Nicholas Hagopian, Vito Iaia, Emily Joseph, Stephanie M. Ribet, Sean R. O’Kelley, Sayan Patra, Loren D. Alegria, Keith Ray, Yaniv Jacob Rosen, Paul M. Voyles
Superconducting qubits (SQ) are favorable technology for quantum computing due to their macroscopic nature, coherence times, and gate speeds [1]. Materials synthesis and processing methods for SQs are common to fabrication in the semiconductor industry, promising easy scalability. Josephson junctions (JJ) are critical components of SQs and consist of a thin dielectric layered between two superconducting electrodes [2]. The prototypical JJ is an Al/AlOx/Al sandwich structure, as aluminum forms a thin, self-limiting oxide [2, 3]. Amorphous alumina (a-AlOx) has also seen widespread adoption as a dielectric in semiconducting devices due to favorable properties such as high breakdown fields and low leakage currents [4]. Aluminum oxidation follows Cabrera-Mott (CM) mechanisms, where adatom-induced fields drive Al-ion diffusion towards the surface [3]. Pappas et al. introduced alternating bias assisted annealing (ABAA), an electrical processing technique to tune JJ resistance [5]. ABAA applies alternating voltage pulses, below the JJ breakdown field but above the CM field, hypothesized to facilitate ionic diffusion and anneal the oxide. Via scanning transmission electron microscopy (STEM), they demonstrated that ABAA induces a measurable change on the Al electron energy loss near-edge structure (EELS-ELNES), suggestive of potential changes in bonding and homogenization of the oxide [5]. Junction resistance with ABAA can be tuned to a narrow distribution, improving device performance [6]. Iaia et al. studied the influence of ABAA in thicker a-AlOx layers in superconducting resonators designed to study two level system defects (TLS) in the amorphous dielectric [7]. They note pronounced impact of ABAA applied and measured at mK temperatures (cryogenic alternating bias stimulation, CABS) on TLS behavior, which are reversible when the device is annealed to 10 K. In this work, we investigate the influence of ABAA on a-AlOx film structure using 4D STEM fluctuation electron microscopy (FEM). Synthesis and processing of the device is described in Iaia et al. [7]. Briefly, a ∼50 nm thick a-AlOx layer was synthesized via thermal atomic layer deposition (ALD), which differs from conventional a-AlOx growth in JJs. ABAA processing was applied with ±100 MV/m fields at 353 K over 30 hours. STEM sample thickness was controlled between 0.4 – 0.7 λEMFPAlOx. 4D STEM experiments were performed at 300 kV on the Lawrence Berkeley National Laboratory TEAM I, collected with a Dectris Arina detector, with a 1.2 mrad and 2 pA probe to minimize beam damage [8]. Thickness was calculated for each real space position based on the transmitted intensity of the diffraction zero-beam using a vacuum dataset for reference. The variance was normalized for thickness at each pixel and averaged across each row of diffraction patterns. Several datasets were collected, aligned, normalized, and averaged to improve statistics. Variance profiles are compared as a function of depth to understand a-AlOx medium range order across the device thickness. Figure 1 shows FEM analysis of a resonator. Fluctuation electron microscopy (FEM) variance was calculated using the Vr method for each pixel [9]. Variance is a metric which tracks how speckle intensity as a function of k fluctuates across a dataset [10]. Speckles originate from individual ordered atomic clusters in the amorphous material aligned on Bragg conditions [10]. A low variance is suggestive of a lower speckle-contrast amorphous halo with little fluctuation across the dataset, or a more uniform and amorphous structure. In contrast, a higher variance suggests stronger speckle contrast intensity fluctuations, with sources arising from both discrete speckles and changes across the dataset, or a more heterogeneous and ordered amorphous material. Figure 1 compares FEM normalized variance profiles between the as-made (AM) and ABAA devices, and Figure 2 isolates the peak variance values at ∼3 and ∼7 nm-1 as a function of distance from the TiN electrode. In the AM device, a pronounced decrease in the variance is observed near the TiN interface, suggesting a substantial reduction in amorphous structural order. This variance reduction extends several nm into the film. We speculate this may be attributable to TiN interfacial effects during the initial stages of the ALD growth. Datasets collected with different cross-sections, probe currents, shorter dwell times, and different scan geometries all show comparable TiN interfacial features, so we believe this effect is not attributable to electron or ion beam damage [8]. The ABAA-treated device shows only a modest variance reduction near the TiN interface. This suggests that ABAA increases the a-AlOx ordering near the interface. While the a-AlOx structure as a function of depth is more homogeneous after ABAA, the a-AlOx structure near the TiN interface can be thought of as more heterogeneous, peppered with well-ordered atomic clusters. We interpret this change as ABAA-induced ionic diffusion allowing the oxide to find more stable configurations [4,5]. ABAA thus offers a straightforward means to modify both the structure and properties of ALD AlOx layers inside devices [11]. Overview of spatially-resolved FEM measurements. A 4D STEM dataset is depicted by a virtual brightfield (vBF) image (top left). Each diffraction pattern’s variance is calculated, normalized based on the pixel thickness, and averaged across each row. Variances are compared as a function of depth in the AlOx layer. A representative diffraction pattern is shown (top right), with rings corresponding to maxima in the variance. Spatially resolved, normalized variance plots are shown for the as-made (bottom left) and ABAA (bottom right) devices with colors corresponding to locations depicted by the vBF colorbar. FEM comparisons between AM and ABAA devices. The AM device shows reduced variance near the TiN interface, mostly mitigated via ABAA processing.
Superconducting qubits (SQ) are favorable technology for quantum computing due to their macroscopic nature, coherence times, and gate speeds [1]. Materials synthesis and processing methods for SQs are common to fabrication in the semiconductor industry, promising easy scalability. Josephson junctions (JJ) are critical components of SQs and consist of a thin dielectric layered between two superconducting electrodes [2]. The prototypical JJ is an Al/AlOx/Al sandwich structure, as aluminum forms a thin, self-limiting oxide [2, 3]. Amorphous alumina (a-AlOx) has also seen widespread adoption as a dielectric in semiconducting devices due to favorable properties such as high breakdown fields and low leakage currents [4]. Aluminum oxidation follows Cabrera-Mott (CM) mechanisms, where adatom-induced fields drive Al-ion diffusion towards the surface [3]. Pappas et al. introduced alternating bias assisted annealing (ABAA), an electrical processing technique to tune JJ resistance [5]. ABAA applies alternating voltage pulses, below the JJ breakdown field but above the CM field, hypothesized to facilitate ionic diffusion and anneal the oxide. Via scanning transmission electron microscopy (STEM), they demonstrated that ABAA induces a measurable change on the Al electron energy loss near-edge structure (EELS-ELNES), suggestive of potential changes in bonding and homogenization of the oxide [5]. Junction resistance with ABAA can be tuned to a narrow distribution, improving device performance [6]. Iaia et al. studied the influence of ABAA in thicker a-AlOx layers in superconducting resonators designed to study two level system defects (TLS) in the amorphous dielectric [7]. They note pronounced impact of ABAA applied and measured at mK temperatures (cryogenic alternating bias stimulation, CABS) on TLS behavior, which are reversible when the device is annealed to 10 K. In this work, we investigate the influence of ABAA on a-AlOx film structure using 4D STEM fluctuation electron microscopy (FEM). Synthesis and processing of the device is described in Iaia et al. [7]. Briefly, a ∼50 nm thick a-AlOx layer was synthesized via thermal atomic layer deposition (ALD), which differs from conventional a-AlOx growth in JJs. ABAA processing was applied with ±100 MV/m fields at 353 K over 30 hours. STEM sample thickness was controlled between 0.4 – 0.7 λEMFPAlOx. 4D STEM experiments were performed at 300 kV on the Lawrence Berkeley National Laboratory TEAM I, collected with a Dectris Arina detector, with a 1.2 mrad and 2 pA probe to minimize beam damage [8]. Thickness was calculated for each real space position based on the transmitted intensity of the diffraction zero-beam using a vacuum dataset for reference. The variance was normalized for thickness at each pixel and averaged across each row of diffraction patterns. Several datasets were collected, aligned, normalized, and averaged to improve statistics. Variance profiles are compared as a function of depth to understand a-AlOx medium range order across the device thickness. Figure 1 shows FEM analysis of a resonator. Fluctuation electron microscopy (FEM) variance was calculated using the Vr method for each pixel [9]. Variance is a metric which tracks how speckle intensity as a function of k fluctuates across a dataset [10]. Speckles originate from individual ordered atomic clusters in the amorphous material aligned on Bragg conditions [10]. A low variance is suggestive of a lower speckle-contrast amorphous halo with little fluctuation across the dataset, or a more uniform and amorphous structure. In contrast, a higher variance suggests stronger speckle contrast intensity fluctuations, with sources arising from both discrete speckles and changes across the dataset, or a more heterogeneous and ordered amorphous material. Figure 1 compares FEM normalized variance profiles between the as-made (AM) and ABAA devices, and Figure 2 isolates the peak variance values at ∼3 and ∼7 nm-1 as a function of distance from the TiN electrode. In the AM device, a pronounced decrease in the variance is observed near the TiN interface, suggesting a substantial reduction in amorphous structural order. This variance reduction extends several nm into the film. We speculate this may be attributable to TiN interfacial effects during the initial stages of the ALD growth. Datasets collected with different cross-sections, probe currents, shorter dwell times, and different scan geometries all show comparable TiN interfacial features, so we believe this effect is not attributable to electron or ion beam damage [8]. The ABAA-treated device shows only a modest variance reduction near the TiN interface. This suggests that ABAA increases the a-AlOx ordering near the interface. While the a-AlOx structure as a function of depth is more homogeneous after ABAA, the a-AlOx structure near the TiN interface can be thought of as more heterogeneous, peppered with well-ordered atomic clusters. We interpret this change as ABAA-induced ionic diffusion allowing the oxide to find more stable configurations [4,5]. ABAA thus offers a straightforward means to modify both the structure and properties of ALD AlOx layers inside devices [11]. Overview of spatially-resolved FEM measurements. A 4D STEM dataset is depicted by a virtual brightfield (vBF) image (top left). Each diffraction pattern’s variance is calculated, normalized based on the pixel thickness, and averaged across each row. Variances are compared as a function of depth in the AlOx layer. A representative diffraction pattern is shown (top right), with rings corresponding to maxima in the variance. Spatially resolved, normalized variance plots are shown for the as-made (bottom left) and ABAA (bottom right) devices with colors corresponding to locations depicted by the vBF colorbar. FEM comparisons between AM and ABAA devices. The AM device shows reduced variance near the TiN interface, mostly mitigated via ABAA processing.
Amorphous oxide tunneling barriers, primarily formed from aluminum, represent one of the most widely adopted platforms for superconducting quantum bits (qubits). To overcome challenges associated with defects and sample variance among the tunneling barriers, the methodology of alternating bias assisted annealing (ABAA) was introduced in Pappas et. al[1]. The process of applying alternating bias to the barrier and subsequently aging before use was shown to reduce defects in the barrier. Namely, defects that give rise to two-level systems, coupling to the qubit and expediting decoherence. In this work we replicate an expedited ABAA process through a combination of ab-initio molecular dynamics and machine-learned potentials, illuminating how ABAA effects the energy landscape of the barrier.
Abstract We demonstrate a transformational technique for controllably tuning the electrical properties of individual aluminum-oxide tunnel junctions. Using conventional test equipment to apply an alternating bias to a heated tunnel barrier, giant increases in the room temperature resistance, greater than 70%, can be achieved. The rate of resistance change is shown to be strongly temperature- dependent, and is weakly dependent on junction size in the sub-micron regime. In order to measure their tunneling properties at mK temperatures, we characterized transmon qubit junctions treated with this alternating-bias assisted annealing (ABAA) technique. The measured frequencies follow the Ambegaokar-Baratoff relation between the shifted resistance and critical current. Further, these studies show a reduction of junction-contributed loss on the order of ≈ 0.3 × 10−6, and indicate that there is a reduction in resonant-and off-resonant-two level system defects when compared to untreated samples. Imaging with high-resolution TEM shows that the barrier is still predominantly amorphous with a more uniform distribution of aluminum coordination across the barrier relative to untreated junctions. This new approach is expected to be widely applicable to a broad range of devices that rely on amorphous aluminum oxide, and is likely applicable to many other metal- insulator-metal devices.
Superconducting qubit technology is among the most promising hardware platform for quantum computing due to its fast gate operation, manufacturing scalability, and architectural flexibility [1]. A key figure of merit that dictates quantum computing functionality is the qubit’s coherence time, which needs to be maximized in order to bring quantum computing to the practical and scalable realm [2]. In particular, the T1 relaxation time—now on the order of milliseconds—can be enhanced with materials engineering, as amorphous dielectrics and crystallographic defects, among other factors, are known to induce decoherence [3-7]. Current state-of-the-art superconducting qubit devices use amorphous AlOx sandwiched between two superconducting Al electrodes as the Josephson junction (JJ), a key element in the transmon qubit, but amorphous dielectrics introduce two-level system (TLS) loss that underlie decoherence [8]. For this reason, numerous efforts have explored crystalline dielectrics as a replacement for amorphous AlOx, since single-crystal materials do not suffer from the lack of long-range order and sub-stoichiometry that comprise TLS defects [9-12]. However, none of these alternative materials choices have been able to surpass, let alone replicate, the performance of Al-AlOx-Al JJs. In this work, we fabricate superconducting quantum devices using a novel heteroepitaxial TiN/γ-Al2O3/TiN tri-layer stack grown on sapphire (Fig. 1a) and find that the measured microwave loss of the single-crystal γ-Al2O3 dielectric is lower than that of amorphous AlOx by two orders of magnitude. For qubits made with this tri-layer to transcend the coherence of incumbent Al-AlOx-Al, the structural, chemical, and electronic defects that induce TLS loss in these materials need to be well understood. Towards this end, we combine scanning transmission electron microscopy (STEM) and scanning electron microscopy (SEM) methods to highlight structural and interfacial complexity across length scales, ultimately identifying loss sources in the TiN/γ-Al2O3/TiN heteroepitaxial tri-layer. With atomic-resolution imaging and four-dimensional (4D-)STEM nanobeam electron diffraction (NBED), we find that the dielectric γ-Al2O3 layer, though single-crystalline (Fig. 1b), suffers from the formation of antiphase boundaries (Fig. 1c), which can host strong local electric fields and enhance lossy dielectric response [13]. Correlated with these nanoscale planar defects is uniformly distributed elastic strain, on the order of 6%, between the lattice-mismatched γ-Al2O3 and TiN (Fig. 1d). In other words, the dielectric layer is fully strained to match the underlying metal layer; such high strain can distort local bonding environments and thus create new TLSs or shift existing TLSs into resonance. Beyond imperfections within the bulk of the dielectric layer, there are noteworthy interfacial defects: roughening of both TiN/γ-Al2O3 interfaces deviate from atomically flat boundaries (Fig. 1c), while anion interdiffusion across both interfaces, as evident from core-loss electron energy-loss spectral (EELS) maps (Fig. 1e), leave behind vacancies and undercoordinated metal atoms, thereby creating 1.5 – 2 nm nonstoichiometric diffuse interfacial layers, both of which can cause TLS loss. In the superconducting TiN layers, we again use NBED to identify 180°-reflected specular twinned domains, indicating distinct heteroepitaxial relationships of [111]TiN//[0006]Al2O3 and [002¯]TiN//[22¯02¯]Al2O3 (Fig. 2a) [14]. Crystal twinning of the TiN epilayer is not only spatially distributed at the nanoscale, as shown using virtual dark-field (VDF) images (Fig. 2b), but is also widespread at the microscale, as shown using SEM electron backscatter diffraction (EBSD) (Fig. 2c). While microwave performance is not limited by a single factor, it is clear that defective quantum microstructure across length scales collectively contribute to dielectric loss, and in turn, TLS-related decoherence. By correlating nanoscale crystallographic and spectroscopic information with microscale diffraction measurements, our work can guide interface engineering during material growth, paving the way towards further improvements in quantum device performance. Cross-sectional samples were prepared by subsequent 30 kV, 8 kV and 5 kV Xe+ plasma focused ion beam (FIB) using the Thermo Scientific Helios 5 Hydra CX DualBeam Plasma FIB/SEM. Samples were then plasma cleaned using South Bay Technology PC-2000 Plasma Cleaner with a 3:1 mixture of Ar:O2 gas at 20∼40 W RF power for 15 s at 150 mTorr. High-angle annular dark-field (HAADF-) and annular bright-field (ABF-) STEM images were collected on a probe-corrected JEOL JEM-ARM200CF S/TEM operating at 200 kV using JEOL detectors. The STEM convergence angle, HAADF collection angle, and ABF collection angle were 27 mrad, 90 – 370 mrad, and 23 mrad, respectively. EELS data were collected on a JEOL JEM-ARM300F GrandARM S/TEM operating at 300 kV using the Gatan GIF Continuum K3 IS system. The data were collected using a 2.5 mm aperture and a 750 eV wide energy slit at a dispersion of 0.18 eV/Ch and a camera length of 2 cm. 4D-STEM data were collected on the same microscope and detector system using a 0.5 – 1.5 mrad convergence angle and a 20 eV wide energy slit at a camera length of 4 cm. All STEM data processing was performed using the Gatan Microscopy Suite software. Top-down EBSD measurements were performed in the Thermo Scientific Helios 5 CX DualBeam FIB/SEM at 25 kV using the Oxford Symmetry S3 detector. EBSD data processing was performed using the Oxford AZtecCrystal software [15]. Defects in γ-Al2O3. a) HAADF-STEM image of a TiN/γ-Al2O3/TiN tri-layer stack on a α-Al2O3 (0001) substrate. The thickness of γ-Al2O3 is approximately 13.5 nm. b) 4D-STEM NBED pattern of the γ-Al2O3 layer, confirming the {111} out-of-plane orientation. c) Fourier-filtered atomic-resolution ABF-STEM image of the γ-Al2O3 dielectric layer, taken from the cyan-colored square region in a), emphasizing the planar defects along both TiN/γ-Al2O3 interfaces and within the γ-Al2O3 layer itself–interfacial roughness (yellow lines) and antiphase boundaries (colored parallelograms), respectively. The scale bar is 2 nm. d) In-plane and out-of-plane strain maps of the heteroepitaxial tri-layer stack, with approximately 6.5∼6.8% strain in both directions. The units of the intensity bar are such that 100% strain would be equivalent to 1.0. e) STEM-EELS maps of Ti-L2,3, N-K, and O-K edges, along with a superimposed image of the three, taken from the same region as c). The scale bars are all 5 nm. Defects in TiN. a) 4D-STEM NBED patterns and b) VDF images of nanoscale TiN {111} and {002} twins. c) SEM EBSD pole figures in the three face-centered cubic crystallographic directions, revealing that the 60° rotation twinning behavior (with the [111] twin axis) is spread across the microscale.
Alternating-bias assisted annealing (ABAA) has been effectively utilized to tune the normal-state resistance of sub-micrometer Josephson junctions (JJs). However, how the annealing efficiency is affected by pre-annealing thermal history, oxidation exposure, and junction geometry remains unexplored for micrometer-scale devices. Here, we systematically investigate the ABAA behavior in micrometer-scale JJs to address the growing demands of modern superconducting devices. The dependence of resistance tuning on the pre-annealing heating time, annealing temperature, bias voltage, oxidation exposure, junction linewidth, and aspect ratio is thoroughly characterized. Our results will contribute a predictive framework for the precise resistance control of micrometer-scale JJs, thereby guiding the deterministic fabrication of high-performance superconducting quantum hardware.
The development of quantum computers, capable of surpassing classical computing in solving specific complex problems, hinges on breakthroughs in materials science, precision device fabrication, and the optimization of quantum architectures [1]. Among the leading candidates for quantum computation, superconducting qubits, particularly those based on transmon designs, have shown promising advances with coherence times reaching hundreds of microseconds [2]. However, to fully unlock the potential of scalable quantum computing, further enhancements in coherence are necessary. The intricate structure-coherence relationship within these devices, especially at critical interfaces such as metal-air, metal-dielectric, and metal-substrate in Josephson junctions (JJs), poses a significant challenge due to the potential for two-level-system loss channels. Here, we detail our pioneering efforts to understand these structure-coherence relationships utilizing advanced electron microscopy techniques. Focusing on the Al-AlOx and Al-Si substrate interfaces in Josephson Junctions, we employ a combination of low-loss and core-loss electron energy loss spectroscopy (EELS) and four-dimensional scanning transmission electron microscopy (4D-STEM) conducted at room and ultra-low temperatures (sub-6 K). This approach allows for an unprecedented examination of the chemical composition, bonding environments, and nanoscale structures of key surfaces and interfaces critical to the quantum coherence. Figure 1 presents a typical superconducting architecture at the chip level; each superconducting qubit consists of a Josephson Junction in parallel with an ultra-low-loss capacitor to create a resonant structure in the 3-6GHz range. JJs are highlighted in the area boxed in Fig. 1A and can be observed in detail in Fig. 1B. Lamella cross-sections (as shown in Fig. 1C) were prepared using a dual-beam focused ion beam (FIB) FEI-SEM Helios Nanolab. The lift-off was performed using 30 kV Ga+ ions, and the final cleaning step was carried out at 2 kV to remove surface damage on the area of interest, resulting in a final sample thickness of 50-100 nm. Transmission electron microscopy (TEM) and scanning transmission electron microscopy (STEM) data were collected on a probe-corrected JEOL ARM200 S/TEM operating at 200 kV. The convergent angle for ADF-STEM imaging was 25 mrad. Electron energy loss spectra (EELS) were acquired at 200 kV using a Gatan GIF Quantum on a K2 pixelated detector. Data processing (background subtraction, plural scattering removal, and signal mapping) was conducted using Gatan GMS software. Four-dimensional STEM (4D-STEM) data were collected by a K3 and STELA cameras. The convergent angles were 0.5-1.5 mrad, and the camera length was 6-10 cm. The step size was 5-10 Å. Merging microscopic insights with electrical data and theoretical frameworks, we aim to reveal how atomic-scale irregularities affect coherence times. Our results highlight key decoherence mechanisms in superconducting qubits and suggest specific improvements for qubit design and manufacturing. By linking electron microscopy findings to device performance, we outline strategies to reduce decoherence, essential for achieving reliable quantum computing. This work bridges materials science and quantum engineering, driving efforts in making scalable quantum computers forward [3]. A) Microscopic view of a superconducting transmon qubit, highlighting its symmetric design with central island and capacitor pads. B) SEM image of a JJ displaying their Al top and bottom contacts. C) STEM image of a JJ showing its key components and interfaces: Al top and bottom contact, AlOx tunnel barrier, Si substrate and Al-Si interface. A) EELS chemical maps of two interfaces, Al/Al and Al/Si using K-edges of oxygen, aluminum and silicon of a Josephson junction. It is possible to detect a formation of oxides at both interfaces. Plots showing Al and Si K-edge at 1560 eV and 1839 eV, respectively, and the ELNES are shown on the right side (Scale bar in A – 20 nm). B) Analysis of the disordered Al phases at the interfaces utilizing ePDF.
Superconducting quantum processors are one of the leading platforms for realizing scalable fault-tolerant quantum computation (FTQC). The recent demonstration of post-fabrication tuning of Josephson junctions using alternating-bias assisted annealing (ABAA) technique and a reduction in junction loss after ABAA illuminates a promising path towards precision tuning of qubit frequency while maintaining high coherence. Here, we demonstrate precision tuning of the maximum∣0⟩→∣1⟩transition frequency(f01max)of tunable transmon qubits by performing ABAA at room temperature using commercially available test equipment. We characterize the impact of junction relaxation and aging on resistance spread after tuning, and demonstrate a frequency equivalent tuning precision of 7.7 MHz (0.17%) based on targeted resistance tuning on hundreds of qubits, with a resistance tuning range up to 18.5%. Cryogenic measurements on tuned and untuned qubits show evidence of improved coherence after ABAA with no significant impact on tunability. Despite a small global offset, we show an empiricalf01maxtuning precision of 18.4 MHz by tuning a set of multi-qubit processors targeting their designed Hamiltonians. We experimentally characterize high-fidelity parametric resonance iSWAP gates on two ABAA-tuned 9-qubit processors with fidelity as high as 99.51±0.20%. On the best-performing device, we measured across the device a median fidelity of 99.22% and an average fidelity of 99.13$\pm$0.12%. Yield modeling analysis predicts high detuning-edge-yield using ABAA beyond the 1000-qubit scale. These results demonstrate the cutting-edge capability of frequency targeting using ABAA and open up a new avenue to systematically improving Hamiltonian targeting and optimization for scaling high-performance superconducting quantum processors.