, 4249 (2024).DOI
  9. [9]F. Trier, D. C. Vaz, P. Bruneel, P. Noël, A. Fert, L. Vila, J.-P. Attané, A. Barthélémy, M. Gabay, H. Jaffrès, and M. Bibes, “Electric-field control of spin current generation and detection in ferromagnet-free srtio3-based nanodevices,” Nano Letters 20, 395–401 (2020), doi: 10.1021/acs.nanolett.9b04079.
  10. [10]P. Noël, F. Trier, L. M. V. Arche, J. Bréhin, D. C. Vaz, V. Garcia, S. Fusil, A. Barthélémy, L. Vila, M. Bibes, and J.-P. Attané, “Non-volatile electric control of spin–charge conversion in a srtio3 rashba system,” Nature 580, 483–486 (2020).
  11. [11]D. C. Vaz, F. Trier, A. Dyrdal, A. Johansson, K. Garcia, A. Barthélémy, I. Mertig, J. Barnaś, A. Fert, and M. Bibes, “Determining the rashba parameter from the bilinear magnetoresistance response in a two-dimensional electron gas,” Physical Review Materials 4, 71001 (2020).
  12. [12]A. Sambri, M. Scuderi, A. Guarino, E. D. Gennaro, R. Erlandsen, R. T. Dahm, A. V. Bjørlig, D. V. Christensen, R. D. Capua, B. D. Ventura, U. S. di Uccio, S. Mirabella, G. Nicotra, C. Spinella, T. S. Jespersen, and F. M. Granozio, “Self-formed, conducting laalo3/srtio3 micro-membranes,” Advanced Functional Materials 30, 19099964 (2020), https://doi.org/10.1002/adfm.201909964.
  13. [13]R. T. Dahm, R. Erlandsen, F. Trier, A. Sambri, E. D. Gennaro, A. Guarino, L. Stampfer, D. V. Christensen, F. M. Granozio, and T. S. Jespersen, “Size-controlled spalling of laalo3/srtio3 micro-membranes,” ACS Applied Materials & Interfaces 13, 12341–12346 (2021), doi: 10.1021/acsami.0c16212.DOI
  14. [14]Y. Li, C. Xiang, F. M. Chiabrera, S. Yun, H. Zhang, D. J. Kelly, R. T. Dahm, C. K. R. Kirchert, T. E. L. Cozannet, F. Trier, D. V. Christensen, T. J. Booth, S. B. Simonsen, S. Kadkhodazadeh, T. S. Jespersen, and N. Pryds, “Stacking and twisting of freestanding complex oxide thin films,” Advanced Materials 34, 2203187 (2022), https://doi.org/10.1002/adma.202203187.
  15. [15]Y. Cao, X. Liu, P. Shafer, S. Middey, D. Meyers, M. Kareev, Z. Zhong, J.-W. Kim, P. J. Ryan, F. A renholz, and J. Chakhalian, “Anomalous orbital structure in a spinel–perovskite interface,” npj Quantum Materials 1, 16009 (2016).DOI
  16. [16]A. Chikina, D. V. Christensen, V. Borisov, M.-A. Husanu, Y. Chen, X. Wang, T. Schmitt, M. Radovic, N. Nagaosa, A. S. Mishchenko, R. Valenti, N. Pryds, and V. N. Strocov, “Band-order anomaly at the γ-al2o3/srtio3 interface drives the electron-mobility boost,” ACS Nano 15, 4347–4356 (2021), doi: 10.1021/acsnano.0c07609.DOI
  17. [17]M. Oltscher, M. Ciorga, M. Utz, D. Schuh, D. Bougeard, and D. Weiss, “Electrical spin injection into high mobility 2d systems,” Phys. Rev. Lett. 113, 236602 (2014).
  18. [18]F. Trier, D. V. Christensen, and N. Pryds, “Electron mobility in oxide heterostructures,” Journal of Physics D: Applied Physics 51, 293002 (2018).DOI
  19. [19]S. Manipatruni, D. E. Nikonov, C.-C. Lin, T. A. Gosavi, H. Liu, B. Prasad, Y.-L. Huang, E. Bonturim, R. Ramesh, and I. A. Young, “Scalable energy-efficient magnetoelectric spin–orbit logic,” Nature 565, 35–42 (2019).DOI
  20. [20]B. Himmetoglu, A. Janotti, H. Peelaers, A. Alkauskas, and C. G. V. de Walle, “First-principles study of the mobility of srtio3,” Physical Review B 90, 241204 (2014).DOI
  21. [21]B. Mikheev, B. Himmetoglu, A. P. Kajdos, P. Moetakef, T. A. Cain, C. G. V. de Walle, and S. Stemmer, “Limitations to the room temperature mobility of two- and three-dimensional electron liquids in srtio3,” Applied Physics Letters 106, 062102 (2015).
  22. [22]K. Eom, H. Paik, J. Seo, N. Campbell, E. Y. Tsymbal, S. H. Oh, M. S. Rzchowski, D. G. Schlom, and C. Eom, “Oxide two-dimensional electron gas with high mobility at room-temperature,” Advanced Science, 2105652 (2022).DOI
  23. [23]A. F. Zurhelle, D. V. Christensen, S. Menzel, and F. Gunkel, “Dynamics of the spatial separation of electrons and mobile oxygen vacancies in oxide heterostructures,” Physical Review Materials 4, 104604 (2020).
  24. [24]Y. Z. Chen, F. Trier, T. Wijnands, R. J. Green, N. Gauquelin, R. E. Gagoil, D. V. Christensen, G. Koster, M. Huijben, N. Bovet, S. Macke, F. He, R. Sutarto, N. H. Andersen, J. A. Sulzpio, M. Honig, G. E. D. K. Prairatoomdojo, T. S. Jespersen, S. Linderoth, S. Ilani, J. Verbeeck, G. Van Tendeloo, G. Rijnders, G. A. Sawatzky, and N. Pryds, “Extreme mobility enhancement of two-dimensional electron gases at oxide interfaces by charge-transfer-induced modulation doping,” Nature Materials 14, 801–806 (2015).DOI
  25. [25]M. Huijben, G. Koster, M. K. Kruize, S. Wenderich, J. Verbeeck, S. Bals, E. Slooten, B. Shi, H. J. A. Molegraaf, J. E. Kleibeuker, S. van Aert, J. B. Goedkoop, A. Brinkman, D. H. A. Blank, M. S. Golden, G. van Tendeloo, H. Hilgenkamp, and G. Rijnders, “Defect engineering in oxide heterostructures by enhanced oxygen surface exchange,” Advanced Functional Materials 23, 5240–5248 (2013), https://doi.org/10.1002/adfm.201203355.
  26. [26]LTD. SHINKOSHA CO., “Strontium titanate,”.
  27. [27]Z. Rongsheng and R. Snyder, “Structures and transformation mechanisms of the η, γ and θ transition aluminas,” Acta Crystallographica Section B: Structural Science 47 (1991).
  28. [28]T. Hvid-Olsen, C. H. Christoffersen, D. J. Carrad, N. Gauquelin, D. Olsteins, J. Verbeeck, N. Bergeal, T. S. Jespersen, and F. Trier, “Coexistence of high electron-mobility, undoped spins, and superconductivity at high carrier density srtio3-based interfaces,” (2024), arXiv:2411.03824 [cond-mat.mtrl-sci].
  29. [29]A. Sambri, D. V. Caruso, F. Trier, Y. Z. Chen, S. Amoruso, N. Pryds, R. Bruzzese, and X. Wang, “Plasma plume effects on the conductivity of amorphous-lao3/srtio3 interfaces grown by pulsed laser deposition in o2,” Applied Physics Letters 100, 231605 (2012), doi: 10.1063/1.4727905.DOI
  30. [30]M. von Soosten, D. V. Christensen, C.-E. Eom, T. S. Jespersen, Y. Chen, and N. Pryds, “On the emergence of conductivity at srtio3-based oxide interfaces – an in-situ study,” Scientific Reports 9, 18005 (2019).
  31. [31]Y. Chen, N. Pryds, J. E. Kleibeuker, G. Koster, J. Sun, E. Stamate, B. Shen, G. Rijnders, and S. Linderoth, “Metallic and insulating interfaces of amorphous srtio3-based oxide heterostructures,” Nano Letters 11, 3774–3778 (2011), doi: 10.1021/nl201821j.DOI
  32. [32]D. V. Christensen, M. von Soosten, F. Trier, T. S. Jespersen, A. Smith, Y. Chen, and N. Pryds, “Controlling the carrier density of srtio3-based heterostructures with annealing,” Advanced Electronic Materials 3, 1700026 (2017), https://doi.org/10.1002/aelm.201700026.
  33. [33]T. Hvid-Olsen, C. Gadea, F. B. Holde, K. M. Hoffmann, T. S. Jespersen, K. Grove-Rasmussen, F. Trier, and D. V. Christensen, “Spatial control of the conductivity in strio3-based heterointerfaces using inkjet printing,” Journal of Physics: Energy (2022), 10.1088/2515-7655/ac9084.

Paper details

vlnc:W_tq35xn7n
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents