Selective doping of Ni2+ in octahedral sites provided by nanocrystals embedded in glass-ceramics (GCs) is crucial to the enhancement of broadband near-infrared (NIR) emission. In this work, a NIR emission with a full-width-at-half-maximum (FWHM) of 288 nm is first reported from ZnGa2O4: Ni2+ nano-spinels embedded GCs with excellent transparency. A comparison is made of the NIR luminescence properties of Ni2+ doped GCs containing ZnGa2O4, germanium-substituted ZnGa2O4 nano-spinels (Zn1+x Ga2−2x Ge x O4), and Zn2GeO4/Li2Ge4O9 composite nanocrystals that are free of Ga3+. The results show that ZnGa2O4: Ni2+ GCs exhibit a significantly enhanced NIR emission. The incorporation of the nucleating agent TiO2 is favored in terms of the increased luminescence intensity and prolonged lifetime. The possible causes for the enhancement effect are identified from the crystal structure/defects viewpoint. The newly developed GCs incorporate good reproducibility to allow for a tolerance of thermal treatment temperature and hence hold great potential of fiberization via the recently proposed “melt-in-tube” method. They can be considered as promising candidates for broadband fiber amplifiers.
Broadband tunable near-infrared (NIR) light sources are extremely useful for a wide range of applications in optical communications, photochemistry, spectroscopy and pump-probe experiments etc.1. The ability of transition metal (TM) ions (e.g., Ti3+, V3+, Cr3+, Cr4+, Ni2+ etc.) doped bulk crystals and glass-ceramics (GCs) to lase in the broadly tunable NIR (1000~1700 nm) wavelength region have attracted intense attention over the last two decades2,3,4,5,6. Because nickel is very stable in its divalent oxidation state (Ni2+), Ni2+ doped phosphors can be readily synthesized in ordinary lab conditions without strict valence control7. However, Ni2+ cannot lase in glasses because Ni2+ is generally five-fold (trigonal bipyramidally, 5Ni2+) and/or four-fold (tetrahedrally, 4Ni2+) coordinated in glassy networks8, whereas only six-fold octahedrally coordinated 6Ni2+ exhibits luminescence in the NIR wavelength range. Even when Ni2+ adopts an octahedral coordination in fluoride glasses9, the strong electron-phonon coupling of 6Ni2+ in amorphous materials severely limits the radiative quantum efficiency. These problems can be resolved provided GCs containing nanocrystals that provide octahedral sites for Ni2+ can be produced, which is very possible through ingenious composition design and well controlled crystallization10. Selective doping of 6Ni2+ in ZnAl2O4 11, LiGa5O8 12,13,14, Ga2O3 15 and most recently ZnF2 and KZnF3 2, 17 embedded transparent GCs and even glass-ceramic (GC) optical fibers have been produced16. Promising results such as the ligand-field driven wavelength tunable and broadband NIR emission of Ni2+ have also been observed5, 17.
The pursuit of Ni2+ doped GCs is being driven continuously by newly invented crystals with excellent luminescence properties, for example, the germanium-substituted ZnGa2O4 spinel of the general formula Zn1+x Ga2−2x Ge x O4 (0 ≤ x ≤ 1) and this has attracted significant attention due to the unprecedented persistent luminescence observed in Zn1+x Ga2−2x Ge x O4:Cr3+ 18. Such nanocrystals can simultaneously provide tetrahedral (occupied by 4Zn2+ and 4Ge4+) and octahedral (occupied by 6Ga3+) sites for Mn2+, Co2+, Ni2+, Cr3+ and Mn4+ etc., and thus can be used as a multi-functional platform for diverse applications in lighting, display, telecommunication and bio-imaging etc.19,20,21. Additionally, transparent GCs containing Zn1+x Ga2−2x Ge x O4 nanocrystals have been recently fabricated21, 22. Selective doping of Ni2+ in Zn1+x Ga2−2x Ge x O4 nanocrystals embedded in GCs has been reported; however, the NIR luminescence properties were not clearly identified in this case22. To date the authors of this article are not aware of any study relating to the luminescence properties of ZnGa2O4: Ni2+ GCs, although GCs containing ZnGa2O4: Cr3+ nanocrystals, showing persistent luminescence and temperature sensing properties, have recently been extensively studied12, 23.
In the work reported in this article a detailed study has been undertaken and comparison made of the NIR luminescence properties of Ni2+ doped GCs containing ZnGa2O4 and germanium-substituted ZnGa2O4 (Zn1+x Ga2−2x Ge x O4) nano-spinels. To underline the important role of 6Ga3+, other Ni2+ doped GCs containing Zn2GeO4/Li2Ge4O9 composite nanocrystals that are free of Ga3+ were also prepared for comparison. The synthesized ZnGa2O4: Ni2+ GCs are highly reproducible which allows for a tolerance of thermal treatment temperature, and thus are perfectly matched for the recently proposed “melt-in-tube” method. For functional GC fibers cannot be obtained using the conventional “rod-in-tube” method16, 24,25,26, the ‘melt-in-tube method has recently facilitated fabrication of GC fibers doped with Ni2+ 16, Bi24, Cr3+ 25, or quantum dots26 which have exhibited excellent optical quality. The study described in this article is expected not only provide a candidate fiber amplifier material, but also advance understanding of the correlation between the structure of the nanocrystals and NIR luminescence of Ni2+, and thus will provide useful guidance for designing novel Ni2+ doped GCs with enhanced luminescence properties such as ultra-broadband tunable NIR emission27. Moreover, the present work may also advance the understanding of the mechanism underlying the persistence luminescence of Cr3+ doped spinels which is currently still open to question18.
Three different types of the nominal composition (in mol. %) of Ni2+-doped glasses and GCs were prepared using high purity (4N) raw materials of SiO2, GeO2, Ga2O3, ZnO, Na2CO3, K2CO3, Li2O, ZrO2, TiO2 and NiO.
51SiO 2 -18Ga 2 O 3 -18ZnO-6Na 2 O-4ZrO 2 -3TiO 2 - x NiO (x = 0, 0.15, 0.3, 0.5) was chosen to generate the ZnGa2O4 nanocrystal, hereafter, this group of glasses and glass-ceramics are denoted as ZGO-xPG and ZGO-xGC, respectively. Glasses were melted in a platinum crucible at 1600 °C for 2 h, quenched onto a cold brass plate and then annealed at 500 °C for 3 h. GCs were fabricated by heating the annealed glasses at 680 °C for 12 h followed by a further heating at 780~800 °C for 12 h, refer to Tanaka et al.28;
62GeO 2 -20ZnO-10Ga 2 O 3 -5K 2 CO 3 -3TiO 2 - x NiO (x = 0, 0.15, 0.3 and 0.5) was chosen to generate Zn1+x Ga2−2x Ge x O4 nanocrystal, hereafter, this group of glasses and glass-ceramics are denoted as ZGGO-xPG and ZGGO-xGC, respectively. Glasses were preheated at 850 °C for 30 min and then melted in an alumina crucible at 1400 °C for 30 min. The glasses were then annealed at 500 °C for 3 h. GC was made via a heat-treatment at 650 °C for 2 h, as described in ref. 21;
70GeO 2 -15ZnO-15Li 2 O-0.15NiO was chosen to generate Zn2GeO4 and Li2Ge4O9 nanocrystal, hereafter, this group of glasses and glass-ceramics are denoted as ZLGO-0.15PG and ZLGO-0.15GC, respectively. The glasses were melted in a platinum crucible at 1300 °C for 30 min and annealed at 450 °C for 2 h. GCs were fabricated by heating the annealed glasses at 545 °C for 2 h, similar to the process reported in ref. 29.
Transmission spectra were measured using a Perkin-Elmer Lambda 950 UV-VIS spectrophotometer in the spectral range of 200–1800 nm. Refractive indices were measured using an Abbe refractometer AR2008 (KRÜSS, Germany). Photoluminescence (PL) spectra were recorded using a Fluorolog-3-P UV-vis-NIR fluorescence spectrophotometer (JobinYvon, Longjumeau, French). The decay curves were measured using a FLS920 Fluorescence spectrometer (Edinburgh Instruments) from room temperature (300 K) down to liquid helium temperature (10 K). The samples used for the PL measurement were plane-parallel well polished plates with the identical dimension of ~10 × 10 mm2 and thickness of 1 mm. The fluorescence was collected in the direction perpendicular to the direction of the pump beam, and the pump light was focused (to a spot of diameter ~4 mm) using a lens and incident at a 45° angle to the normal of the front surface of the sample. In the experiment, both the power of the pump light and the configuration of the light path were kept the same. Because only very thin samples were used for the measurement, reabsorption is not significant and any effects due to this can be omitted in the present study, in accordance with the work of Loiko30.
X-ray diffraction (XRD) patterns of all the samples were recorded under the same measurement conditions using an X-ray diffractometer (D/MAX 2550VB/PC, Rigaku Corproation, Japan) with Cu-Kα irradiation. The microstructure of the crystallized glasses was studied using a JEM-2100 high-resolution transmission electron microscope (HRTEM). Raman spectra were measured by RenishawInvia Raman microscope (Renishaw, Gloucestershire, UK) with an excitation wavelength of 515 nm.
Results and Discussion
From the XRD patterns of the crystallized glasses, the precipitation of ZnGa2O4 (Fig. 1(a)), and Zn1+x Ga2−2x Ge x O4 nano-spinels (Fig. S1(a), supporting information), as well as Zn2GeO4/Li2Ge4O9 composite phases (Fig. S2 in the supporting information) can be discerned in accordance to the literature21, 28, 29. The formation of the target nanocrystals were also confirmed from the Raman spectra where the crystallized glasses show sharp scattering peaks well match those of the standard polycrystals (Fig. 1(b)). According to the work of Zhuang et al.21, it is very difficult to determine unambiguously the exact Zn1+x Ga2−2x Ge x O4 phase in GCs, owing to the undistinguishable XRD patterns between the two end-members, ZnGa2O4 (x = 0) and Zn2GeO4 (x = 1). By comparing the Raman spectra of the crystallized glasses with the standard Zn1+x Ga2−2x Ge x O4 polycrystals synthesized in our lab by solid-state reaction (for more detail, refer to our previous work19), we provide the first direct evidence for the formation of Zn1+x Ga2−2x Ge x O4 with x ≥ 0.4 in GCs (Fig. S1(b), supporting information).
The crystallinity (volume fraction of the crystalline phase) of the GCs can be estimated by the ratio of the area under the indexed diffraction peaks to that under the whole XRD patterns14. For ZnGa2O4 and Zn1+x Ga2−2x Ge x O4 GCs, the crystallinities are approximately 37% and 32%, respectively, which are close in value to each other. The total molar concentration of ZnO and Ga2O3 is only 36 mol. % for the ZnGa2O4 GCs, which is less than the calculated crystallinity. The reason for the discrepancy is not clear and the validity of this result has yet to be confirmed. Here, it should be noted that the presence of nucleating agent such as TiO2 in gallium-containing GCs favors the substitution of 6Ni2+ (ionic radius: 0.69 Å) for 6Ga3+ (ionic radius: 0.62 Å) via the following substitutional mechanism: Ti4+ + Ni2+ → 2Ga3+, where Ti4+ acts as charge compensator31, 32. The incorporation of Ti4+ in the precipitated nanospinels was confirmed by the TEM-EDS analysis on the selected crystallization area in the ZGO-0.15GC sample (Fig. S3, supporting information). It is possible that a certain degree of inversion may occur in realistic spinels during crystallization, i.e., a fraction of the Ni2+ can occupy non-luminescent tetrahedral sites as found in NiAl2O3 crystals33, 34, and hence the selective doping of Ni2+ in octahedral sites is highly desirable for enhanced NIR luminescence33.
The morphology, distribution and particle sizes of nanocrystals were determined from the HRTEM measurements. The precipitated nanoparticles, approximately 15 nm in diameter, are distributed uniformly in all the GCs (Fig. 1(c) and (d)). The ultra-fine particle size allows these materials to be polished as they are in the glass state and then crystallized without any significant degradation of the surface quality (shown photographically in Fig. 2(a) and Figs S4 and S5, supporting information). The crystallization process was highly reproducible as verified by the fact that GCs show similar performance can be obtained repeatedly under the same experimental condition.
The coordination states of Ni2+ can be approximately inferred from the color of the glasses and GCs, for example, blue, brown and yellow-green in the case of 4Ni2+, 5Ni2+, and 6Ni2+ coordination, respectively8. The as-made ZGO (Fig. 2(a)) and ZGGO glasses (Fig. S4, supporting information) are light brown in appearance, suggesting 5Ni2+ and 4Ni2+ coordination states, whereas the color of the crystallized glasses becomes light green and blue, indicative of 6Ni2+. Since 6Ni2+ possesses a larger crystal field stabilization energy (CFSE) value than that of 5Ni2+, the unstable 5Ni2+ in glasses tends to transform into 6Ni2+ during crystallization of the spinel phases. The absorption related to 6Ni2+ (e.g., around 1160 nm due to the 3 A 2(3F) → 3 T 2(3F) transition) in GCs increases with the concentration of NiO, indicative of an efficient partition of Ni2+ in ZnGa2O4 nanocrystals, e.g. more than 90% of Ni2+ can be successfully embedded in gallium-containing GCs7, whereas it is well known that substitutional doping a large fraction of TM ions into semiconductor nanocrystals is extremely difficult because of the intrinsic self-purification mechanism35. The absorption bands can be well fitted to the Tanabe-Sugano (TS) diagram for d 8 ions (Fig. 2(d)), with the values of Racah parameter (B) and crystal field strength (Dq) equal to 767 cm−1 and 917 cm−1, respectively.
An inspection of the transmission spectra of the Zn2GeO4/Li2Ge4O9 GCs (Fig. S5, supporting information) also indicates the presence of 6Ni2+, which is possible via the substitution of 6Ni2+ for 6Ge4+ in Li2Ge4O9 nanocrystals. Meanwhile, since both the valence and ionic radius of 4Zn2+ (0.60 Å) matches those of 4Ni2+ (ionic radius: 0.55 Å), the substitution of 4Ni2+ for 4Zn2+ in ZnGa2O4 and Zn2GeO4 nanocrystals may also occur, similar to the embedding of Ni2+ in Zn2SiO4 crystals36. For a detailed analysis and discussion of the absorption spectra, refer to Supporting Information (Fig. S6). The fabricated GCs with transmission larger than 80% demonstrate great potential to be drawn into fibers for use as fiber lasers and amplifiers.
The use of the nucleant TiO2 is very important; it drastically increases both the emission intensity (Fig. S7, supporting information) and lifetime (Fig. S8, supporting information) of the GCs as compared to those free of TiO2 but otherwise the GCs containing TiO2 were thermally treated under identical conditions. The enhancement effect can be understood based on the substitution mechanism by which Ni2+ substitutes for Ga3+ favorably as mentioned above. An intense broadband NIR emission (from 1100 to 1700 nm) was recorded from ZnGa2O4: Ni2+ and Zn1+x Ga2−2x Ge x O4: Ni2+ GCs, but was very weak from Zn2GeO4/Li2Ge4O9: Ni2+ GCs. Both the emission intensity (Fig. 3(a)) and lifetime (Fig. 3(b)) (defined as the time taken for the emission intensity to decay to 1/e of its initial value) increase with NiO for the Zn1+x Ga2−2x Ge x O4: Ni2+ GCs, in contrast to ZnGa2O4: Ni2+ GCs where concentration quenching has already set in at the lowest doping level (~0.15 mol. %). However, in the cases with a fixed NiO, the ZnGa2O4: Ni2+ GCs exhibit stronger NIR emission and longer lifetime than those of Zn1+x Ga2−2x Ge x O4: Ni2+ GCs, e.g., a five-fold increase in the intensity and a two-fold increase in the lifetime when NiO was 0.15 mol. %. For the Zn1+x Ga2−2x Ge x O4: Ni2+ GCs, the emission intensity appears not to saturate at 0.5 mol. %. GCs have also been fabricated doped with 0.7 mol. % NiO. However, the samples suffer from significant devitrification due to NiO-assisted growth of large sized crystals commonly found in glasses heavily doped with NiO7. As a result, the 0.7 mol. % doped GCs become opaque and this restricts their use for optical applications, and hence does not warrant further study.
The differences in the crystal structures between the ZGO and ZGGO GCs may account for the contrast in the luminescent property. The Zn1+x Ga2−2x Ge x O4 crystals were assumed to be a solid solution between the normal ZnGa2O4 and inverted Zn2GeO4 spinel structures19, 37. Pure Zn1+x Ga2−2x Ge x O4 spinels can be synthesized for x ranging from 0 to 0.519. Our recent study of Mn doped Zn1+x Ga2−2x Ge x O4 phosphors shows that the substitution of Ge4+ for octahedrally coordinated 6Ga3+ helps to separate Mn4+ which also substitutes for 6Ga3+, thus resulting in an enhanced emission of Mn4+ 38. It is assumed that a similar separating effect exists for Ni2+ as well, i.e., 6Ni2+ ions are well separated in the ZGGO GCs, and thus the concentration quenching is postponed. As more Ni2+ ions diffuse from the surface to the inside of the nanocrystals and/or are shielded by the nanocrystals from the outside high-phonon energy environment, the non-radiative relaxation rate is reduced, and the lifetime increases accordingly. However, in the case of ZnGa2O4 GCs, Ni2+ substituting for 6Ga3+ is not well separated because there is no “separating agent” akin to Ni2+ doped Ga2O3 GCs where concentration quenching already starts at 0.10 mol. % low content of Ni2+ 7, and this accounts for the observed decreasing lifetime in the ZGO GCs.
Previous studies have shown that the Cr3+-doped and germanium-substituted compounds (Zn1+x Ga2−2x Ge x O4, x ≤ 0.5) exhibit much brighter and longer persistence luminescence than pure Cr3+-doped ZnGa2O4 spinels37. The 2Ga3+ → Ge4+ + Zn2+ substitution induces an inversion increase in the spinel structure, that is, an increased amount of Ga3+ now occupies the tetrahedral 4Zn2+ sites, forming the so-called anti-site defects . According to ref. 37, the enhancement of Cr3+ emission relies on the formation of anti-site defects, however, the presence of such defects definitely has an adverse effect on the luminescence of Ni2+ because of the reduced proportion of 6Ga3+ sites. On the other hand, it is possible that the substitution of Ge4+ for 6Ga3+ may generate octahedrally coordinated 6Ge4+, which may in turn be substituted by Ni2+. However, considering the fact that only weak NIR emission was observed from the Zn2GeO4/Li2Ge4O9: Ni2+ GCs, and the large mismatch in valence and ionic radii between 6Ni2+ and 6Ge4+ (ionic radius: 0.53 Å), the substitution of 6Ni2+ for 6Ge4+ is severely limited, akin to the partition of Ni2+ in K2SiF6 nanocrystals embedded GCs17. Moreover, no NIR emission related to the tetrahedrally coordinated 4Ni2+, e.g., Ni2+ doped Zn2SiO4 or Zn2GeO4 crystals, has been recorded even at cryogenic temperatures36. All these effects account for the observed weaker emission intensity of the ZGGO GCs than that of the ZGO GCs.
The ZnGa2O4: Ni2+ GCs, were selected for further study of internal fluorescence quantum efficiency (η) due to the stronger NIR emission and longer lifetime of Ni2+. Figure 4 shows the NIR emission lifetime of Ni2+ as a function of temperature from room temperature (300 K) down to liquid helium temperature (10 K). The sudden drop in lifetime at around 100 K indicates the occurrence of phonon-assisted non-radiative relaxation39. As shown in the inset, the decay curve has a strong non-exponential characteristic, implying multiple site effects of Ni2+ and non-radiative multipolar interactions among Ni2+. The value of η can be calculated as η = τ 300K/τ 0K, where τ 300K (~0.17 ms) and τ 0K (~0.62 ms, obtained by linear extrapolation to 0 K) are the lifetimes at the room and absolute zero temperatures, respectively. It is about 25% for the ZnGa2O4: Ni2+ GCs, which is less than that of ZnAl2O4: Ni2+ (~55%)40, LiGa5O8: Ni2+(~60%)12 and BaAl2Ti6O16: Ni2+ (~65%)39 GCs. However, it is comparable to that of Ga2S3: Cr4+ chalcogenide GCs (~25%)4 and even larger than that of pure Ni2+ doped ZnGa2O4 crystals (~18%)41.
The stimulated emission cross section (σe) was calculated using the McCumber formula4 and was found to be 0.52 × 10−20 cm2. The product of σe and τ 300K (proportional to the amplification gain and inversely proportional to the laser oscillation threshold) taken as a figure of merit (FOM) for the ZGO-0.15GC sample is 1.23 × 10−24 cm2·s if the 1/e lifetime is used for the calculation, and is about 3.79 × 10−24 cm2·s if the average lifetime is used for the calculation, which is comparable to that of ZnAl2O4: Ni2+ (~3.1 × 10−24 cm2·s)40, LiGa5O8: Ni2+ (~3.7 × 10−24 cm2·s)12 and BaAl2Ti6O16: Ni2+ (~3.3 × 10−24 cm2·s)39 GCs, and much larger than Ga2S3: Cr4+ chalcogenide (ChG) GCs (~0.62 × 10−24 cm2·s) which are known for the difficulty of preparation4. Light amplification at similar O-band wavelengths can be also achieved for Pr3+ or Dy3+ doped fluoride and ChG glasses of very low phonon energy. In comparison, the FOM of the ZGO-0.15GC is less than that of Pr3+ doped Ge-Ga-S ChG glass (4.79 × 10−24 cm2·s)42, however, it is much larger than in the case of Pr3+ doped ZBLAN (0.38 × 10−24 cm2·s) and Dy3+ doped Ge-Ga-S ChG glasses (1.4 × 10−24 cm2·s)43, 44. Moreover, the present GCs are superior to rare-earth doped glasses in terms of the availability of a broad tuning range of wavelength. A comparison of the luminescent properties (λ peak, peak emission wavelength, τ 298K, decay lifetime at the room temperature, and FOM) and crystal field parameters (Dq and B) of Ni2+ in GCs containing different spinels is shown in Table 1. The magnitude of crystal field strength Dq is a measure of the interaction of the 3d-electrons with the rest of the lattice, and the main contribution arises from the nearest neighbors. Although Ni2+ substitutes for Ga3+ in both ZGO and ZGGO GCs, the Dq value of the latter is slightly less than that of the former GCs, which, according to the ligand field theory, is due to the distortion of ligands inducing a weakening effect on the crystal field strength of the central ion45.
It is important to stress that the synthesized ZnGa2O4: Ni2+ GCs are highly reproducible to allow for a fluctuation in the thermal treatment temperature, for example, transparent GCs with the broadband near NIR emission can be obtained at a crystallization temperature ranging from 750 to 800 °C (Fig. S9 in the supporting information). This is a very important advantage for the “melt-in-tube” method, for which the core fiber is covered with the SiO2 cladding, and the heat transfer process during the heat treatment can be different from that of the glass sample. Because of different thermal treatment temperature, higher for the “melt-in-tube” method, GCs with required luminescent properties and transparency should be obtained in a temperature range as broad as possible. In this respect, the studied ZGO GCs are perfectly matched to the “melt-in-tube” method, which will be the subject of our next study to succeed in making them into fibers.
The selective doping of Ni2+ in ZnGa2O4 and Zn1+x Ga2−2x Ge x O4 nano-spinels via the controlled crystallization results in a broadband NIR emission. The use of nucleating agents such as TiO2 promotes occupation of the octahedral Ga3+ sites by Ni2+ and leads to enhanced luminescence and prolonged lifetime, whereas the partition of Ge4+ in ZnGa2O4 spinels leads to a reduced NIR emission, which is assumed to be related to the formation of anti-site defects. The large mismatch of valence and ionic radii between 6Ni2+ and 6Ge4+ considerably limits the substitution of 6Ni2+ for 6Ge4+, which also partly accounts for the comparatively weaker NIR emission from the Zn1+x Ga2−2x Ge x O4: Ni2+ GCs. The stronger NIR emission, excellent optical quality and reproducibility, as well as a tolerance for thermal treatment temperature make ZnGa2O4: Ni2+ nano-spinels embedded GCs highly promising candidates for broadband fiber amplifiers. Future work will focus on fabricating GC fibers by the “melt-in-tube” method.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This study was supported by the National Natural Science Foundation of China (61227013, 51302082, 61307104, 61422505, 61405044 and 61575050), National Key Scientific Instrument and Equipment Development Project (No. 2013YQ040815), Program for New Century Excellent Talents in University (NCET-12-0623), Open Project Program of the Jiangsu Key Laboratory of Advanced Laser Materials and Devices (KLALMD-2015-07), the Fundamental Research Funds for the Central Universities and the 111 project (B13015) to the Harbin Engineering University.
Electronic supplementary material
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.