Study on the stability of nano-polycrystalline Ag films in thermal and humid environment

Abstract

This study systematically investigates the degradation behavior of nanopolycrystalline silver films (npAgF) in a thermal and humid environment, focusing on analyzing the impact of specific environmental factors on their performance evolution, especially the formation pathways of degradation products and their impact on optical properties. NpAgF samples were prepared by ion-assisted deposition and electron beam evaporation, and a six-month environmental test was conducted under controlled temperature and humidity. During the process, changes in the chemical state, surface morphology, reflectivity, and residual stress of the film were continuously monitored. The results show that the adsorption of moisture and corrosive gases on the silver film changes its interfacial properties, promotes the migration of Ag ions, and leads to the formation and growth of degradation products. By combining the characterization results of the degradation products with the fitting analysis of the optical constants, the evolution path of the silver film during degradation can be better simulated. This work reveals the degradation mechanism of npAgF under thermal and humid conditions from an experimental perspective and provides a basis for improving their long-term stability in the atmospheric environment.

Introduction

Silver (Ag) films are indispensable for high-performance aerospace optical systems, such as satellite telescopes, optical communications, and remote sensing, due to their exceptional broadband reflectivity and low thermal emissivity^1–4. Although their on-orbit performance is excellent, silver films are easily affected by thermal and humid conditions during ground assembly and calibration, which may degrade their performance or even make them unusable.

To address the degradation of silver films, researchers have proposed an effective method, which is to deposit nano-metal oxide films such as ZnO, ZrO₂, and TiO₂ on the surface of silver nanostructures to block the intrusion of moisture and corrosive gases^5,6. In fact, the traditional coating technology is difficult to form a large-area and completely dense protection. Even when covered with a protective film, the silver surface may still contain defects such as pinholes, particles, or pores, and these localized weaknesses can spread over time and may eventually damage the entire silver layer^4,7,8. This inherent drawback has prompted extensive research on the degradation of silver films under specific conditions, such as low‑energy ion irradiation^9,10, ultraviolet irradiation^11–13, and water vapor saturation^4,7,14–17. While these works offer valuable insights into silver protection, the effectiveness of protecting optical‑grade nanopolycrystalline silver films (npAgF) ultimately depends on a thorough understanding of their degradation pathways, which is still incomplete. This gap may fundamentally limit the development of silver film maintenance strategies^16,18–20.

To study the degradation pathway of silver films in the atmospheric environment, we established an experimental framework. The degradation process of npAgF under the combined conditions of high temperature and humidity was systematically analyzed, revealing the failure mechanism driven by low-temperature annealing and high-humidity exposure. First, npAgF with a thickness of 130 nm were prepared using a combination of ion-assisted deposition (IAD) and electron beam evaporation. Then, npAgF were placed in a thermal and humid (TAH) environment (temperature: 50 °C, relative humidity: 95%). During the six‑month experiment, the chemical state^19,21–25, structure, stress^18,26,27, and optical properties of npAgF were measured and recorded every two months.

Experimental results show that composite absorption, grain coarsening, and stress reversal are the main reasons for the degradation of silver films under atmospheric conditions. Based on optical methods, the degradation mechanisms described above were integrated into a “five-phase” model. The model links the degradation process to the main corrosion product (Ag₂S) and thus reconstructs each corrosion stage accurately. This method provides a theoretical basis for the future transition from empirical conservation methods to predictive conservation strategies.

Experimental details

Preparation

In this experiment, a Leybold ARES2500 high-precision coating system was used to prepare npAgF. The system is equipped with a large vacuum chamber and a planetary disk structure and can complete the coating at room temperature by combining electron-beam evaporation and IAD processes.

During evaporation, a high-power electron beam (beam current 140 ± 10 mA) was used to vaporize high-purity silver (99.99%). At the same time, a radio-frequency ion source (grid voltage 1500 V, beam current 1000 mA) was used to supply argon gas for IAD and to improve the film density. Silver was deposited on silicon (111) wafers, K9 glass, and fused-quartz substrates at a rate of about 10 Å/s. The film thickness and deposition rate were monitored in real time during deposition using a quartz crystal sensor.

The prepared npAgF samples were placed in a Hitachi EC‑16MHH test chamber, and environmental tests were conducted at 50 ± 2 °C and 95 ± 3% relative humidity. The chemical state, surface morphology, reflectivity, and residual stress of the films were measured every two months. In order to meet the optical and surface‑structure testing requirements, the surface flatness of the silicon wafers was processed to λ/10 at a wavelength of 632.8 nm, and the polishing grade of the quartz substrates was not lower than level II. Table 1 summarizes the detailed processing parameters and corresponding characterization methods for the different types of substrates.

Table 1.

Substrate material parameters and applications.

Sample	Substrate	Dimensions (mm)	Surface flatness (λ)	Surface quality*	Experimental parameters
S1	Si (111)	5 × 10 × 0.5	\	60/40	XRD/XPS/Ellipsometry
S2	Si (111)	Ø150 × 0.625	1/10	60/40	Stress/ TAH
S3	JGS-1	Ø30 × 2	\	40/20	AFM/ Raman
S4	K9	40 × 40 × 1	\	60/40	Reflectivity/ TAH

*ISO 10110:2010.

The X-ray diffraction (XRD) pattern of the sample was recorded by measuring the diffraction intensity as a function of the 2θ angle in the range of 20°–70°. From the XRD pattern of the 130‑nm‑thick silver film (Fig. 1), it can be seen that the silver film prepared by IAD electron‑beam evaporation has a polycrystalline structure and shows a strong preferred orientation on the (111) plane.

Fig. 1.

XRD pattern of sample S1 deposited by electron-beam evaporation with IAD.

Testing

Chemical state

The Raman spectra of the samples were collected at room temperature using a Renishaw inVia laser confocal Raman spectrometer. The instrument is equipped with a 532 nm Ar⁺ laser light source and a CCD detector. During the test, the grating was 1800 lines/mm, the exposure time was 6 s, and the total integration was 10 times.

A Thermo Scientific K‑Alpha X‑ray photoelectron spectroscopy (XPS) system was used to analyze the chemical state of the film surface. During the experiment, the sample was fixed on the sample stage in the analysis chamber. When the vacuum was better than 2.0 × 10⁻⁷ mbar, it was transferred to the analysis chamber for testing. The specific test parameters were as follows: spot size 400 μm, operating voltage 12 kV, and filament current 6 mA. The pass energy for full‑spectrum scanning was 150 eV with a step size of 1 eV, and the pass energy for narrow‑spectrum scanning was 50 eV with a step size of 0.1 eV²⁸.

Morphology

At room temperature, an atomic force microscope (NT-MDT NTEGRA) was used in tapping mode to scan the sample surface of over a 5 μm × 5 μm area with a contact force of 10⁻⁹ N, and the surface morphology of the film was recorded.

The images were processed and analyzed using Gwyddion and ImageAnalysis software, which provided particle-size distribution and surface-roughness data for the npAgF.

Optical constant

The reflectance of the npAgF was measured in the wavelength range of 200–1000 nm using a double-beam spectrophotometer (LAMBDA 1050, Perkin Elmer) at an incident angle of 6°.

Polarization data for light of different wavelengths were measured using a spectroscopic ellipsometer (J.A. Woollam Co.). The thickness and optical constants of the degraded npAgF were fitted using spectroscopic ellipsometry (SE) software.

Stress

To monitor the evolution of residual stress in npAgF under the TAH environment, we performed stress measurements on the samples at each experimental stage using a kSA‑MOS Ultra‑Scan film‑stress imager (k‑Space Associates).

Results and discussion

Chemical states

At room temperature, the Raman spectrum of sample S3 was measured at each stage using an argon-ion laser operating at 10% power (10 mW). The characteristic spectral peaks were normalized, as shown in Fig. 2.

Fig. 2.

Raman spectra of sample S3 at different stages in the TAH environment.

No Raman signal was detected in the as-deposited npAgF. After 2 months of the TAH environment exposure, weak and broad Raman peaks emerged. After 4 months of exposure, significant intensity changes and characteristic peaks appeared at ~ 93 cm⁻¹, ~ 219 cm⁻¹, and ~ 490 cm⁻¹. The ~ 93 cm⁻¹ peak, corresponding to Ag lattice vibrations, matches the Raman signature of acanthite (α-Ag₂S)^29–33. The prominent 219 cm⁻¹ peak represents Ag–O stretching vibrations, with its narrow half-width being characteristic of Ag₂O. Similarly, the broad 490 cm⁻¹ peak, also assigned to Ag–O vibrations, indicates AgO formation in the degraded npAgF surface layer^24,31,34,35.

Raman spectra confirm that the degradation products of npAgF in the TAH environment comprise Ag₂S, Ag₂O, and trace AgO. The observed gradual increase in intensity of the Raman peaks further indicates the progressive accumulation of these silver compounds.

To further analyze the chemical states of the degradation products on the npAgF surface and of Ag, a Thermo Scientific K-Alpha XPS system was used to assess the effects of the TAH environment, with Al Kα radiation (hν = 1486.6 eV) as the excitation source.

In Fig. 3, the XPS full spectrum of the freshly prepared npAgF shows only adsorbed carbon and oxygen, with no other chemical elements detected. The Ag 3d fine spectrum peak is asymmetric, showing a clear energy loss characteristic at the high binding energy side. The spin-orbit separation is 6.0 eV, indicating that the XPS spectrum corresponds to pure Ag.

Fig. 3.

Ag 3d5 XPS full spectrum and fine spectrum of sample S1 (as-deposited).

After 6 months of exposure to the TAH environment, high-resolution XPS spectra of Ag 3d, S 2p, O 1s, and Cl 2p were acquired from the film surface. Figure 4a and e show the Ag 3d and Ag Auger spectra, respectively, with dotted lines indicating the binding-energy (BE) positions for pure Ag 3d at 368.3 eV and for the Auger peak at 528.2 eV³⁶. The anion charge and lattice potential of the compound cause a slight shift in the BE of Ag 3d toward lower energy, broadening the half-peak width. The Auger spectrum also shows noticeable changes in peak shape and kinetic energy, with a wider half-peak width and a more pronounced shift toward lower energy. The Auger kinetic energy of Ag 3d is around 520 eV, with the Auger parameter α’ of approximately 725 eV (Table 2)^24,37, consistent with the chemical state of Ag sulfide. In Fig. 4b, two doublets at the lower and higher BE of S 2p are observed in all samples. The dotted lines mark the BE positions of Ag₂SOₓ (~ 168.6 eV) and Ag₂S (~ 161 eV). The lower BE of S 2p, ranging from 161.07 to 161.45 eV (Table 2), is attributed to Ag₂S. The higher BE peaks correspond to sulfuric acid (SO₄²⁻) or sulfite (SO₃²⁻), suggesting the coexistence of Ag₂S and Ag sulfates on the Ag surface³¹. The BE of the O 1s fine spectrum in Fig. 4c is around 532 eV, with a broadened half‑peak width. This suggests that not all regions of the O 1s spectrum originate from Ag sulfate. The presence of AgO or Ag₂O on the npAgF is also likely, with BE positions at approximately 528.6 eV for AgO and 529.5 eV for Ag₂O. Based on the BE position of oxygen (Table 2), the oxide in the npAgF was identified as Ag₂O. The dotted line in Fig. 4d shows the BE position of Ag chloride, the chlorine BE peak on the npAgF (Table 2) appears near 197.8 eV, clearly indicating the presence of AgCl.

Fig. 4.

The XPS fine spectra of Ag, S, O, and Cl elements of sample S1.

Table 2.

XPS data of atomic content, binding energy, and Auger parameters of sample S1.

Exposure	Atomic concentration (%)				Binding energy (eV)				Ag MNN	Auger paramter
(Month)	Ag	O	S	Cl	Ag 3d5/2	O 1s	S 2p3	Cl 2p3	Kinetic energy	Binding energy
0	100.00	–	–	–	368.30	–	–	–	358.20	726.50
2	61.60	23. 90	10.94	3.56	368.10	532.37	161.33	197.89	357.54	725.53
4	54.45	25.99	15.30	4.25	367.99	532.33	161.07	197.74	358.43	726.42
6	38.64	50.69	8.00	2.66	367.97	532.23	161.25	197.89	357.04	725.01

*.

In summary, peak fitting was applied to each element, and a semi-quantitative analysis of the main chemical elements in both the full and fine spectra was performed. Figure 5 shows the evolution of surface elemental composition on the npAgF surface during the experiment. The proportion of elemental Ag gradually decreases, a trend consistent with the reduction in the film’s reflectance in the visible and near-infrared regions. Meanwhile, the sulfur and chlorine content increase from the preparation stage to the four-month TAH test, after which the rate of change slows as the surface oxygen content increases (Fig. 5c).

Fig. 5.

Chemical element content on sample S3 at different stages in the TAH environment.

Figure 6 illustrates the changes in the surface content of Ag compounds after elemental self‑consistency correction. The trends of the silver compounds are in excellent agreement with the evolution of elemental composition shown in Fig. 5. By comparing Figs. 5c and 6b, we observe that although the total oxygen content increases linearly with exposure time, the concentration of silver oxide (mainly Ag₂O) shows a non-monotonic change, first increasing and then decreasing. Further analysis shows that this fluctuation is closely related to the evolution of the surface microstructure.

Fig. 6.

Content of compounds on sample S3 at different stages of the TAH environment.

In the TAH environment, the water layer on the npAgF surface as a medium for absorbing corrosive gases from the atmosphere, leading to the dissolution of solid Ag and the formation of Ag ions. Given the negligible content of sulfate particles in the experimental chamber, sulfur mainly comes from gaseous reduced sulfur species, whereas chlorine is derived from water vapor. Regarding the oxidation state of Ag, although some studies suggest that Ag oxide formation is unlikely under normal temperature and pressure^38,39, it is indeed detected in the XPS measurements. Other studies indicate that O₃ in air can lead to Ag oxidation, which explains the observed presence of Ag oxide⁴⁰.

Surface morphology

The surface morphology of sample S3 was analyzed using an atomic force microscope (AFM). AFM images of the npAgF (5 × 5 μm scan area) at different stages are shown in Fig. 7.

Fig. 7.

Surface morphologies of sample S3 at different stages in the TAH environment. (a) AFM image of sample S3 as-deposited; (b) AFM image of sample S3 after 2 months in the TAH environment; (c) AFM image of sample S3 after 4 months in the TAH environment; (d) AFM image of sample S3 after 6 Months in the TAH environment.

The mixture of Ag₂S and sulfate, uniformly distributed across the npAgF surface, exhibited prominent “spikes” in the TAH environment (Fig. 7d), consistent with the behavior of sulfide films on metal surfaces¹⁵. These sulfide formations “grow” vertically, and their morphological changes correspond to the trend in sulfur content observed in the XPS spectra (Table 2). The sulfide on the npAgF surface does not form a continuous layer during the tests. Initially, the surface roughness of the substrate increases rapidly and then tends to stabilize, while the “peak” height of the sulfide continues to grow. Throughout the experiment, the degradation of the Ag film was relatively uniform, giving the surface a rough and dull appearance under natural light.

After 6 months in the TAH environment, the average particle size on the npAgF surface increased from approximately 48.05 to 59.37 nm, representing a 24% increase. The root-mean-square surface roughness increased from approximately 1.60 to 10.96 nm, representing a sixfold increase. In the early stage of the test, no elements other than silver were detected in the XPS spectra, indicating that low-temperature annealing in the TAH environment effectively reduced the defect density in the film⁴¹. As water molecules continue to adsorb onto the surface, the migration of silver ions along the grain boundaries is significantly accelerated, which further drives the dissolution of small silver particles and the selective growth of larger particles, ultimately leading to the aggregation of silver atoms and an increase in grain size. After two months of exposure, compounds such as Ag₂S and Ag₂O were detected in the XPS spectra of npAgF, confirming that corrosion products gradually formed on the film surface. Although the average particle size did not change significantly throughout the experiment, the particle size distribution frequency shown in Fig. 8 indicates that the number of particles exceeding the average size continues to increase, reflecting the dynamic characteristics of grain growth evolution.

Fig. 8.

Particle size frequency distribution diagram of sample S3.

The mean roughness and root-mean-square roughness (S_a, S_q) of npAgF can be calculated using Eqs. (1) and (2) equations⁴²:

1

2

where M is the number of columns on the surface, and N is the number of lines.

Figure 9 shows that the particle size on the Ag-film surface gradually increased throughout the test. The frequency of the average particle size initially increased, then slightly decreased, and later increased again. The proportion of Ag oxide in the npAgF XPS data, as presented in Table 2, correlates with the frequency distribution of particle size. Therefore, the observed changes in average particle size are attributed to the formation and decomposition of Ag oxide in the TAH environment (Fig. 6b)^24,43.

Fig. 9.

Variation in particle size and RMS (S3) in the TAH environment.

Optical properties and stress analysis

Optical reflectance

Figure 10 shows the reflection spectra of sample S4 at various stages of the TAH environment, measured by a spectrophotometer. As shown in the figure, the reflectance changes slowly at first, then decreases rapidly after 2 months, and the rate of decline slows after 4 months. Throughout the process, the reflectance in the ultraviolet and visible bands changes significantly, whereas the reflectance in the near-infrared band (wavelengths greater than 800 nm) changes only very slowly.

Fig. 10.

Reflectance (S4) at different stages of the TAH environment.

Figure 11 shows the changes in the reflectance of npAgF at a wavelength of 400 nm in the TAH environment. After 4 months, the reflectance decreased by approximately 23%. However, in the following 2 months, it decreased by only about 5%, indicating that the degradation rate slowed down. This trend was consistent with the changes observed in the XPS spectra of npAgF. After 4 months, the Ag compounds on the film surface reached an almost stable state. As degradation and corrosion progressed, reactive elements on the film surface, such as Ag, sulfur, and oxygen were depleted. This reduced availability slowed further degradation and corrosion, thereby slowing the decline in reflectance.

Fig. 11.

Variation in reflectance changes (S4) at different stages of the TAH environment.

Optical constant

The surface of npAgF in the TAH environment mainly consists of naturally formed Ag sulfide, with trace amounts of contaminants. If the optical constants of natural Ag sulfide are known, they can be used to analyze the rough layer to determine its formation time and estimate its thickness. The optical constants of the rough surface layer are determined using variable-angle spectroscopic ellipsometry (SE). The ellipsometric parameters ψ and Δ of Si-based Ag film samples were measured at three incidence angles (55°, 60°, and 65°) over the wavelength range of 207–1000 nm, and then fitted using SE.

The SE data were analyzed using a five-phase model that includes a Si substrate, a SiO₂ thin layer, an Ag layer, an Ag₂S layer, and a surface rough layer of Ag₂S. A rough layer was applied to the surface to obtain a refractive index and extinction coefficient that more closely represent the actual conditions. The roughness layer was characterized using a Bruggeman effective medium approximation (EMA) model, comprising 50% Ag₂S and 50% voids, as shown in Fig. 12.

Fig. 12.

Schematic diagram of the five-phase Model.

The Tauc-Lorentz (T-L) dispersion model, commonly used for dielectric films, is used to describe the dispersion relationship of the Ag₂S thin layer on the npAgF^44,45.

The real part (ε₁) and the imaginary part (ε₂) of the dielectric function in the T‑L model are given as follows.

3

4

Here, E represents the photon energy, A is the amplitude, E₀ is the peak transition energy, C is the broadening term, Eₛ is the optical band‑edge parameter, E_g is the optical band gap, and ε₁(∞) is the high‑frequency dielectric constant, all expressed in units of energy⁴⁶.

The unknown parameters include the Ag₂S film thickness (dse), the thickness of the Ag₂S roughness layer, and the dielectric constant, which is treated as a fitting variable.

As shown in Fig. 13, the degradation products formed on the bare npAgF were approximated in the model by an Ag₂S layer. The optical constants of the Ag sulfide layer (0.4–1.0 μm) are similar to those reported in the literature^21,47.

Fig. 13.

Refractive index and extinction coefficient (S1) of the Ag₂S layer for sample S1.

In the visible range, a larger extinction coefficient (k) and a lower refractive index (n) enhance the surface absorption of the npAgF. The disturbance in the UV range (0.24–0.4 μm) caused by the trapping effect of the rough layer increases the uncertainty of the product nk in this region. The obtained n and k values were used to fit the thickness of the Ag₂S layer, and the results indicated a continuous increase in the Ag₂S layer thickness at various stages, from an initial 1.57 nm to approximately 12.40 nm. Figure 14 shows the fitting results of the five-phase model for the S1 after 6 months.

Fig. 14.

Experimental and fitted SE data (S1) after 6 Months in the TAH environment.

Stress analysis

For sample S2 exposed to the TAH environment, surface-shape parameters were recorded throughout the experiment using the kSA-MOS UltraScan system.

Changes in the local curvature of the substrate surface and its physical parameters were used with the Stoney formula (5) to calculate the film-stress characteristics of npAgF at each stage.

5

Where E_s is the elastic modulus of the substrate (MPa), v_s is the Poisson’s ratio of the substrate, T_s is the substrate thickness (µm), T_f is the film thickness (µm), R₀ is the initial radius of curvature of the substrate (µm), and R_(s+f) is the final radius of curvature (µm) of the substrate plus the thin film⁴⁸.

The resulting residual-stress values are mapped to the x and y coordinates on the wafer to obtain the surface residual stress σ, as illustrated in Fig. 15.

Fig. 15.

Changes in minimum residual stress (sample S2) in the TAH environment. (a) Surface stress distribution of S2 as-deposited; (b) Surface stress distribution of S2 after 2 Months in the TAH environment; (c) Surface stress distribution of S2 after 4 Months in the TAH environment; (d) Surface stress distribution of S2 after 6 Months in the TAH environment.

Table 3 shows the results of the residual stress measurements for the S2 samples. The initial stress of the as-deposited npAgF sample is tensile, with an average value of + 155.10 MPa. During exposure to the TAH environment, the stress in the film layer transitions from tensile to compressive. After 2 months, the average stress of the film decreased to + 22.73 MPa, the lowest value recorded. After six months, the average stress reached − 364.65 MPa, and the reliability of the npAgF no longer met the standard usage requirements.

Table 3.

Residual stress (S2) at different stages in the TAH environment.

Exposure	Residual stress (MPa)
Time (month)	Max value	Min value	Mean value	Difference value
0	+ 213.25	+96.95	+ 155.10	116.30
2	−4.95	+50.40	+22.73	55.35
4	−172.00	−136.40	−154.20	35.60
6	−415.45	−313.85	−364.65	116.30

As shown in Fig. 16, the total stress initially decreased and then increased over time during the experiment, with the rate of change gradually accelerating. The stress induced by H₂O adsorption increased continuously⁴⁹, while the internal stress was gradually relieved. The results show that during the transition from tension to compression, the total stress first decreases and then increases in the opposite direction, and the overall change is nearly linear.

Fig. 16.

Variation of total residual stress (S2) at different stages of the TAH.

After six months, based on the XPS and AFM results at each stage, it can be concluded that the formation of silver sulfide compounds leads to a reconstruction of the film structure. Additionally, the compressive stress caused by the volume change of the silver compounds further accelerates stress release, while the residual stress keeps increasing, resulting in a relatively large overall compression stress in the film.

Conclusion

In the thermal and humid (TAH) environment, the water layer adsorbed on the surface of the nanopolycrystalline silver film (npAgF) serves as a medium for dissolving solid silver and absorbing corrosive gases. The low-temperature annealing effect reduces the defect density inside the film and promotes the formation of silver ion clusters, while the water layer further enhances the migration and aggregation of these ions. The corrosive gas dissolved in the water layer alters the chemical state of the npAgF, generating corrosion products mainly composed of Ag₂S, Ag₂O, and AgCl with a small amount of Ag₂SO₃. The formation of silver ion clusters and silver oxide on the surface of npAgF promotes a gradual increase in grain size, while the change in surface roughness is primarily attributed to the growth of spiked Ag₂S. Research indicates that increased surface roughness is not the only factor contributing to the decrease in reflectivity; the oxidation of npAgF causes photons to lose more energy when exciting free electrons, which is the primary mechanism behind the reduction in reflectivity. In addition, the stress induced by H₂O adsorption and the volume change of the corrosion products together promote the increase in the residual stress of npAgF. Our work systematically elucidates the corrosion mechanism of npAgF in the TAH environment, providing an important foundation for enhancing their environmental stability. Future research could develop targeted protection methods based on the degradation patterns of npAgF to broaden its application range.

Author contributions

Y.P.L . : Design of methodology, creation of models; Writing Original Draft, Review , Editing.S.G.W. : Provision of study materials, reagents, materials, laboratory samples, other analysis tools.X.X.S. : Formal analysis, Application of statistical, mathematical, computational.T.F.Z. : Investigation, conducting a research and investigation process.W.B.D . : Project administration.J.L. : Oversight and leadership responsibility for the research activity planning and execution.All authors reviewed the manuscript.

Data availability

Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request. Email: yaopengli@mail.sitp.ac.cn.

Declarations

Competing interests

The authors declare no competing interests.