Characteristics of Nb/Al superconducting tunnel junctions fabricated using ozone gas
Ukibe Masahiro†, Fujii Go, Ohkubo Masataka
Nanoelectronics Research Institute (nano), National Institute of Advanced Industrial Science and Technology (AIST), 1-1-1 Umezono, Tsukuba, Ibaraki 305-8568, Japan

Corresponding author. E-mail: ukibe-m@aist.go.jp

Abstract

To improve the energy resolution (Δ E) of Nb/Al superconducting tunnel junctions (STJs), an ozone (O3) oxidation process has been developed to fabricate a thin defect-free tunnel barrier that simultaneously shows high critical current JC > 1000 A/cm2 and high normalized dynamic resistance RD A > 100 MΩ · μm2, where A is the size of the STJ. The 50-μm2 STJs produced by O3 exposure of 0.26 Pa· min with an indirect spray of O3 gas, which is a much lower level of exposure than the O2 exposure used in a conventional O2 oxidation process, exhibit a maximum JC = 800 A/cm2 and a high RD A = 372 MΩ · μm2. The 100-pixel array of the 100-μm2 STJs produced using the same O3 oxidation conditions exhibits a constant leak current Ileak = 14.9 ± 3.2 nA at a bias point around Δ / e (where e is half the energy gap of an STJ), and a high fabrication yield of 87%. Although the Ileak values are slightly larger than those of STJs produced using the conventional O2 oxidation process, the STJ produced using O3 oxidation shows a Δ E = 10 eV for the C-K α line, which is the best value of our Nb/Al STJ x-ray detectors.

PACS: 33.20.Kf; 33.70.Jg
Keyword: Nb/Al superconducting tunnel junctions; high critical current density; high energy resolution; ozone
1. Introduction

Fluorescence-yield x-ray absorption fine structure (FY-XAFS) spectroscopy is useful to characterize trace elements (dopants) in functional materials, such as next-generation compound semiconductor materials. In conventional XAFS measurements, semiconductor detectors are used to detect fluorescent x-rays. However, the energy-resolving power of the semiconductor detectors is not sufficient to clearly distinguish the K lines of light trace elements (e.g., boron and nitrogen) from the L lines of various elements, and the K lines of light elements in matrices in the energy range below about 3  keV. This insufficiency makes measurement of low-concentration dopants difficult. Superconducting tunnel junction (STJ) detectors are promising for XAFS measurements in the soft x-ray region, because of their high energy resolution and high sensitivity, both of which exceed those of semiconductor detectors.[13] In fact, our 100-pixel Nb/Al STJ array detectors had a mean energy resolution Δ E = 12  eV for C-Kα (277  eV) photons. At this high-energy resolution, the XAFS spectra of 300  ppm nitrogen dopants in n-SiC samples were successfully obtained using a superconducting XAFS (SC-XAFS) apparatus.[4] With such high Δ E, it was possible to resolve the K lines of light trace elements, but it is difficult to achieve clear separation of the L lines of transition elements and the K lines of light elements. Therefore, further improvements in energy resolution are desirable. In general, the high critical current density (JC) of STJs increases the x-ray signal output, and the low normalized dynamic resistance (RDA) of STJs increases the circuit (system) noise when a charge preamplifier is used. Here, A is the size of the STJs. In a previous article, [5] it was shown that one of the possible ways to improve Δ E is to increase JC to over 1000 A/cm2, while maintaining a high RDA of more than 100  MΩ · μ m2 (or a low leak current (Ileak)) of the STJs. When Nb/Al STJs produced using the conventional fabrication process (an O2 oxidation process) show high JC values, they tend to exhibit low RDA values (i.e., they have a large Ileak).[6] Motivated by these drawbacks, a new barrier formation process using ozone (O3), which exhibits much stronger oxidation than O2 gas, was introduced to fabricate a thin defect-free tunnel barrier. The dependence of JC, Ileak, and RD on the O3 oxidation condition, which is represented by the product of the oxidant gas pressure (P) and the oxidation time (t), has been evaluated.[5] However, the STJs produced using the O3 oxidation process exhibited extremely small JC and Ileak values, which indicated a thick tunnel barrier. Even in the case of the lowest O3 exposure of 0.72  Pa· min, which is a much lower level of exposure than the O2 exposure applied in the conventional O2 oxidation process (7.2 × 104  Pa· min), the JC values were as small as 8  A/cm2 and the Ileak value was lower than 50  pA, which was the lower limit of the measurement system. These results showed that the O3 oxidation conditions must be optimized to realize a thin defect-free tunnel barrier and high JC values of more than 1000  A/cm2. In this study, the O3 oxidation process was modified by changing the method of supplying the O3 gas and the O3 exposure. The performance of Nb/Al STJs made using the modified O3 oxidation process was evaluated, and the dependences of JC and RD on the O3 oxidation conditions, the fabrication yields, and the Δ E for soft x-rays were investigated.

2. Experiment

All STJs have an asymmetric layer structure. Here, this structure was Nb (300  nm)/Al (70  nm)/AlOx/Al (70  nm)/Nb (50  nm). Two different types of STJ arrays were fabricated. The first was an 8-pixel array consisting of STJs with a size of 50 × 50  μ m2, which was used to measure the full current– voltage (IV) curves. Because the critical current (IC) of the superconducting Nb wirings (cross-section ∼ 1.2× 10− 7  cm2) used in the STJs was less than 0.1  Å , [7] it was necessary to use STJs with a size of 50 × 50  μ m2 to prevent the current flow in the Nb wirings from exceeding the IC values of the Nb wirings when JC was measured. The second was a 100-pixel array of STJs with a size of 100 × 100  μ m2, which was used to evaluate the x-ray detection performance. Figure  1 shows a top view of a Nb/Al STJ with a size of 50 × 50  μ m2. The STJs were fabricated using a conventional photolithographic technique, DC magnetron sputtering, a lift-off technique, reactive ion etching, and wet etching processes. These processes are described in detail in the literature.[8] The Nb/Al multilayer deposition and the tunnel barrier formation were performed by the same DC-sputtering apparatus, without breaking the vacuum, at room temperature. In the O3 oxidation process, we reported in a previous article, [5] the oxidant gas was a mixture of gases prepared by diluting the O3 (10%)/O2 (90%) gas supplied by the O3 generator with Ar gas, and this mixture was sprayed directly onto the surface of the bottom Al layer. However, the obtained relationship between JC and Pt (the product of the oxidant gas pressure (P) and the oxidation time (t)) was JC ∝ (Pt)− 0.6, which was considered to be almost the same as the relationship of the low-JC data described in Ref.  [6] by taking into account the measurement error of theJC values. This means that in the previous experiments, the influence of Al oxidation by O2 gas during the total oxidation process was much larger than we first thought. Therefore, two alterations were made to the O3 oxidation process. The first was that, to reduce the influence of O2 gas as much as possible, one of the source gases of the oxidant gas was changed from a mixture of O3 (10%)/O2 (90%) to a mixture of O3 (1.1%)/Ar (98.9%), which was supplied by the O3 gas cylinder.[9] The second was that the oxidant gas was sprayed indirectly onto the surface of the bottom Al layer to make the gas composition of the oxidant gas more uniform and avoid the oxidation by dense clouds of O3 gas. The properties of the junctions prepared using the modified O3 oxidation process were compared with those of junctions prepared using O2 oxidation. In the O2 process, O2 gas with 99.99% purity was used as the oxidant. The O2 exposure was 7.2 × 104  Pa· min, which is the standard value used to fabricate STJ detectors for the soft x-ray region. The oxidation conditions used in the modified process are listed in Table  1.

Fig.  1. Micrograph of an STJ used for the current– voltage (IV) measurements.

Table 1. Summary of oxidation conditions.

To create appropriate O3 oxidation conditions, the O3 exposure should be significantly lower than the exposure typically applied in O2 processes, because oxidation by O3 gas is much stronger than that by O2 gas.[10] Measurement of the junction properties was performed using a cryogen-free 3He cryostat with a base temperature of 0.31  K. The IV characteristics were measured to evaluate JC, the normal resistance (RN), and RD for all of the oxidation conditions. Note that JC was measured in a zero field, and RD was obtained using a linear approximation of the slope in the subgap region of the IV curve in a magnetic field (B) of about 10 mT, which was applied along the diagonal direction of the STJs. The quality factor (Q) was then calculated by RD/RN. The x-ray detection performance (Δ E and the rise time (τ r) of the x-ray signals) of the 100-pixel array was evaluated using the fluorescence x-rays generated by an x-ray tube. The x-ray tube consisted of a coniferous carbon nanostructure (CCNS) cathode[11] and a carbon compound target deposited on a pure Al plate. The fluorescent x-rays were generated by irradiating a carbon compound target with low-energy electrons (energies < 1  keV). The STJs were biased around Δ /e using constant bias currents. The magnetic field conditions were the same as those used to measure RD, Ileak is defined as the subgap current value around Δ /e. The x-ray signals from the STJ array were processed using 100-channel charge-sensitive amplifiers and 100-channel FPGA-DSP based multi-channel analyzers.[12]

3. Result and discussion
3.1. Relationship between JC,   RD,   and  O3  exposure

Figure  2 shows the IV curves of two STJs made using different oxidation conditions: O2 exposure of 7.2 × 104  Pa· min and O3 exposure of 0.264  Pa· min. Figures  2(a) and 2(c) show that JC = 220  A/cm2 and RDA = 2300  MΩ · μ m2. The Q value of 2.4 × 106 is large enough for x-ray detection. In Figs.  2(b) and 2(d), JC and RDA are 800  A/cm2 and 372  MΩ · μ m2, respectively. The Q value of 8.2 × 105 is large enough for x-ray detection. These values of the STJ made using the O3 oxidation process are satisfactory. This is particularly true for the JC value, which is approximately four times larger than that measured for the STJ fabricated using the conventional O2 oxidation process. The JC values are evaluated using the current values at the points of inflection outside the subgap regions of the IV curves. It is assumed that the steep slopes observed around zero voltage in Figs.  2(a) and 2(b) are generated by the small resistances of about 0.01  Ω and 0.07  Ω , respectively, produced by the Au bonding pad. In the case of O3 oxidation at 0.264  Pa· min, it is difficult to accurately ascertain the position of the inflection points outside the subgap region of the IV curves because the resistance of the Au bonding pads is close to RN, which lead to uncertainty in the JC value. However, the results obtained using an O3 exposure of 0.264  Pa· min indicate that it is possible to realize a thin tunnel barrier while maintaining a high RD value by using an appropriate O3 exposure. For 264  Pa· min O3 oxidation, the Ileak value is less than 50  pA, which is the lower limit of our current measurement system. Therefore, all of the RDA values of the STJs are larger than 20000  MΩ · μ m2. The properties of all of the STJs are shown in Table  2.

The JC value measured for 0.264  Pa· min is the highest in our evaluation but is not significantly different from that measured for 2.64  Pa· min because of the above-mentioned uncertainty in the JC value measured at 0.264  Pa· min. However, the overall relationship between JC and O3 exposure in the O3 process using the indirect spray of the gas was considered to be valid, and the dependence of JC on O3 exposure (except for the case of 0.264  Pa· min) was JC ∝ (Pt)− 1.2, which is quite different to that for the low-JC data described in Ref.  [6]. This indicates that the Al oxidation process by O3 gas is different from that by O2 gas. The relationship between JC and O3 exposure in the O3 process is plotted in Fig.  3 for the processes performed using indirect and direct spray of the gas.[5]

Fig.  2. IV curves of two STJs made using different oxidation conditions: (a) and (c) O2 exposure of 7.2 × 104  Pa· min, and (b) and (d) O3 oxidation at 0.264  Pa· min. The JC and RD values were measured at B = 0  mT ((a) and (b)) and B = 10  mT ((c) and (d)), respectively.

Table 2. Summary of junction properties.

Fig.  3. Relationship between JC and O3 exposure in the O3 process for indirect (solid circles) and direct (open circles) spray of the gas. The solid and dashed lines are fit to the data for indirect and direct spray of the gas, and indicate the dependencies JC ∝ (Pt)− 1.2 and JC ∝ (Pt)− 0.6, respectively.

It was found that an indirect spray of O3 was very effective in reducing the tunnel barrier thickness, which led to the increase in JC. The best O3 oxidation conditions in our evaluation were an O3 exposure of 0.264  Pa· min using an indirect spray of gas. Under these conditions, the highest JC value of 800 ± 28  A/cm2 was achieved.

3.2. X-ray detection performance of STJs fabricated under the best O3 oxidation conditions

The x-ray detection performance of the STJs fabricated under the best O3 oxidation conditions was evaluated. The size of these STJs was 100 × 100  μ m2. 87 out of 100 STJs in the array detector exhibited Ileak < 30  nA, i.e., RDA > 100  MΩ · μ m2, which means that the fabrication yield was 87%. These STJs were good enough to detect x-ray photons. The average and standard deviations of the leakage current were 14.9  nA and 3.2  nA, respectively. The excellent uniformity of the leakage currents allowed each STJ in the array detector to be operated at a uniform bias current. Figure  4 shows an IV curve of an STJ in the 100-pixel array at B = 10  mT. The fluorescent x-rays from the x-ray tube were measured, and a typical x-ray signal pulse is shown in Fig.  5.

The rise time (τ r) of the x-ray signal pulse was 21  μ s, which is approximately twice as long as that of conventional STJs (∼ 12  μ s). This increase in τ r occurred because the frequency of the quasiparticle tunneling increased. This was the result of the increased tunneling provability, which was itself a result of the high JC.[13, 14] A fluorescent x-ray spectrum is shown in Fig.  6.

Fig.  4. IV curve for an STJ of the 100-pixel array made using an O3 exposure of 0.264  Pa· min at B = 10  mT.

Fig.  5. Typical x-ray signal pulse obtained using an STJ in the array made using an O3 exposure of 0.264  Pa· min.

Fig.  6. X-ray spectrum of an STJ in the array for measuring fluorescent x-rays from the x-ray tube. The C-Kα , O-Kα , and Fe-Lα x-rays in the spectrum were emitted from a carbon compound target, oxide in the carbon compound target, and iron in the Al plate, respectively. The x-ray signals from approximately 700– 900  ch correspond to bremsstrahlung x-ray photons generated in the Al plate of the target.

The energy resolution of C-Kα and the circuit noise were 10 and 2.8  eV (FWHM), respectively. The O-Kα and Fe-Lα x-rays shown in Fig.  6 were emitted from oxide in the carbon compound target and iron in the Al plate of the target, respectively. The x-ray signals from approximately 700– 900  ch in Fig.  6 correspond to bremsstrahlung x-ray photons generated in the Al plate of the target. The Δ E measured for the C-Kα line is the best Δ E value measured for our Nb/Al STJs to date. This value is higher than that for silicon drift detectors, which are the semiconductor detectors that currently give the best Δ E value in the soft x-ray region.[15] We therefore succeeded in realizing an STJ array detector with high JC and good energy resolution using the O3 oxidation process.

4. Conclusion

To improve the Δ E of Nb/Al STJs, the O3 oxidation process was modified to fabricate a thin defect-free tunnel barrier that simultaneously showed a high critical current JC > 1000  A/cm2 and a high normalized dynamic resistance RDA > 100  MΩ · μ m2. The 50-μ m2 STJs produced using O3 oxidation of 0.26  Pa· min with the indirect spray of the gas, which is a much lower level of exposure than the O2 exposure used in the conventional O2 oxidation process, exhibited a maximum JC = 800  A/cm2 and a high RDA = 372  MΩ · μ m2. The 100-pixel array of 100-μ m2 STJs made using the same O3 oxidation conditions showed a constant Ileak = 14.9 ± 3.2  nA at the bias point, and a high fabrication yield of 87%. Although the Ileak values were slightly larger than those of STJ arrays produced using the conventional O2 oxidation process, the STJ produced using O3 oxidization showed a Δ E = 10  eV for the C-Kα line, which is the best value yet shown for our Nb/Al STJs.

Acknowledgments

The authors thank S. Shiki for his help with the experiments, and T. Adachi and the clean room members of analog-digital superconductivity (CRAVITY) for performing the STJ fabrication.

Reference
1 Ukibe M, Shiki S, Kitajima Y and Ohkubo M 2012 Jpn. J. Appl. Phys. 51 010115 DOI:10.7567/JJAP.51.010115 [Cited within:1]
2 Fons P, Tampo H V. Kolobov A, Ohkubo M, Niki S, Tominaga J, Carboni R, Boscherini F and Friedrich S 2006 Phys. Rev. Lett. 96 045504 DOI:10.1103/PhysRevLett.96.045504 [Cited within:1]
3 Friedrich S, Funk T, Drury O, Labov S E and Cramer S P 2002 Rev. Sci. Instrum. 73 1629 DOI:10.1063/1.1445826 [Cited within:1]
4 Ohkubo M, Shiki S, Ukibe M, Matsubayashi N, Kitajima Y and Nagamachi S 2012 Sci. Rep. 2 831 DOI:10.1038/srep00831 [Cited within:1]
5 Ukibe M, Fujii G and Ohkubo M 2014 J. Low Temp. Phys. 176 142 DOI:10.1007/s10909-013-1073-5 [Cited within:4]
6 Miller R E, Mallison W H, Kleinsasser A W, Delin K A and Macedo E M 1993 Appl. Phys. Lett. 63 1423 DOI:10.1063/1.109697 [Cited within:3]
7 Youssef A E, Švindrych Z, Hadač J and Jansů Z 2007 WDS’s 07 Proceedings(Part III) 48 [Cited within:1]
8 Ukibe M, Kushino A, Chen Y and Ohkubo M 2007 X-ray Spectr. 36 260 DOI:10.1002/(ISSN)1097-4539 [Cited within:1]
9 http://www.iwatani.co.jp/jpn/div/ing_wem/ele_gas/cylinder/main.html [Cited within:1]
10 Horvath M, Bilitzky L and Huttner J 1985 Ozone New York Elsevier [Cited within:1]
11 Kato H, O’Rourke B E and Suzuki R 2014 JJAP Conf. Proc. 2 011302 DOI:10.7567/JJAPCP.2.0113022014 [Cited within:1]
12 Shiki S, Ukibe M, Kitajima Y and Ohkubo M 2012 J. Low Temp. Phys. 167 748 DOI:10.1007/s10909-012-0526-6 [Cited within:1]
13 Mears C A, Labov S E and Barfknecht A T 1993 Appl. Phys. Lett. 63 2961 DOI:10.1063/1.110286 [Cited within:1]
14 Goldie D J, Brink P L, Patel C, Booth N E and Salmon G L 1994 Appl. Phys. Lett. 64 3169 DOI:10.1063/1.111328 [Cited within:1]
15 Schlosser D M, Lechner P, Luts G, Niculae A, Soltau H, Struder L, Eckhardt E, Hermenau K, Schaller G, Schopper G, Jaritschin O, Liebel A, Simsek A, Fiorini C and Longini A 2010 Nucl. Instrum. Methods Phys. Res. , Sect. A 624 270 DOI:10.1016/j.nima.2010.04.038 [Cited within:1]