Solid-state reactions at niobium–germanium interfaces in hybrid quantum electronics

In the past decade, research on hybrid superconducting–semiconducting (S–Sm) materials and devices has focused on realizing functionalities that enhance the operation of superconducting quantum devices. Hybrid S–Sm–S Josephson junctions (JJs) have been at the core of such efforts being integrated into devices such as voltage-tunable qubits (Gatemons) and voltage-tunable couplers.^1–5 S–Sm–S JJs rely on superconductivity induced within a semiconducting channel through Andreev reflection.⁶ This configuration simultaneously benefits from superconducting properties (i.e., dissipationless charge transport) and semiconducting properties (i.e., tunable carrier densities, long mean free paths, and large spin–orbit coupling).

Hybrid S–Sm systems composed of superconducting Al contacts coupled to InAs quantum wells (QWs) have been extensively studied as a platform for hybrid S–Sm quantum devices.^7,8 However, there are concerns with the high losses in the InAs heterostructures at microwave frequencies, as evidenced by low internal quality factors in superconducting resonators and qubits fabricated on Al–InAs platforms.^3,5,9–11 Current performance limitations have been attributed to the large defect densities within the heterostructures and the piezoelectricity in indium phosphide (InP) substrates.³ Therefore, exploring alternative materials systems is critical to realizing coherent gate-tunable quantum devices.

Ge QWs have recently emerged as promising alternatives to InAs QWs due to their high hole mobility, the possibility for growth on Si substrates with low microwave loss, and compatibility with wafer-scale fabrication processes. Several studies have demonstrated voltage-tunable hybrid Al–Ge QW JJs with critical currents ranging from 6 to 500 nA.^12,13 Al has been the sole superconducting contact layer in those studies, as it forms a stable ohmic contact to the Ge QWs.^12,14 Integrating superconducting materials with larger gaps, such as Nb, may be of interest for realizing a more robust induced superconducting gap in the Ge QW devices. Thus far, studies of hybrid Nb/Ge QW systems have been hindered by two challenges: (1) sputtering, which is conventionally used for Nb deposition and imparts significant physical damage to the substrate during the deposition; (2) solid-state chemical reactions between Nb and Ge under vacuum and at elevated temperatures are only studied over limited temperature windows primarily above 900 °C.¹⁵ 

In this study, we leverage a cryogenic ultrahigh vacuum (UHV) e-beam evaporation process to deposit thin films of Nb on Ge substrates directly. Using this method, we demonstrate chemically abrupt and atomically sharp Nb/Ge interfaces with near-Ohmic characteristics (barrier height ∼121 meV). We study thermally driven changes in the physical and chemical properties of the Nb/Ge heterostructures by annealing the UHV-grown films at a wide range of temperatures from 150 to 675 °C. With increased temperature, we observed a continuous rise in the Ge content of the Nb films, reaching 65% at 675 °C. This is accompanied by an increase in the barrier height at the Nb/Ge interface up to 575 °C, followed by a sharp drop due to the significant incorporation of Ge. The superconducting properties of the films show a slight improvement after a 300 °C anneal, mainly due to structural refinements in the Nb films. Further annealing led to the degradation of the superconducting properties.

To prepare the Nb/Ge heterostructures, we started with undoped Ge(001) wafers (ρ > 50 Ω cm). The Ge(001) substrates were cleaned by a 5 min sonication in acetone, isopropanol, and deionized water, followed by N₂ drying. The Ge substrates were then loaded into a UHV molecular beam epitaxy (MBE) system and annealed at 150 °C for 30 min before introduction into the growth chamber. With pressures less than 1 × 10⁻⁹ mbar, the MBE allows for the deposition of high-quality Nb thin films. The MBE system is equipped with a manipulator capable of cooling the sample to cryogenic temperatures (T < 200 K) to minimize intermixing between the Nb and Ge. Throughout the deposition, we used reflective high-energy electron diffraction (RHEED) to monitor the surface reconstruction of the Ge and Nb film. Figure 1(a) depicts a typical substrate temperature profile for the Nb growth process. It begins with Ge oxide desorption at 500 °C, followed by cooling to ∼−170 °C for the deposition of the Nb thin films. The Nb thin films, varying in thickness from 8 to 200 nm, were deposited using a vertical electron gun at rates of 0.8–1.1 Å/s.

We examined the nanoscale structure of the cryogenically grown Nb/Ge heterostructure with cross-sectional high-angle annular dark field (HAADF) transmission electron microscope (TEM) imaging (JEOL-ARM 200F operated at 200 kV). The cross-sectional scanning trasmission electron microscope (STEM) specimen was prepared using an FEI Nova 200 dual-beam focused ion beam. Based on the HAADF images, the as-grown 100 nm thick Nb film is nanocrystalline with an average grain size of 8 nm [see Fig. 1(b)]. Figure 1(c) shows energy-dispersive x-ray spectroscopy (EDS) maps and line scans across a narrow Nb/Ge interface region. The EDS results confirm that the growth at −170 °C leads to a chemically abrupt interface with minimal intermixing and a narrow interface layer (<4 nm) between the Nb and Ge. For our TEM sample, with a thickness of 50 nm, the beam broadening is estimated to be 4.07 nm, confirming an atomically sharp Nb/Ge interface. This abruptness provides a baseline for testing the influence of thermal processing on the physical and chemical properties of the Nb/Ge heterostructures.

To explore the changes in the surface and interface electronics, we use femtosecond ultraviolet photoelectron spectroscopy (fs-UPS) (see supplementary material for fs-UPS details).^16–18  Figure 2(a) shows the valence band spectra measured by UPS for a 30 nm thick Nb film deposited on Ge(001) and grown at −170 °C. The spectra is measured as a function of in situ UHV annealing at temperatures ranging from 150 to 675 °C. After the samples were grown, they were stored in a stainless steel canister and pressurized to 2.0 bars with ultrahigh-purity N₂ to avoid any oxidation during the transfer to the fs-UPS lab. The samples were then immediately loaded into the UHV system. As the sample temperature increases, the valence band edge becomes progressively sharper, consistent with a larger density of states at the Fermi level and a good metallic behavior. Additionally, above 625 °C, new peaks emerge at higher binding energies (BE = 6.5 and 11.7 eV) that may be due to Ge incorporation into the Nb film.

Figure 2(b) shows the band bending at the Nb/Ge interface as a function of anneal temperature through a pump and probe process. To determine the band bending at the Nb/Ge interface, synchronized 1.55 eV pump pulses are concentrated on the probe area. This action generates an electron–hole pair within the interface, effectively counteracting the static dipole field within the depletion region, thus flattening the energy bands. Consequently, the UPS spectrum undergoes a consistent shift in energy. The extent of band bending is determined by measuring this shift. Initially, the sample exhibits a relatively low band bending of 121 meV at 150 °C, which progressively rises with temperature to 300 meV at 575 °C. This is followed by a sharp fall in the band bending to 144 meV upon longer anneal at 575 °C. With further increase in temperature, band bending continuously decreases to 110 meV at 675 °C. This decrease in the Schottky barrier height may be attributed to the degrading chemical abruptness of Nb/Ge interfaces at elevated temperatures consistent with the increased work function of our annealed Nb films (see supplementary material, Fig. S1).

To confirm that the additional peaks in the valence band spectra are due to Ge out-diffusion in Nb, we repeated our fs-UPS measurements on 200 nm thick Nb films on Ge(001) as shown in Fig. 2(c). Similar to thinner Nb films, high binding energy peaks emerged in the valence band spectra when 200 nm thick Nb films were annealed at 575 and 675 °C. Nevertheless, the peak at 11.7 eV exhibited a greater intensity in thicker films compared to the peak at 6.5 eV. Moreover, the annealing led to a sharper valence band edge on the 200 nm thick films. The sharp valence bands and lower band bending are indicators of an Nb/Ge interface that may be suitable for hybrid S–Sm JJs. However, the influence of Ge incorporation during the annealing process on surface and interface electronics could potentially restrict its beneficial effects.

To investigate the impact of annealing on the composition of the Nb films, we performed compositional analysis on the Nb/Ge heterostructures using ex-situ x-ray photoelectron spectroscopy (XPS). The data were collected using a PHI VersaProbe III with a monochromatic Al Ka x-ray source (hv = 1486.6 eV). (See supplementary material for details.) Figures 3(a) and 3(b) show the Nb3p and Ge2p spectra for the Nb/Ge samples as a function of annealing temperature. The samples were annealed ex-situ in the UHV MBE system and were transferred to the XPS lab under N₂. They were immediately loaded into the XPS system and underwent three cycles of Ar sputtering at 2 kV over a 3 × 3 mm² area. For the as-grown sample (black traces), no Ge signal is detected while the Nb3p peaks appear at the expected binding energies for elemental Nb (BE_Nb3/2 = 360.7 eV).¹⁹ On the other hand, annealing the sample to 300 °C shows the emergence of Ge on the top surface. The Ge2p peaks continue shifting to higher binding energies, with temperature eventually aligning with the expected binding energies of elemental Ge2p peaks (BE_Ge3/2 = 1217.4 eV).²⁰ Similarly, the Nb3p peaks shift to higher binding energies by increasing the annealing temperature. This is consistent with a notable accumulation of Ge near the top surfaces of Nb films, with a fraction of the Ge atoms bonding to Nb atoms.

Figure 3(c) displays the elemental composition of the Nb films derived from the Nb3p, Ge2p, and O1s peaks. To calculate the intensity of each peak, its area was first corrected by its relative sensitivity factor (RSF). The intensity of each element was then divided by the total sum of intensities across all three peaks to yield the elemental ratios. The as-grown sample started with Nb:Ge of 1:0. After annealing at 300 °C, Nb:Ge changed to 1:0.11. The final two anneal temperatures (500 and 675 °C) show further Ge incorporation with Nb:Ge of 1:0.18. Considering the small inelastic mean free paths²¹ for the Nb3p (1.07 nm) and Ge2p (0.40 nm) photoelectrons, our results point to sizable Ge accumulation on the top surfaces of Nb even at temperatures as low as 300 °C.

To complement the surface composition, derived from XPS, with bulk film composition, we utilized Rutherford Backscattering Spectrometry (RBS). Three 30 nm thick Nb films were studied by RBS: (1) as-grown, (2) annealed at 585 °C, and (3) annealed at 675 °C. The samples were annealed in situ in UHV and transferred to the RBS system in a stainless steel canister filled with 2.0 bars of ultrahigh-purity N₂ to prevent oxidation. Table I summarizes the RBS compositional analysis results for the three Nb/Ge samples. The as-grown sample shows no Ge incorporation consistent with XPS results. On the other hand, the two samples annealed at 585 and 675 °C show a 1:1 and 1:1.85 of Nb:Ge, respectively. These results are in agreement with the cross-sectional STEM-EDS measurements on the films, where significant Ge diffusion was observed at 650 °C (see supplementary material, Fig. S2). Moreover, Ge diffusion resulted in increased grain size and surface roughness in the Nb films (see supplementary material, Fig. S3). We anticipate the presence of Ge on the top surface and within the bulk to change the superconducting properties of the Nb films.

To evaluate the influence of thermal processing on the superconducting properties of the Nb/Ge heterostructures, we utilized a cryogenic measurement system (Oxford Instruments, Teslatron PT). For a detailed evaluation of superconducting parameters such as superconducting transition temperature (T_C), critical magnetic field (B_C), and critical current density (J_C), Nb thin films were fabricated into microwires of 100 μm length and varying widths (5, 10, and 20 μm). Details of the microfabrication process are provided in the supplementary material. Figure 4 displays the transport measurement results for 10 μm wide microwires fabricated on 100 nm thick Nb films that were as-grown (blue), annealed at 300 °C (green), and annealed at 500 °C (red). The annealing was conducted on the thin films in UHV (5 × 10⁻¹⁰ mbar) immediately after the cryogenic growth. Table II lists the key superconducting parameters extracted from the transport measurements, including the zero temperature critical magnetic field (B₀), zero temperature critical current density (J₀), BCS superconducting gap (Δ ≈ 1.76k_BT_C, where k_B is the Boltzmann constant), and the superconducting coherence length (⁠$ξ0=Φ02πB0$⁠, where Φ₀ is the flux quantum).

The superconducting transition increases from 8.3 K for the as-grown sample to 9.2 K (value for bulk Nb) after UHV annealing at 300 °C. On the other hand, annealing at higher temperatures, such as 500 °C, degrades the superconducting properties of the films, as evidenced by a significant drop in T_C to 2.3 K. Similarly, J_C increases after annealing at 300 °C and significantly decreases at 500 °C. The behavior can be attributed to the increased average grain size in the films (see supplementary material, Fig. S3). However, the influence of larger grain size is counteracted by large Ge content at 500 °C, leading to degraded superconducting properties. Our reported J₀ values are based on fitting the experimental data to the Ginzburg–Landau (GL) theory for depairing critical current: J_C(t) = J₀(1 − t²)^α (1 + t²)^1/2, where t = T/T_C (see supplementary material, Fig. S4). We note that the temperature dependence of J_C deviates from the conventional GL theory where α = 1.5 for samples that were as-grown or annealed at 500 °C (α ≈ 1).^22,23 This deviation may be due to the disorder in the two films. In the former case, the disorder stems from the nanocrysallinity of the cryogenically grown Nb samples. In the latter case, the large Ge content leads to atomic scale disorder within the Nb–Ge grains (see supplementary material, Fig. S2).

Unlike T_C and J_C(T), B_C(T) and B₀ values significantly declined with annealing temperature. The as-grown sample has the highest B₀ of 5.3 T. Annealing the Nb/Ge samples to 300 and 500 °C decreased B₀ to 2.2 and 0.1 T, respectively. Despite the significant change in its absolute values, the B_C for our Nb/Ge samples maintained a linear temperature dependence irrespective of the annealing conditions. B₀ and ξ₀ were estimated by fitting the B_C(T) data for each sample to the linear Ginzburg–Landau (GL) relationship B_C(t) = B₀(1 − t), where t = T/T_C. Linear B_C(T) has been previously reported in two-dimensional disordered superconducting films of Ge:Ga and InO.^24,25 This picture is once again consistent with a disorder in the Nb/Ge heterostructures due to either nanocrystalline Nb structures (as-grown cryogenically) or solid-state Nb–Ge alloying at temperatures as low as 300 °C.

In summary, we demonstrated chemically abrupt and atomically sharp Nb/Ge interfaces by UHV evaporation of Nb on Ge substrates held at temperatures as low as 100 K. Cryogenic growth also minimizes the physical damage to the interface. As-grown Nb/Ge interfaces show a relatively low band bending (121 meV) that can benefit future hybrid S–Sm devices. Annealing the samples to temperatures above 575 °C led to even smaller interface band bending at the cost of significant Ge incorporation (up to 65 at. %) into the Nb thin films. Additionally, annealing at such high temperatures led to the degradation of superconducting properties in the Nb films, including critical temperature, current, and magnetic field. Our results show a low thermal budget for realizing low-barrier Nb–Ge interfaces, making a low-temperature or cryogenic Nb growth approach essential in fabricating future hybrid Nb–Ge–Nb devices.