High quality NbTiN films fabrication and rapid thermal annealing investigation
Ge Huan1, Jin Yi-Rong1, 2, Song Xiao-Hui1, †
Beijing National Laboratory for Condensed Matter Physics, Institute of Physics, University of Chinese Academy of Sciences, Beijing 100190, China
Beijing Academy of Quantum Information Sciences, Beijing 100193, China

 

† Corresponding author. E-mail: xhsong@iphy.ac.cn

Abstract

NbTiN thin films are good candidates for applications including single-photon detector, kinetic inductance detector, hot electron bolometer, and superconducting quantum computing circuits because of their favorable characteristics, such as good superconducting properties and easy fabrication. In this work, we systematically investigated the growth of high-quality NbTiN films with different thicknesses on Si substrates by reactive DC-magnetron sputtering method. After optimizing the growth conditions, such as the gas pressure, Ar/N2 mixture ratio, and sputtering power, we obtained films with excellent superconducting properties. A high superconducting transition temperature of 15.5 K with narrow transition width of 0.03 K was obtained in a film of 300 nm thickness with surface roughness of less than 0.2 nm. In an ultra-thin film of 5 nm thick, we still obtained a transition temperature of 7.6 K. In addition, rapid thermal annealing (RTA) in atmosphere of nitrogen or nitrogen and hydrogen mixture was studied to improve the film quality. The results showed that Tc and crystal size of the NbTiN films were remarkably increased by RTA. For ultrathin films, the annealing in N2/H2 mixture had better effect than that in pure N2. The Tc of 10 nm films improved from 9.6 K to 10.3 K after RTA in N2/H2 mixture at 450 °C.

1. Introduction

NbTiN thin films are widely investigated as candidates for many superconducting devices, including nanowire superconducting single photon detectors (SNSPD),[14] hot electron bolometer (HEB) mixers,[57] superconductor–insulator–superconductor (SIS) devices in the THz band,[810] and coplanar waveguide (CPW) superconducting resonators.[11] Unlike NbN films, the introduction of the third element can locally modify the atomic binding energy, leading to electronic structure and crystal lattice modifications.[1214] Such modifications can improve the physical properties such as the electrical resistivity and the superconducting transition temperature.[14] In particular, it has been shown that NbTiN films surpass NbN films in many aspects for single-photon detectors because of the better uniformity and the smaller kinetic inductance.[15] In addition, NbTiN presents lower calculated surface impedance and better surface properties than NbN.

Several methods have been investigated to obtain high quality NbTiN thin films in the past few years. One of them is to introduce a buffer layer with the lattice constant similar to that of both the NbTiN film and the substrate. As a result, the lattice matching is improved and consequently the film quality is improved. For example, the AlN buffer layer can greatly improve the quality of NbTiN films.[16,17] This technique, although very effective, would introduce additional impurities, which thus degrades the device performance. Another method for high quality films is to grow the films on lattice-matched substrates. The NbTiN thin films fabricated on MgO or sapphire substrates usually have the high transition temperature and the low resistivity compared to those fabricated on Si or SiO2/Si substrates.[18,19] In many applications, however, the silicon based substrates are preferred or even necessary for large-scale integrating due to the compatibility to semiconducting industry.

In the present study, we focused on the optimization of growth conditions for the NbTiN thin films on SiO2/Si or intrinsic Si substrates. In addition, to further improve the superconducting properties of the NbTiN films, a rapid thermal annealing (RTA) method was also investigated at different temperatures. Post-annealing can promote the grain growth and decease the defects and disorders in the thin films, as discussed in Ref. [20]. However, some other previous results of annealing of Nb or NbN films showed degraded superconducting properties, possibly due to the oxygen atoms diffused into the grains.[21] Here, to clarify the annealing mechanism in nitride thin films, post-annealing of the NbTiN films was studied systematically. Our study shows that the transition temperature Tc increases evidently after RTA, but the residual resistivity increases too. As a comparison, we tried the thermal annealing treatment in both N2 and N2/H2 mixture. The results show that Tc improvement is greater after annealing in N2/H2 mixture.

2. Experiment

Our NbTiN thin films were deposited by reactive DC magnetron sputtering in a high vacuum chamber with base pressure down to 2.0 ×10−7 Pa. The target was an NbTi alloy with nominal compensation of 70% Nb and 30% Ti. The substrates were SiO2/Si with 500 nm thick SiO2 or Si(100) with high resistivity ( ). Before sputtering, the substrates were pre-cleaned sequentially in acetone, ethanol, and distilled water by ultrasonic and then dried and degassed in-situ to over 100 °C for about 12–16 h. After some pre-sputtering to clean the target surface, the films were then deposited in an atmosphere of Ar/N2 mixture of about 0.2–1 Pa with various powers. In addition, for investigating the influence of annealing, some films were annealed in N2 atmosphere at 200–1000 °C with a rapid annealing furnace of Accu Thermo AW410, or in mixed atmosphere of 85% N2 and 15% H2 at 100–450 °C with a homemade annealing furnace.

The transport properties of the thin films were measured by the standard four-probe method in a Quantum DesignTM PPMS system. The lattice structural properties were determined by an x-ray diffractometer (XRD Ultima IV, Mac Science Co., Japan) and surface morphologies were observed by an atomic force microscope (American Asylum Research, MFP-3d-SA standard atomic force microscopy).

3. Results and discussion

The deposition parameters, such as the atmosphere pressure, Ar/N2 mixture ratio, and sputtering power, can significantly influence the quality of the final films. To improve the superconducting transition temperature and surface smoothness, we systematically varied all these parameters in appropriate ranges and did transport, structural, and surface measurements for the corresponding films. Figures 1(a) and 1(b) show the Tc and surface roughness versus different sputtering pressures and powers, respectively. All these films were controlled with the thicknesses of about 300 nm.

Fig. 1. The dependence of superconducting transition temperature Tc and surface roughness on (a) the total Ar and N2 gas pressure and (b) the sputtering power during magnetron sputtering (film thickness 300 nm).

When the sputtering pressure is low, the mean free path of the atoms increases and the averaged energy of the atoms enhances when hitting the substrates. As a result, the diffusion rate of atoms on top of the films increases, leading to a smooth surface morphology. However, if the pressure is lowered, the stress in the films increases, leading to the increasing of defects and thus the drop of Tc. If the pressure is increased, the surface roughness increases too. This is because the deposited atoms do not have enough energy to move to the proper positions, as a result, form a rough surface. A smooth surface is particular important for some applications, for example, the fabrication of superconducting microelectronic devices such as SIS junction. Kohlstedt et al.[22] showed that the roughness of the film surface strongly influences the tunnel barrier formation and the electrical properties of the barrier. With the optimal sputtering pressure, fairly smooth surfaces with 0.15 nm roughness were obtained for our 300 nm films.

High sputtering power leads to high growth rate of the films, which avoids the impurity gas contamination, especially for active elements like Nb, Ti, and so on. As a result, increasing the sputtering power to an appropriate level usually improves the film quality. However, with the further increase of the growth rate, atoms will not have enough time to move to low energy positions before they are buried by newly arrived atoms. This will generate more defects in the films, thus decrease Tc and increase the surface roughness. Our data show this phenomenon (Fig. 1(b)).

Figure 2(a) shows some typical ρT curves for different thickness NbTiN films fabricated with optimized sputtering conditions (the sputtering power was decreased to 40 W for the ultrathin films when the thicknesses were lower than 30 nm). The superconducting transition temperature Tc and transition width (the temperature interval of 90% to 10% ) as the function of thickness from 5 nm to 300 nm are shown in Fig. 2(b). Consistent with the behavior of the Nb films,[23] Tc increased with the film thickness and was 14.2 K for 100 nm thickness films. A maximum Tc of 15.5 K was obtained for 300 nm thickness films. The Tc was 7.6 K for the thinnest 5 nm films and 9.6 K for the 10 nm films; to our knowledge these are fairly good results for ultrathin NbTiN films grown on Si substrates at room temperature without buffer layer to decrease the lattice mismatch. The superconducting transition widths of our NbTiN films were quite narrow, about 0.07 K for 100 nm films and 0.03 K for 300 nm films, which indicates the exceeding uniform of our NbTiN films. Figure 3 shows the AFM images for films with different thicknesses (under the same deposition conditions). From the surface morphology we could not identify clearly the grains until the thickness increased to 300 nm (with grain size about 14 nm). Such surface topography indicates the amorphous growth characteristics because of the ternary elements constitute of the NbTiN films. The ultralow surface roughnesses, 0.1 nm for 10 nm film and 0.18 nm for 300 nm films, also show the amorphous growth mechanism of the NbTiN films.

Fig. 2. (a) Some representative resistivity versus temperature for different thickness NbTiN films. (b) Superconductor transition temperature Tc and superconductor transition width (from 90% to 10% ) (inset) for different thickness films.
Fig. 3. The AFM topographic images of NbTiN thin films with thicknesses of (a) 10 nm, (b) 30 nm, (c) 60 nm, and (d) 300 nm.

To study the effect of thermal treatment on NbTiN thin films, the NbTiN films of different thickness (10–300 nm) were annealed by rapid thermal annealing (RTA) method with temperature ranging from 200 °C to 1000 °C. The annealing atmosphere was chosen as N2 or N2/H2 mixture (85% of N2 and 15% of H2) for comparison. In pure N2, the films were rapidly heated to the settled temperature within 30 s, and held at that temperature for 10 min, then decreased to 40 °C within 3 min. In N2/H2 mixture, the temperature increased to the annealing temperature within 180 s, held for 10 min, and then naturally cooled down to room temperature. Figure 4(a) shows the x-ray diffraction (XRD) results for the 300 nm films before and after RTA in N2 atmosphere. The [111] and [200] diffraction peaks were both present for the as-deposited and annealed films. Increasing the annealing temperature resulted in an increase in the ratio of the [200] and [111] x-ray diffraction line intensities (see Fig. 4(b)). These results indicate that as the annealing temperature increased, a change in texture from [111] to [100] took place, which supports Iosad et al. argument[24] that the thermalization of films during annealing causes the increase of the atomic energy, resulting in the preferred growth of [100] direction. The inset of Fig. 4(b) shows the proportional growth of grain size in [200] direction with increasing annealing temperature. The grain size increased from 14.5 nm for the as-deposited films to 22.7 nm after RTA at 1000 °C.

Fig. 4. (a) XRD patterns of 300 nm NbTiN films rapid thermal annealing in N2 atmosphere. (b) The ratio of [200] and [111] XRD peaks as the function of annealing temperature. The inset shows the grain size increasing with the RTA temperature.

Figure 5 shows the resistivity versus temperature curves of 100 nm films and the change percentage of Tc for different thickness films before and after RTA in N2 atmosphere. The Tc of the 100 nm films increased from 14.2 K to 14.8 K after 1000 °C thermal annealing. The improvement of Tc was more obviously for the relative thinner films, such as 30 nm and 60 nm films. After 1000 °C RTA, Tc increased about 12% for 30 nm, 60 nm thin films (11.3 K to 12.7 K for 30 nm films, 12.0 K to 13.5 K for 60 nm films), only 4.5% for 100 nm films. Along with the increase of Tc after thermal annealing, we found that the residual resistivity of the films increased with the increase of the RTA temperature. Such behavior contradicts with the thermal annealing effect of normal metals, where post-annealing increases the grain size and decreases the grain boundary scattering, as a result, leads to the decrease of the resistivity. We believe that this phenomenon may connect with the transport mechanism in nitride films. Chockalingam et al. proposed[25] that the Tc of NbN films is governed primarily by the carrier density rather than the disorder scattering. The Hall effects were measured for both as-deposited and 1000 °C RTA in N2 atmosphere 300 nm NbTiN films. The results showed that the carrier density decreased after RTA, from 1.74 ×1028 e/m3 for the as-deposited film to 1.34 ×1028 e/m3 for the 1000 °C annealed one. These results offered a reasonable explanation for the increase of the resistivity after RTA. However, these results could not give hints for Tc enhancement of NbTiN films after RTA. The increase of grain sizes after RTA may be the reason for the improvement of superconductivity properties. The discussion in Refs. [23] and [25] showed that the factor influencing the superconductivity transition temperature was complex for Nb or NbN films. For three-element superconductivity material NbTiN films, the intrinsic mechanism of superconductivity may be more complex, which needs further investigation for more details.

Fig. 5. (a) The resistivity versus temperature curves of 100 nm films before and after RTA. (b) The variation percent of Tc before and after RTA in N2 atmosphere for different thickness films.

For the thinner films of 10 nm, the situation changed. We found that Tc decreased after annealing at 450 °C in N2 atmosphere. When we changed the atmosphere to N2/H2 mixture, however, Tc kept increasing. These results are shown in Fig. 6(a). Furthermore, as a comparison, we annealed films of different thicknesses in these two kinds of atmospheres, and the results are shown in Fig. 6(b). It can be seen that Tc after RTA in N2/H2 mixture is generally higher than that in N2. As we know, the main difference of pure N2 and N2/H2 mixture is the introduction of deoxidizing hydrogen gas. Post-annealing in N2 atmosphere avoids the further oxidation and decreases the nitrogen defect in nitride films. However, there is usually a natural oxidized layer on the film surface after it is taken out from the sputtering chamber. Zhang et al.[26] showed that there is an about 1.3 nm thick oxidized layer of Nb2O5, TiO2, and NbO on top of NbTiN films. We believe that the high temperature thermal treatment induces the surface oxygen atoms diffusion into the grain boundary, as a result leads to the decrease of Tc even if the annealing is carried out in N2 atmosphere. While the films are treated in a reducing atmosphere, in our case, N2/H2 mixture, hydrogen atoms can take the oxygen atoms from the surface and prevent them diffuse into the films, resulting in Tc improvement. The ultrathin films were more sensitive to the influence of the surface oxidized layer because of the high ratio of the surface layer. The Tc of 10 nm ultrathin films increased from 9.6 K to 10.3 K after RTA in N2/H2 at 450 °C. These results can help us further understand the surface properties of ultrathin NbTiN films for improving the performance of NbTiN-based devices.

Fig. 6. (a) The relationship between superconductor transition temperature and annealing temperature for 10 nm NbTiN films in N2 or N2/H2 mixture. (b) The rise of Tc after RTA at 450 °C in two kinds of atmospheres.
4. Conclusion

High quality NbTiN films with high transition temperature, narrow transition width, and smooth surfaces were fabricated by optimizing the sputtering conditions including the pressure, N2/Ar mixture ratio, and sputtering power. We obtained Tc as high as 15.5 K and ΔTc as low as 0.03 K in a 300 nm NbTiN film on SiO2/Si substrate. We also obtained Tc of 7.6 K in a 5 nm film on intrinsic Si substrate without extra treatment and the growth of a buffer layer. Rapid thermal annealing after film deposition was carried out in N2 or N2/H2 mixture atmosphere. X-ray diffraction results showed that the grain size increased with the increase of the annealing temperature, and the film texture generally changed from [111] to [100] facet. The Tc increased evidently after RTA, but the residual resistivity also increased. As a comparison, we tried the thermal annealing treatment in both N2 and N2/H2 mixed atmospheres. The results showed that Tc improvement is higher after the treatment in N2/H2 mixture.

Reference
[1] Tanner M G Natarajan C M Pottapenjara V K O’Connor J A Warburton R J Hadfield R H Baek B Nam S Dorenbos S N Bermudez Urena E Zijlstra T Klapwijk T M Zwiller V 2010 Appl. Phys. Lett. 96 221109
[2] Schuck C Pernice W H P Tang H X 2013 Sci. Rep. 3 1893
[3] Zadeh I E Johannes W N Los Gourgues R B M Steinmetz V Dobrovolskiy S M Zwiller V Dorenbos S N 2017 APL Photon. 2 111301
[4] Jia X Q Kang L Yang X Z Wang Z H Ren T K Jin B B Chen J Xu W W Wu P H 2015 IEEE Trans. Appl. Supercond. 25 3
[5] Jiang L Shiba S Shimbo K Sakai N Yamakura T Sugimura M Ananthasubramanian P G Maezawa H Irimajiri Y Yamamoto S 2009 IEEE Trans. Appl. Supercond. 19 3
[6] Jiang L Shiba S Shiino T Shimbo K Sakai N Yamakura T Irimajiri Y Ananthasubramanian P G Maezawa H Yamamoto S 2010 Supercond. Sci. Technol. 23 4
[7] Shiba S Irimajiri Y Yamakura T Maezawa H Sekine N Hosako I Yamamoto S 2012 IEEE Trans. Appl. Supercond. 2 1
[8] Westig M P Selig S Jacobs K Klapwijk T M Honingh C E 2013 J. Appl. Phys. 114 124504
[9] Jackson B D Baryshev A M Lange G de Gao J R Shitov S V Iosad N N Klapwijk T M 2001 Appl. Phys. Lett. 79 436
[10] Akaike H Sakamoto S Munemoto K Fujimaki A 2016 IEEE Trans. Appl. Supercond. 26 5
[11] Bruno A Lange G de Asaad S Enden K L van der Langford N K DiCarlo L 2015 Appl. Phys. Lett. 106 182601
[12] Myoren H Shimizu H Iizuka T Takada S 2001 IEEE. Trans. Appl. Supercond. 11 3828
[13] Yu L Singh R K Liu H Wu S Y Hu R Durand D Bulman J Rowell J M Newman N 2005 IEEE Trans. Appl. Supercond. 15 44
[14] Makise K Terai H Takeda M Uzawa Y Wang Z 2011 IEEE Trans. Appl. Supercond. 21 139
[15] Yang X Y You L X Zhang L Lv C L Li H Liu X Y Zhou H Zhen Wang 2018 IEEE Trans. Appl. Supercond. 28 1
[16] Shiino T Shiba S Sakai N Yamakura T Jiang L Uzawa Y Maezawa H Yamamoto S 2010 Supercond. Sci. Technol. 23 045004
[17] Loudkov D Tong C Y E Blundell R Megerian K G Stern J A 2005 IEEE Trans. Appl. Supercond. 15 476
[18] Zhang L Peng W You L X Wang Z 2015 Appl. Phys. Lett. 107 122603
[19] Guziewicz M Laszcz A Domagala J Z Golaszewska K Ratajczak J Kruszka R Juchniewicz M Czerwinski A Slysz W 2013 Proc. SPIE 8902 89022S
[20] Zhang X Song X H Zhang D L 2010 Chin. Phys. B 19 086802
[21] Hatano M Nishino T Kawabe U 1988 J. Vac. Sci. Technol. 6 2381
[22] Kohlstedt H Konig F Henne P Thyssen N Caputob P 1996 J. Appl. Phys. 80 9
[23] Song X H Jin Y R Fan Z J Mi Z Y Zhang D L 2015 Chin. Phys. Lett. 32 047403
[24] Iosad N N Pers N M van der Grachev S Roddatis V V Jackson B D Polyakov S N Dmitriev P N Klapwijk T M 2002 J. Appl. Phys. 92 9
[25] Chockalingam S P Chand M Jesudasan J Tripathi V Raychaudhuri P 2008 Phys. Rev. B 77 214503
[26] Zhang L You L X Ying L L Peng W Wang Z 2018 Physica C 545 1