A top gate structure is currently the most widely used in the bulk production of metal oxide semiconductor (MOS) thin-film transistors (TFTs), because it not only reduces the parasitic capacitance between the source and gate but also prevents the active layer from environmental influences.11. T. J. Yen, A. Chin, and V. Gritsenko, Nanomaterials 10(11), 2145 (2020). https://doi.org/10.3390/nano10112145 However, the mainstream manufacturing methods for top-gate MOS devices are based on expensive and time-consuming vacuum-based deposition processes, as compared to solution-processed fabrication methods, which are simpler and less expensive.22. R. Chen and L. Lan, Nanotechnology 30(31), 312001 (2019). https://doi.org/10.1088/1361-6528/ab1860Although there are many reports on high-performance top-gate solution-processed MOS TFTs, how the stability of solution-processed top-gate structured TFTs can be improved has been rarely reported.33. J.-S. Park, J. K. Jeong, H.-J. Chung, Y.-G. Mo, and H. D. Kim, Appl. Phys. Lett. 92(7), 072104 (2008). https://doi.org/10.1063/1.2838380 On the one hand, top-gate TFTs using the solution process may have the surface of the active layer damaged by acidic by-products that were produced by the following condensation reaction of a solution-processed dielectric, which affects device stability.44. X. Yang, Y. Gao, and X. Li, Optoelectron. Adv. Mater. Rapid Commun. 13, 343–347 (2019). Solution-processed top gate TFTs can choose the active layer materials such as binary Zn–Sn–O (ZTO), which is not only anti-corrosive but also exhibits superior chemical durability against toxic solutions while maintaining excellent electrical properties.55. X. Yu, T. J. Marks, and A. Facchetti, Nat. Mater. 15(4), 383–396 (2016). https://doi.org/10.1038/nmat4599 On the other hand, typically, cation doping is used to inhibit the formation of oxygen vacancy trap states to improve the stability of TFTs because of their strong metal-oxygen binding dissociation energy. Our previous work demonstrated that the doping of the W component significantly improves the bias stability of ZTO TFTs.66. X. Yang, S. Jiang, J. Li, J.-H. Zhang, and X.-F. Li, RSC Adv. 8(37), 20990–20995 (2018). https://doi.org/10.1039/C8RA02925C However, the previous studies show that the TFT mobility deteriorates severely when this strategy is used. Therefore, to solve this problem, anionic dopants, such as F and N, have also been extensively investigated, which can replace oxygen in metal oxide semiconductors and reduce oxygen vacancies to simultaneously improve the mobility and stability of MOS TFTs.7,87. C. Rao, J. Phys. Chem. Lett. 6(16), 3303–3308 (2015). https://doi.org/10.1021/acs.jpclett.5b010068. J. Su, Y. Ma, H. Yang, R. Li, L. Jia, D. Liu, and X. Zhang, J. Vacuum Sci. Technol. A: Vacuum, Surf., Films 37(6), 061511 (2019). https://doi.org/10.1116/1.5127889 To date, however, there are rare reports of using solutions to study the stability of top-gate TFTs, and it is meaningful and highly necessary for improving high stability of top-gate TFTs to study cation and anion co-doping ZTO devices.

In this Letter, we present a strategy to improve the stability of solution-processed top-gate bottom-contact (TGBC) MOS TFTs with cationic W and anionic F co-doped ZTO films as an active layer. The W: F co-doping stability improvement mechanism is systematically studied on the devices/films.

Tungsten hexachloride (WCl6, 99%, Alfa Aesar), zinc acetate dihydrate (Zn (CH3COO)2·2H2O), chloride pentahydrate (SnCl4·5H2O, 98%, Sigma), and 1,1,1-trifluoro-2,4-pentanedione are dissolved in 2-methoxyethanol (2-ME) to yield solutions with a concentration of 0.15 M. The molar ratio of F: W: Sn: Zn in the precursor solution with a concentration of 0.15 M was 1: 1: 30: 70.

The TGBC structure WZTOF TFTs are fabricated on a quartz glass substrate, and the 50 nm indium tin oxide (ITO) deposited by sputtering is patterned into the source and drain electrodes. The WZTOF precursor solution is spin-coated and annealed at 300 °C for 1 h. The process is repeated twice to obtain the thickness for 30 nm and then patterned by photolithography and wet etching. The active layers are patterned by etching with oxalic acid. Then, after spin-coating the HfAlO solution, it is annealed at 270 °C for 1 h. This process is repeated three times to obtain the desired GI thickness for 100 nm. The dielectric constant of HfAlO is 12.7, and the capacitance density is 118 nF/cm2. After via-hole formation in the HfAlO layer, a 50 nm thickness ITO is sputtered and patterned as the gate electrode. The channel width (W) is 50 μm, and the length (L) is 5 μm of the TFTs. The schematic cross section of the device structure and a top-view micrograph of the TFTs are shown in Fig. 1.

The transmittances of thin films on quartz substrates are measured by UV−vis spectroscopy (H-3900, Hatachi). Grazing incidence x-ray diffraction (GIXRD) probes the thin-films crystallinity. The chemical and structural properties of the active layer films are examined by x-ray photoelectron spectroscopy (XPS). The capacitance characteristics are measured under a frequency of 1 kHz by WAYNE KERR 6500. Electrical measurement of devices is carried out with a semiconductor characterization system (Keithley, 4200-SCS).

Figure 2(a) shows the XRD spectra of ZTO, ZTOF, WZTO, and WZTOF thin films without exhibiting any obvious diffraction peaks. This analysis reveals the amorphous structure of ZTO, ZTOF, WZTO, and WZTOF thin films, which contributes to the formation of a smooth film surface as shown in the inset of Fig. 2(a), which facilitates the performance of the devices. The Tauc plots of ZTO, ZTOF, WZTO, and WZTOF films are plotted in Fig. 2(b), from which the optical bandgap of each film can be determined. The optical bandgap of the film can be calculated by the following expression: where A and h are the constant and Planck's constant, respectively, Eg is the optical bandgap, and v is the light frequency. The optical bandgap of the devices ZTO, ZTOF, WZTO, and WZTOF are 3.42, 3.47, 3.44, and 3.52 eV, respectively. The widening of the optical bandgap could be explained by the Burstein–Moss effect.99. K. Saw, N. Aznan, F. Yam, S. Ng, and S. Pung, PLoS ONE 10(10), e0141180 (2015). https://doi.org/10.1371/journal.pone.0141180X-ray photoelectron spectroscopy analysis is done to investigate the difference in the chemical bonding of ZTO, ZTOF, WZTO, and WZTOF films in Fig. 3. As shown in Fig. 3(a), a significant F 1s spectrum located at 685.24 eV is obtained in the ZTOF and WZTOF films. It indicates that the F dopant has been effectively incorporated into the ZTOF and WZTOF films. Figure 3(b) shows the W 4f spectra with thin films. The W 4f7/2 and 4 f5/2 peaks of the WZTO film are centered at 35.3 and 37.6 eV, respectively, and show a slight shift toward higher binding energies with the addition of the F element. This is attributed to the change in electronegativity of the F and O elements.66. X. Yang, S. Jiang, J. Li, J.-H. Zhang, and X.-F. Li, RSC Adv. 8(37), 20990–20995 (2018). https://doi.org/10.1039/C8RA02925C Since the electronegativity of F is greater than that of O, the F ion can attract electrons from the W atom, which is much stronger than O, leading to an increase in the binding energy. This is consistent with the previous report on Sn: F co-doped TiO2 TFTs by Duan et al.1010. Y. Duan, J. Zheng, M. Xu, X. Song, N. Fu, Y. Fang, X. Zhou, Y. Lin, and F. Pan, J. Mater. Chem. A 3(10), 5692–5700 (2015). https://doi.org/10.1039/C4TA07068BIn Fig. 3(c), the O 1s peaks of WZTOF films are divided into OI, OII, and OIII. OI at 530.2 eV is related to the M–O bond, while OII at 531.34 eV can be attributed to oxygen vacancy-related defects. In addition, OIII at 532.0 eV is almost associated with oxygen in the hydroxyl group. The corresponding area proportions of OI, OII, and OIII are plotted in Fig. 3(d). The integral area of OII decreases monotonically from 15% to 7% with W: F co-doping; similarly, the integral area of OIII decreases from 10% to 7%, 6%, and 4% with F doing, W doping, and W: F co-doping, respectively. This result indicates that the oxygen-related defects associated with W–F co-doping are significantly reduced, which may improve mobility and stability.The electrical hysteresis of the source-drain current (IDS) as a function of VGS is shown in Fig. 4(a) for ZTO, WZTO, and WZTOF TFTs. Clockwise hysteresis shows that negative charge carriers are injected into the gate insulator or trapped interface between the channel and the gate insulator. The ZTO TFT has substantially the largest electrical hysteresis, while WZTOF TFTs exhibited minimal hysteresis, suggesting that fewer negative charge carriers were trapped at the channel-gate insulator contact, which was further confirmed by the smallest subthreshold.1111. J.-T. Li, L.-C. Liu, J.-S. Chen, J.-S. Jeng, P.-Y. Liao, H.-C. Chiang, T.-C. Chang, M. I. Nugraha, and M. A. Loi, Appl. Phys. Lett. 110(2), 023504 (2017). https://doi.org/10.1063/1.4973992 The inset of Fig. 4(a) shows the output characteristic curves of the WZTOF devices. The IDS increases with increasing drain voltage and shows a linear relationship in the lower gate voltage range and no current blocking occurs. The extracted electrical parameters of the ZTO, WZTO, and WZTOF TFTs were summarized in Fig. 4(b). Compared with the undoped ZTO TFT, the mobility (μ) of the WZTO TFT after cationic W doping is reduced to 0.93 cm2 V−1 s−1, which may be due to W as an ionization impurity that enhances the coulomb scattering. Moreover, the repeating experiments proved that anion F and cation W co-doped ZTO TFTs can be improved to 3.14 cm2 V−1 s−1, and the highest mobility can reach 6.87 cm2 V−1 s−1, as shown in Fig. 4(c). This is because the oxygen vacancies (Vo··) with doubly positive charges in the channel layer are passivated by a fluorine atom sitting on an oxygen vacancy with a neutral charge (Fox), which is in accord with XPS results, as1212. Y.-J. Choi, K.-M. Kang, and H.-H. Park, Sol. Energy Mater. Sol. Cells 132, 403–409 (2015). https://doi.org/10.1016/j.solmat.2014.09.029 Vo··+2e−+Fox↔Fo·+e−.(1)The decrease in the on-current is not significant compared to the decrease in oxygen vacancies. Because an oxygen atom sitting on an oxygen lattice site (Oox) is substituted by a Fox, and a fluorine ion sitting on an oxygen lattice site with a positive charge (Fo·) can release an electron (e-) as1212. Y.-J. Choi, K.-M. Kang, and H.-H. Park, Sol. Energy Mater. Sol. Cells 132, 403–409 (2015). https://doi.org/10.1016/j.solmat.2014.09.029 Figure 4(d) shows the capacitance–voltage (C–V) curves for TFT devices with different active layers that can be employed to investigate the trap states in the ZTO layer. The C–V curve of W: F co-doping ZTO TFTs has a larger and steeper slope and a smaller area between the hysteresis curves compared to un-doped ZTO TFTs, which is attributed to the decrease in trap states. At the same time, the XPS result confirms the suppression of the oxygen-related defect by W and F co-doping. Moreover, when F replaces O, the electronic perturbation is largely confined to the filled valence band, and the scattering of conduction electrons is minimized,13,1413. R. G. Gordon, MRS Bull. 25(8), 52–57 (2000). https://doi.org/10.1557/mrs2000.15114. F.-H. Wang, C.-F. Yang, and Y.-H. Lee, Nanoscale Res. Lett. 9(1), 300 (2014). https://doi.org/10.1186/1556-276X-9-300 and the influence of grain boundary scattering on mobility could be weakened and reduced due to the stronger electronegativity.1515. J. Ma, D. Lin, P. Li, G. Yang, and Y. Liu, Chin. Sci. Bull. 65(25), 2678–2690 (2020). https://doi.org/10.1360/TB-2020-0262 Obviously, the fact further demonstrates that the trap states are effectively suppressed by W: F co-doping, which is of benefit to improve the stability and mobility of the TFTs.Negative bias illumination stress (NBIS) is measured to evaluate the stability of the fabricated devices. Figure 5(a) shows the variation of the transmission characteristics of ZTO, WZTO, and WZTOF TFTs under NBIS with 10000 lux of white light at room temperature. Due to the abundant oxygen vacancy (VO) in the ZTO film, the NBIS instability can be explained by the photo-transition mechanism. Photo-irradiation drives the transition of oxygen vacancy from the neutral stable (VO) to the metastable (VO2+) active state in TFTs. Other reports have shown that the VO stable state is a nonconducting deep state, while the increase in VO2+ can contribute to the NBIS instability and the negative shift in Vth.1616. W. Pan, X. Zhou, Q. Lin, J. Chen, L. Lu, and S. Zhang, J. Mater. Chem. C 10(8), 3129–3138 (2022). https://doi.org/10.1039/D1TC05651D Therefore, an effective way to improve the stability is to control the oxygen vacancies in the films. W as an inhibitor has strong W–O binding energy (720 kJ mol−1), so it can effectively reduce the oxygen vacancies in the WZTO component and improve the NBIS stability of WZTO TFTs. Furthermore, according to the first-principles calculation reported by Liu et al., the doped fluorine ions can easily occupy the VO site with less energy.1010. Y. Duan, J. Zheng, M. Xu, X. Song, N. Fu, Y. Fang, X. Zhou, Y. Lin, and F. Pan, J. Mater. Chem. A 3(10), 5692–5700 (2015). https://doi.org/10.1039/C4TA07068B Therefore, F can further reduce the VO defect in the WZTOF film and improve the NBIS stability based on the W doping. The enhanced NBIS stability by W: F doping follows the above XPS result.Figure 5(b) exhibits the comparisons of ΔVth under NBIS and SS between this work and other related typical investigations.16–2416. W. Pan, X. Zhou, Q. Lin, J. Chen, L. Lu, and S. Zhang, J. Mater. Chem. C 10(8), 3129–3138 (2022). https://doi.org/10.1039/D1TC05651D17. D.-Y. Zhong, J. Li, C.-Y. Zhao, C.-X. Huang, J.-H. Zhang, X.-F. Li, X.-Y. Jiang, and Z.-L. Zhang, IEEE Trans. Electron Devices 65(2), 520–525 (2018). https://doi.org/10.1109/TED.2017.277974318. H. Zhang, L. Liang, X. Wang, Z. Wu, and H. Cao, IEEE Trans. Electron Devices 69(1), 152–155 (2022). https://doi.org/10.1109/TED.2021.312627819. M.-J. Park, J.-Y. Bak, J.-S. Choi, and S.-M. Yoon, presented at the 2014 21st International Workshop on Active-Matrix Flatpanel Displays and Devices (AM-FPD), Kyoto, Japan, 2014.20. Y. Liu, C. Liu, H. Qin, C. Peng, M. Lu, Z. Chen, and Y. Zhao, Membranes 11(11), 902 (2021). https://doi.org/10.3390/membranes1111090221. J. Li, Y.-H. Zhou, D.-Y. Zhong, X.-F. Li, and J.-H. Zhang, IEEE Trans. Electron Devices 66(10), 4205–4210 (2019). https://doi.org/10.1109/TED.2019.293648422. J. Li, Q. Chen, Y.-H. Yang, Y.-H. Zhou, D.-Y. Zhong, W.-Q. Zhu, J.-H. Zhang, and Z.-L. Zhang, J. Phys. D: Appl. Phys. 52(31), 315105 (2019). https://doi.org/10.1088/1361-6463/ab209823. H. Xu, M. Xu, M. Li, Z. Chen, J. Zou, W. Wu, X. Qiao, H. Tao, L. Wang, H. Ning, D. Ma, and J. Peng, ACS Appl. Mater. Interfaces 11(5), 5232–5239 (2019). https://doi.org/10.1021/acsami.8b1832924. W. Song, L. Lan, M. Li, L. Wang, Z. Lin, S. Sun, Y. Li, E. Song, P. Gao, Y. Li, and J. Peng, J. Phys. D: Appl. Phys. 50(38), 385108 (2017). https://doi.org/10.1088/1361-6463/aa83ee In our experiment, the WZTOF TFT presents superior stability under NBIS at just −0.09 V, which is much smaller in all solution-processed TFTs and even as excellent as the IGZO TFTs by sputtering. On the one hand, the co-doping of anions and cations can further suppress the oxygen vacancy defects in the films and improve the stability of NBIS based on cation doping. On the other hand, solution process MOS TFTs are susceptible to atmospheric effects, leading to device instability.25,2625. Y. Ding, T. Li, B. Yan, G. Liu, and F. Shan, Appl. Phys. Lett. 121(26), 263301 (2022). https://doi.org/10.1063/5.012845726. Y. Ding, Y. Ren, G. Liu, and F. Shan, IEEE Trans. Electron Devices 69(7), 3722–3726 (2022). https://doi.org/10.1109/TED.2022.3175674 Moreover, the NBIS stability of WZTOF TFTs after three months in the air remains essentially the same as shown in the inset of Fig. 5(a), which is the main attraction of top-gate devices. Therefore, this also opens the door for further research on the device stability of the top-gate structure fabricated by the solution process.

We prepared W: F co-doped ZTO TFTs with the TGBC structure using a solution process, which has excellent electrical properties. The XPS results indicate that W: F co-doping can significantly reduce the oxygen vacancies in the ZTO film. The μ, Vth, SS, and Ion/Ioff of WZTOF TFTs are 3.14 cm2 V−1 s−1, 0.24 V, 0.157 V/dec, and 3.38 × 106, respectively. More importantly, the ΔVth under NBIS is only −0.09 V, which is much as the same as that of the TFTs prepared by the sputtering method. The XPS and C–V measurement further confirm that W: F co-doping can effectively decrease the trap state and improve the performance of the ZTO TFT.

This work was supported by the National Natural Science Foundation of China under Grant Nos. 62174105 and 61674101. This work was also supported by the Shanghai Education Development Foundation and the Shanghai Municipal Education Commission under Grant No. 18SG38.

Conflict of Interest

The authors have no conflicts to disclose.

Author Contributions

Sunjie Hu: Writing – original draft (equal); Writing – review & editing (equal). Meng Xu: Investigation (equal); Resources (equal). Cong Peng: Investigation (equal); Resources (equal). Longlong Chen: Investigation (equal); Resources (equal). Hai Liu: Investigation (equal); Resources (equal). Xifeng Li: Supervision (equal); Validation (equal).

The data that support the findings of this study are available within the article.

REFERENCES

1. T. J. Yen, A. Chin, and V. Gritsenko, Nanomaterials 10(11), 2145 (2020). https://doi.org/10.3390/nano10112145, Google ScholarCrossref2. R. Chen and L. Lan, Nanotechnology 30(31), 312001 (2019). https://doi.org/10.1088/1361-6528/ab1860, Google ScholarCrossref3. J.-S. Park, J. K. Jeong, H.-J. Chung, Y.-G. Mo, and H. D. Kim, Appl. Phys. Lett. 92(7), 072104 (2008). https://doi.org/10.1063/1.2838380, Google ScholarScitation, ISI4. X. Yang, Y. Gao, and X. Li, Optoelectron. Adv. Mater. Rapid Commun. 13, 343–347 (2019). Google Scholar5. X. Yu, T. J. Marks, and A. Facchetti, Nat. Mater. 15(4), 383–396 (2016). https://doi.org/10.1038/nmat4599, Google ScholarCrossref6. X. Yang, S. Jiang, J. Li, J.-H. Zhang, and X.-F. Li, RSC Adv. 8(37), 20990–20995 (2018). https://doi.org/10.1039/C8RA02925C, Google ScholarCrossref7. C. Rao, J. Phys. Chem. Lett. 6(16), 3303–3308 (2015). https://doi.org/10.1021/acs.jpclett.5b01006, Google ScholarCrossref8. J. Su, Y. Ma, H. Yang, R. Li, L. Jia, D. Liu, and X. Zhang, J. Vacuum Sci. Technol. A: Vacuum, Surf., Films 37(6), 061511 (2019). https://doi.org/10.1116/1.5127889, Google ScholarCrossref, ISI9. K. Saw, N. Aznan, F. Yam, S. Ng, and S. Pung, PLoS ONE 10(10), e0141180 (2015). https://doi.org/10.1371/journal.pone.0141180, Google ScholarCrossref10. Y. Duan, J. Zheng, M. Xu, X. Song, N. Fu, Y. Fang, X. Zhou, Y. Lin, and F. Pan, J. Mater. Chem. A 3(10), 5692–5700 (2015). https://doi.org/10.1039/C4TA07068B, Google ScholarCrossref11. J.-T. Li, L.-C. Liu, J.-S. Chen, J.-S. Jeng, P.-Y. Liao, H.-C. Chiang, T.-C. Chang, M. I. Nugraha, and M. A. Loi, Appl. Phys. Lett. 110(2), 023504 (2017). https://doi.org/10.1063/1.4973992, Google ScholarScitation, ISI12. Y.-J. Choi, K.-M. Kang, and H.-H. Park, Sol. Energy Mater. Sol. Cells 132, 403–409 (2015). https://doi.org/10.1016/j.solmat.2014.09.029, Google ScholarCrossref13. R. G. Gordon, MRS Bull. 25(8), 52–57 (2000). https://doi.org/10.1557/mrs2000.151, Google ScholarCrossref14. F.-H. Wang, C.-F. Yang, and Y.-H. Lee, Nanoscale Res. Lett. 9(1), 300 (2014). https://doi.org/10.1186/1556-276X-9-300, Google ScholarCrossref15. J. Ma, D. Lin, P. Li, G. Yang, and Y. Liu, Chin. Sci. Bull. 65(25), 2678–2690 (2020). https://doi.org/10.1360/TB-2020-0262, Google ScholarCrossref16. W. Pan, X. Zhou, Q. Lin, J. Chen, L. Lu, and S. Zhang, J. Mater. Chem. C 10(8), 3129–3138 (2022). https://doi.org/10.1039/D1TC05651D, Google ScholarCrossref17. D.-Y. Zhong, J. Li, C.-Y. Zhao, C.-X. Huang, J.-H. Zhang, X.-F. Li, X.-Y. Jiang, and Z.-L. Zhang, IEEE Trans. Electron Devices 65(2), 520–525 (2018). https://doi.org/10.1109/TED.2017.2779743, Google ScholarCrossref18. H. Zhang, L. Liang, X. Wang, Z. Wu, and H. Cao, IEEE Trans. Electron Devices 69(1), 152–155 (2022). https://doi.org/10.1109/TED.2021.3126278, Google ScholarCrossref19. M.-J. Park, J.-Y. Bak, J.-S. Choi, and S.-M. Yoon, presented at the 2014 21st International Workshop on Active-Matrix Flatpanel Displays and Devices (AM-FPD), Kyoto, Japan, 2014. Google Scholar20. Y. Liu, C. Liu, H. Qin, C. Peng, M. Lu, Z. Chen, and Y. Zhao, Membranes 11(11), 902 (2021). https://doi.org/10.3390/membranes11110902, Google ScholarCrossref21. J. Li, Y.-H. Zhou, D.-Y. Zhong, X.-F. Li, and J.-H. Zhang, IEEE Trans. Electron Devices 66(10), 4205–4210 (2019). https://doi.org/10.1109/TED.2019.2936484, Google ScholarCrossref22. J. Li, Q. Chen, Y.-H. Yang, Y.-H. Zhou, D.-Y. Zhong, W.-Q. Zhu, J.-H. Zhang, and Z.-L. Zhang, J. Phys. D: Appl. Phys. 52(31), 315105 (2019). https://doi.org/10.1088/1361-6463/ab2098, Google ScholarCrossref23. H. Xu, M. Xu, M. Li, Z. Chen, J. Zou, W. Wu, X. Qiao, H. Tao, L. Wang, H. Ning, D. Ma, and J. Peng, ACS Appl. Mater. Interfaces 11(5), 5232–5239 (2019). https://doi.org/10.1021/acsami.8b18329, Google ScholarCrossref24. W. Song, L. Lan, M. Li, L. Wang, Z. Lin, S. Sun, Y. Li, E. Song, P. Gao, Y. Li, and J. Peng, J. Phys. D: Appl. Phys. 50(38), 385108 (2017). https://doi.org/10.1088/1361-6463/aa83ee, Google ScholarCrossref25. Y. Ding, T. Li, B. Yan, G. Liu, and F. Shan, Appl. Phys. Lett. 121(26), 263301 (2022). https://doi.org/10.1063/5.0128457, Google ScholarScitation26. Y. Ding, Y. Ren, G. Liu, and F. Shan, IEEE Trans. Electron Devices 69(7), 3722–3726 (2022). https://doi.org/10.1109/TED.2022.3175674, Google ScholarCrossref© 2023 Author(s). Published under an exclusive license by AIP Publishing.