International Journal of Photoenergy

International Journal of Photoenergy / 2014 / Article
Special Issue

Materials, Designs, Fabrications, and Applications of Organic Electronic Devices

View this Special Issue

Research Article | Open Access

Volume 2014 |Article ID 412084 |

Shun-Wei Liu, Chih-Chien Lee, Yu-Ting Chung, Jiun-Haw Lee, Chin-Ti Chen, Juen-Kai Wang, "Improvement in Device Performance and Reliability of Organic Light-Emitting Diodes through Deposition Rate Control", International Journal of Photoenergy, vol. 2014, Article ID 412084, 7 pages, 2014.

Improvement in Device Performance and Reliability of Organic Light-Emitting Diodes through Deposition Rate Control

Academic Editor: K. R. Justin Thomas
Received25 Feb 2014
Revised09 Apr 2014
Accepted10 Apr 2014
Published27 Apr 2014


We demonstrated a fabrication technique to reduce the driving voltage, increase the current efficiency, and extend the operating lifetime of an organic light-emitting diode (OLED) by simply controlling the deposition rate of bis(10-hydroxybenzo[h]qinolinato) beryllium (Bebq2) used as the emitting layer and the electron-transport layer. In our optimized device, 55 nm of Bebq2 was first deposited at a faster deposition rate of 1.3 nm/s, followed by the deposition of a thin Bebq2 (5 nm) layer at a slower rate of 0.03 nm/s. The Bebq2 layer with the faster deposition rate exhibited higher photoluminescence efficiency and was suitable for use in light emission. The thin Bebq2 layer with the slower deposition rate was used to modify the interface between the Bebq2 and cathode and hence improve the injection efficiency and lower the driving voltage. The operating lifetime of such a two-step deposition OLED was 1.92 and 4.6 times longer than that of devices with a single deposition rate, that is, 1.3 and 0.03 nm/s cases, respectively.

1. Introduction

Molecular packing plays an important role in the function of organic electronics and optoelectronic devices [15]. For organic thin-film transistor (OTFT) applications, appropriate packing produces higher field-effect mobility resulting in increased electrical current [1]. Molecular packing of organic materials for OTFT (such as pentacene) can be adjusted by introducing the substituent to the molecules, surface treatment (e.g., octadecyltrichlorosilane), or insertion of self-assembled monolayer [6, 7]. For polymer solar cell application, molecular packing can be controlled by fabrication techniques such as thermal annealing or solvent annealing. Better molecular packing of donor material (e.g., poly(3-hexylthiophene)) not only improved hole mobility and charge balance but also increased absorption [8, 9]. For the small-molecular organic solar cell, molecular packing plays an important role in determining the open circuit voltage (). For organic materials with higher crystallinity, it resulted in higher saturation dark current, and in turn reduced the and hence power conversion efficiency, which can be engineered by inserting an interfacial layer or varying evaporation rate [10, 11]. For the organic light-emitting diode (OLED) application, we demonstrated in our previous study that molecular packing of a planar molecule, bis(10-hydroxybenzo[h]qinolinato) beryllium (Bebq2), can be controlled via the deposition rate during vacuum sublimation [1214]. Atomic force microscopy (AFM) measurements confirmed that a slower deposition rate (i.e., 0.03 nm/s) resulted in better molecular packing. When slower-deposition rate Bebq2 was used as the emitting layer (EML) and the electron-transporting layer (ETL) material in our OLED, such an ordered molecular packing improved electron mobility and reduced driving voltage. On the other hand, the ordered Bebq2 resulted in a photoluminescence (PL) quenching and a lower current efficiency (in terms of cd/A), so there appeared to be a tradeoff between the electrical and optical properties.

The deposition rate of organic materials was generally controlled at ~0.1 nm/s, with some fluctuations (e.g., 0.08–0.12 nm/s). An ultralow deposition rate (e.g., less than 0.01 nm/s) meant that device fabrication took a long time and was impractical from a manufacturing point of view. On the other hand, at ultrahigh deposition rates (e.g., higher than 10 nm/s) precise control over the thickness of the organic layer became problematic. This paper therefore looks at deposition rates between 0.03 and 1.3 nm/s due to practical considerations such as takt time and precise thickness control. During the evaporation process, temperature was carefully adjusted to achieve constant deposition rates of 0.03, 0.3, and 1.3 nm/s with fluctuations of 0.02–0.04, 0.27–0.33, and 1.1–1.4 nm/s, respectively. The separation of deposition rates was sufficient to avoid any possible overlap due to fluctuation in deposition rate.

In this report, we made use of a two-step evaporation technique for depositing Bebq2 at high and low rates sequentially. First, 55 nm of Bebq2 was deposited at a fast rate (1.3 nm/s) to form the EML (and ETL as well) alleviating the PL quenching at the recombination zone near the HTL/EML interface. This was followed by the deposition of a Bebq2 thin layer (5 nm) at a slow rate (0.03 nm/s) to modify the interface between the ETL and the cathode for efficient electron injection, which was deduced from electron-only device and AFM measurement. Although Bebq2 with slower deposition rate exhibited the lowest PL efficiency, it did not affect the device performances because it was far away from the recombination zone.

2. Experiment

To study the effects of the deposition rate on the device performance and lifetime of the electroluminescence (EL) device, five different device structures were fabricated, as summarized in Table 1. The authors selected ITO/NPB (60 nm)/Bebq2, (55 nm)/Bebq2, (5 nm)/LiF, and (1 nm)/Al (100 nm) as the EL samples, in which the 5 nm thick Bebq2 layer was deposited at three different rates of 0.03 nm/s (device A), 0.3 nm/s (device B), and 1.3 nm/s (device C). OLEDs with different single deposition rates (0.3 and 0.03 nm/s) were also fabricated and designated as devices D and E, respectively. The fabrication process was similar to that outlined in our previous reports, except for the purification process of the organic materials [15]. For this experiment, the organic materials were purified three times by temperature-gradient sublimation. This change indeed resulted in significant changes to the lifetime performances compared to our previous studies [14]. Despite similar device configurations and aging conditions, the operating lifetimes observed in this project were approximately 10 times longer than those in our previous paper [14]. We also measured the characteristics of the single carrier (electron) device to investigate the injection characteristics of the organic/electrode interface [16]. The device structures were Ca (150 nm; 0.1 nm/s)/Bebq2, (55 nm; 1.3 nm/s)/Bebq2, (5 nm; 0.03, 0.3 and 1.3 nm/s)/Ca, and (150 nm; 0.1 nm/s) on glass substrate, designated as devices F, G, and H, respectively. For our fabrication, the ITO/glass substrate (sheet resistance ~30 Ω/□) was commercially obtained from the Merck Corporation, with the low energy method (AC2; Riken Keiki) yielding a work function of 4.8 eV for the ITO. During the film deposition of the devices, the pressure of the chamber was maintained at below 4 × 10−6 torr. The thickness of each organic layer was calibrated by a surface profiler (Dektak 150; Veeco) before device fabrication. The active area of the devices was 4 mm2 and the devices were completed with an encapsulation in a glove box (O2 and H2O concentration below 0.1 ppm). A DC current/voltage source meter (2400; Keithley) was used to measure current density versus voltage characteristics, while the brightness was monitored with a spectrophotometer (PR650; Photo Research). The device lifetime was determined with a constant current density of 100 mA/cm2 in a DC driving condition. For AFM measurement (noncontact mode; Park system XE-100), the Si wafer was sequentially sonicated in acetone, methanol, and deionized water, and etched in dilute H2SO4 solution. Immediately after cleaning, the substrates were transferred into an evaporation chamber for deposition of the organic thin film. The surface roughness of the cleaned Si wafer was 0.2 nm.


LiF/Al,  nm
Bebq2, 5 nm,  nm/s
Bebq2, 55 nm,  nm/s
NPB, 60 nm
ITO, 80 nm


Ca, 150 nm
Bebq2, 5 nm,  nm/s
Bebq2, 55 nm, 1.3 nm/s
Ca, 150 nm

3. Results and Discussion

3.1. Device Performances

Figure 1 shows the characteristics and EL efficiency of devices A–E. The only difference in the OLED fabrication process among devices A, B, and C was the deposition rate of the ultrathin Bebq2 at the LiF/Al interface, with these being 0.03, 0.3, and 1.3 nm/s, respectively, while the deposition rate of the thick Bebq2 at the HTL interface was maintained at 1.3 nm/s. One should note that the current density decreased monotonically from 25.8 to 11 mA/cm2 at 6 V when the deposition rate of the final Bebq2 layer was increased from 0.03 to 1.3 nm/s. When devices C, D, and E were compared with cases where the entire Bebq2 layer was deposited at a constant evaporation rate of 1.3, 0.3, and 0.03 nm/s, respectively, current density increased monotonically as the deposition rate decreased, because lower evaporation rates resulted in higher electron mobility [17]. By comparing the characteristics of devices A, B, and C, we can see that the ultrathin Bebq2 film (5 nm) near the LiF/Al interface helped to reduce the driving voltage; that is, this thin layer may facilitate better electron injection from the cathode into the ETL and EML. To confirm this hypothesis, three electron-only devices with the device structure shown in Table 1 were fabricated. A thick Bebq2 (55 nm; 1.3 nm/s) layer followed by a thin Bebq2 (5 nm) layer deposited at rates of 0.03, 0.3, and 1.3 nm/s, respectively, was sandwiched between two Ca electrodes. Figure 2 shows the characteristics of the electron-only devices. As expected, the electron-only device with a thin Bebq2 (5 nm) layer of the lowest deposition rate (0.03 nm/s) exhibited the highest current density due to better electron injection capability, consistent with the device performances observed in Figure 1(a). By optimizing the carrier injection and carrier transport, one can theoretically obtain an OLED with the lowest driving voltage (device E). For the cases with a single deposition rate (1.3, 0.3, and 0.03 nm/s, corresponding to devices C, D, and E, resp.), one can note that the current efficiency decreased with the deposition rate, as shown in Figure 1(b), due to the decrease in PL efficiency with better molecular packing at a lower deposition rate [14]. By changing the deposition rate of the thin (5 nm) Bebq2 layer close to LiF/Al (devices A, B, and C), the current efficiency decreased monotonically from 7.5 to 5.4 cd/A at 100 mA/cm2 when the deposition rate of the final Bebq2 layer was increased from 0.03 to 1.3 nm/s. The change can be attributed to better electron injection at the cathode interface and hence better electron/hole balance [1820]. By simply decreasing the evaporation rate at the ETL and LiF/Al interface, device performances were significantly improved with lower driving voltages and higher current efficiencies.

3.2. Accelerated Lifetime Test

To evaluate the effects of the Bebq2/electrode on the device lifetime, the five types of Bebq2 based devices (A–E) were measured at a constant DC driven current density of 100 mA/cm2. Figures 3(a) and 3(b) show the time dependent variations of the device luminance and driving voltage. The initial luminance was 7470, 7300, 5320, 3360, and 3220 cd/m2 for devices A, B, C, D, and E, respectively. No detectable change in the EL spectra was observed for the five OLEDs during the accelerated lifetime test. This evidence suggested that there was no change in the chemical nature of these organic layers and no drastic alteration in the emission zone location during device operation; that is, the EL efficiency degradation was most likely due to the decrease in the created exciton density and radiative efficiency. When the deposition rate for the last 5 nm Bebq2 layer was decreased from 1.3 to 0.03 nm/s, the device exhibited a significantly longer lifetime compared to similar devices with no change in the deposition rate. The extracted half lifetimes of the OLEDs with a two-step deposition process (devices A and B) were 1200 and 833 min. at deposition rates of 0.03 and 0.3, respectively, for ultrathin Bebq2 (5 nm) on the bulk material (55 nm). This suggested that modifying the organic/electrode interface at the electron contact of OLED can significantly improve their operational reliability. In contrast, the corresponding half lifetimes of bulk Bebq2 devices with constant deposition rate were 410, 288, and 214 min. at deposition rates of 1.3, 0.3, and 0.03 nm/s, respectively. The extracted device lifetimes (devices C, D, and E) decreased as the film deposition rate decreased. The PL intensities were measured before and after the aging process, with decreases of 3.6%, 4%, 4.3%, 20%, and 35% observed for devices A, B, C, D, and E, respectively. One should note that the PL decrease was the most pronounced for device E. On the contrary, as shown in Figure 3(b), the voltage increase was highest for the OLED with the highest deposition rate (device C). By using a low deposition rate Bebq2 at the LiF/Al interface, the voltage increase was significantly alleviated in devices A and B. The voltage increase in an aged OLED is due to the trap density formation [2123]. By comparing devices C, D, and E, one may conclude that the lower deposition rates helped to reduce the trap formation. Device A exhibited the lowest voltage increase rate (even lower than device E), however, indicating that the trap distribution upon electrical aging was different under different deposition rates. For device C, with the highest deposition rate Bebq2, traps were formed near the interface in direct contact with the cathode. On the other hand, for device E, the traps were mainly located away from the cathode side. With the two-step evaporation process, the trap formation rate for the entire Bebq2 layer was decreased, extending the operating lifetime. This explained the differences in the EL lifetimes among devices A, B, and C, despite similar PL decay rates.

3.3. Surface Morphology

To further investigate the possible physical mechanisms in play, we examined the correlation of the morphology of the Bebq2 layer with other devices prepared on Si wafer [4]. Although the situation of Bebq2 deposited on Si substrate may be different from that on ITO glass, unfortunately, the surface of ITO is quite rough making it unsuitable for studying the surface condition of Bebq2 thin film. Figure 4 shows the AFM surface morphology measurement of the bulk Bebq2 films from the two-step deposition process. Such layers were fabricated with a constant deposition rate of 1.3 nm/s (55 nm) for the bulk material, followed by deposition of the last 5 nm ultrathin layer at the three different deposition rates of 0.3, 0.03, and 1.3 nm/s upon the bulk Bebq2. The scanning area of these images was 2 × 2 μm and the morphology of all films revealed a similar island-like surface texture. The average root-mean-square (RMS) roughness of the ultrathin Bebq2 films with deposition rates of 0.03, 0.3, and 1.3 nm/s on bulk Bebq2 surface was about 0.26, 0.30, and 0.38 nm, respectively (see Figures 4(a), 4(b), and 4(c)). We also extracted the variation values (the extracted values correspond to the highest and lowest heights) of these devices: 0.95, 1.35, and 1.55 nm for the ultrathin layer deposited at 0.03, 0.3, and 1.3 nm/s, respectively (see Figures 4(d), 4(e), and 4(f)). Note that in Figure 4(c) the deposition rate was kept at 1.3 nm/s for the whole 60 nm. The surface is rough and there are numerous “white spots” with a diameter ~200 nm. When decreasing the deposition rate for the final 5 nm at 0.3 and 0.03 nm/s, it can be seen that the surface became smooth and the number of “white spots” decreased. The dependence of the morphology results on the deposition rate can be explained by the fact that a lower deposition rate in the ultrathin layer enabled the planar Bebq2 molecules to settle into their lowest energy-packing configuration and produce a smooth surface (see Figure 4(d)). The optimization of the two-step process resulted in a significant decrease to the RMS roughness of the Bebq2 surface and the formation of a “good contact” interface between Bebq2/electrode. Consequently, we expect that such smooth contact would eliminate the obstruction buildup at the interface between organic/electrode. Conversely, the rough surface morphology of Bebq2 film (see Figures 4(e) and 4(f)) may result in weak adhesion of the subsequently deposited electrode layer and decrease the electron injection efficiency. The present results suggest that these factors are strongly affected by the interface qualities of Bebq2/electrode under different deposition rates. With the modification of the contact layer, the 5 nm Bebq2 with lower deposition rate produced a uniform buffer layer that reduced the driving voltage and gave a higher efficiency due to better charge balance. At a low deposition rate, the smooth Bebq2/electrode interface also facilitated stable contact for the subsequently deposited LiF/Al electrode. This explained why the trap distribution upon electrical aging was predominantly concentrated on the cathode side with high deposition rate Bebq2 (device C). On the other hand, for device E with low deposition rate Bebq2, the efficiency was low (as shown in Figure 1(b)), resulting in more serious joule heating near the recombination zone (HTL/Bebq2 interface) that assisted the crystallization process [24, 25] and decreased the PL efficiency.

4. Conclusion

We investigated the influence of the deposition rate of the Bebq2 film on the performance of OLEDs with a Bebq2 film and reached two conclusions. First, the current densities versus driving voltage and luminance versus current densities characteristics of the devices were found to be consistent with the extracted dependence of the carrier injection properties on the film deposition rate. In particular, when ultrathin films were fabricated at low deposition rates, the formation of a smooth Bebq2 surface may strongly enhance charge injection. This conclusion was supported by the electron-only measurements. Second, when the deposition rate for the last 5 nm Bebq2 layer was decreased from 1.3 to 0.03 nm/s, the devices exhibited significantly longer lifetimes compared to similar devices with no changes in the deposition rate. This indicated that the two-step deposition process was responsible for a significant increase in the surface smoothness of the Bebq2/electrode interface, possibly due to the strong and stable adhesion of Bebq2 and subsequently deposited Al electrode.

Conflict of Interests

The authors declare that they have no conflict of interests regarding the publication of this paper.


The authors thank the financial support by the National Science Council, ROC, under Grant nos. NSC 101-2221-E-002-156-MY3, 102-2221-E-002-182-MY3, NSC 102-2221-E-131-026-MY2, NSC 102-2221-E-131-030-MY2, NSC 102-2511-S-131-002, NSC 103-ET-E-011-004-ET, NSC 102-2627-E-002-002, and NSC 102-2622-E-011-018-CC3 and Academia Sinica of Taiwan.


  1. H. Klauk, “Organic thin-film transistors,” Chemical Society Reviews, vol. 39, pp. 2643–2666, 2010. View at: Publisher Site | Google Scholar
  2. G. Li, R. Zhu, and Y. Yang, “Polymer solar cells,” Nature Photonics, vol. 6, no. 3, pp. 153–161, 2012. View at: Publisher Site | Google Scholar
  3. B. J. Chen, W. Y. Lai, Z. Q. Gao, C. S. Lee, S. T. Lee, and W. A. Gamblig, “Electron drift mobility and electroluminescent efficiency of tris(8-hydroxyquinolinolato) aluminum,” Applied Physics Letters, vol. 75, p. 4010, 1999. View at: Publisher Site | Google Scholar
  4. C. B. Lee, A. Uddin, X. Hu, and T. G. Andersson, “Study of Alq3 thermal evaporation rate effects on the OLED,” Materials Science and Engineering: B, vol. 112, pp. 14–18, 2004. View at: Publisher Site | Google Scholar
  5. T. Matsushima, K. Shiomura, S. Naka, and H. Murata, “Optical, morphological, structural, electrical, molecular orientation, and electroluminescence characteristics of organic semiconductor films prepared at various deposition rates,” Thin Solid Films, vol. 520, no. 6, pp. 2283–2288, 2012. View at: Publisher Site | Google Scholar
  6. H. W. Zan, Y. H. Hsu, H. F. Meng, C. H. Huang, Y. T. Tao, and W. W. Tsai, “High output current in vertical polymer space-charge-limited transistor induced by self-assembled monolayer,” Applied Physics Letters, vol. 101, no. 9, Article ID 093307, 2012. View at: Publisher Site | Google Scholar
  7. S. Pola, C. H. Kuo, W. T. Peng, M. M. Islam, I. Chao, and Y. T. Tao, “Contorted tetrabenzocoronene derivatives for single crystal field effect transistors: correlation between packing and mobility,” Chemistry of Materials, vol. 24, pp. 2566–2571, 2012. View at: Publisher Site | Google Scholar
  8. F. Padinger, R. S. Rittberger, and N. S. Sariciftci, “Effects of postproduction treatment on plastic solar cells,” Advanced Functional Materials, vol. 13, no. 1, pp. 85–88, 2003. View at: Publisher Site | Google Scholar
  9. G. Li, V. Shrotriya, J. Huang et al., “High-efficiency solution processable polymer photovoltaic cells by self-organization of polymer blends,” Nature Materials, vol. 4, no. 11, pp. 864–868, 2005. View at: Publisher Site | Google Scholar
  10. H. C. Han, C. A. Tseng, C. Y. Du et al., “Enhancing efficiency with fluorinated interlayers in small molecule organic solar cells,” Journal of Materials Chemistry, vol. 22, pp. 22899–22905, 2012. View at: Publisher Site | Google Scholar
  11. C.-F. Lin, S.-W. Liu, C.-C. Lee et al., “Open-circuit voltage and efficiency improvement of subphthalocyanine-based organic photovoltaic device through deposition rate control,” Solar Energy Materials and Solar Cells, vol. 103, pp. 69–75, 2012. View at: Publisher Site | Google Scholar
  12. C. C. Lee, S. W. Liu, and Y. T. Chung, “Effect of deposition rate on device performance and lifetime of planar molecule-based organic light-emitting diodes,” Journal of Physics D: Applied Physics, vol. 43, no. 7, Article ID 075102, 2010. View at: Publisher Site | Google Scholar
  13. J.-H. Lee, C.-I. Wu, S.-W. Liu, C.-A. Huang, and Y. Chang, “Mixed host organic light-emitting devices with low driving voltage and long lifetime,” Applied Physics Letters, vol. 86, Article ID 103506, 2005. View at: Publisher Site | Google Scholar
  14. S.-W. Liu, C.-C. Lee, C.-H. Wang, J.-H. Lee, C.-T. Chen, and J.-K. Wang, “Enhancing performance of planar molecule-based organic light-emitting diodes through deposition-rate optimization: role of molecular packing,” Chemical Physics Letters, vol. 474, pp. 207–211, 2009. View at: Publisher Site | Google Scholar
  15. S.-H. Liao, J.-R. Shiu, S.-W. Liu et al., “Hydroxynaphthyridine-derived group III metal chelates: wide band gap and deep blue analogues of green Alq3 (tris(8-hydroxyquinolate)aluminum) and their versatile applications for organic light-emitting diodes,” Journal of the American Chemical Society, vol. 131, pp. 763–777, 2009. View at: Publisher Site | Google Scholar
  16. W. Brütting, S. Berleb, and A. G. Mückl, “Device physics of organic light-emitting diodes based on molecular materials,” Organic Electronics, vol. 2, pp. 1–36, 2001. View at: Google Scholar
  17. S. T. Lee, Z. Q. Gao, and L. S. Hung, “Metal diffusion from electrodes in organic light-emitting diodes,” Applied Physics Letters, vol. 75, p. 1404, 1999. View at: Publisher Site | Google Scholar
  18. J.-H. Lee, Y.-H. Ho, T.-C. Lin, and C.-F. Wu, “High-efficiency fluorescent blue organic light-emitting device with balanced carrier transport,” Journal of the Electrochemical Society, vol. 154, pp. 226–228, 2007. View at: Publisher Site | Google Scholar
  19. H. Aziz, Z. D. Popovic, and N.-X. Hu, “Organic light emitting devices with enhanced operational stability at elevated temperatures,” Applied Physics Letters, vol. 81, p. 370, 2002. View at: Publisher Site | Google Scholar
  20. C.-C. Lee, M.-Y. Chang, P.-T. Huang, Y.-C. Chen, and S.-W. Liu, “Electrical and optical simulation of organic light-emitting devices with fluorescent dopant in the emitting layer,” Journal of Applied Physics, vol. 101, p. 114501, 2007. View at: Publisher Site | Google Scholar
  21. S. A. Van Slyke, C. H. Chen, and C. W. Tang, “Organic electroluminescent devices with improved stability,” Applied Physics Letters, vol. 69, no. 15, pp. 2160–2162, 1996. View at: Google Scholar
  22. D. Y. Kondakov, J. R. Sandifer, C. W. Tang, and R. H. Young, “Nonradiative recombination centers and electrical aging of organic light-emitting diodes: direct connection between accumulation of trapped charge and luminance loss,” Journal of Applied Physics, vol. 93, no. 2, pp. 1108–1119, 2003. View at: Publisher Site | Google Scholar
  23. J.-H. Lee, J.-J. Huang, C.-C. Liao, P.-J. Hu, and Y. Chang, “Operation lifetimes of organic light-emitting devices with different layer structures,” Chemical Physics Letters, vol. 402, pp. 335–339, 2005. View at: Publisher Site | Google Scholar
  24. J. R. Gong, L. J. Wan, S. B. Lei, C. L. Bai, X. H. Zhang, and S. T. Lee, “Direct evidence of molecular aggregation and degradation mechanism of organic light-emitting diodes under joule heating: an STM and Photoluminescence study,” The Journal of Physical Chemistry B, vol. 109, pp. 1675–1682, 2005. View at: Publisher Site | Google Scholar
  25. H. Aziz and Z. D. Popovic, “Degradation phenomena in small-molecule organic light-emitting devices,” Chemistry of Materials, vol. 16, pp. 4522–4532, 2004. View at: Publisher Site | Google Scholar

Copyright © 2014 Shun-Wei Liu et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

More related articles

 PDF Download Citation Citation
 Download other formatsMore
 Order printed copiesOrder

Related articles

Article of the Year Award: Outstanding research contributions of 2020, as selected by our Chief Editors. Read the winning articles.