Si nanocorals/PbS quantum dots composited high efficiency c-Si solar cell

Abstract

A composited colloidal Si nanocorals (NCs)/PbS quantum dots (QDs) p–n active layer was demonstrated to have the potential to reduce the surface recombination velocity and further improve the cell efficiency, not only due to the light trapping structure of Si NCs, but the ability of passivation as well as down-conversion in PbS QDs. An appropriate length of Si NCs was acquired first by investigating the interaction of light absorption and effective minority-carrier lifetime. After the integration with PbS QDs, the composited solar cell showed a 30% increase in power conversion efficiency (PCE), compared to its bare Si substrate counterpart. Thus, we believe that the Si NCs and PbS QDs composited structure is promising for the enhancement of the PCE of current Si based photovoltaic devices.

---

1. Introduction

Silicon-based photovoltaic devices occupy over 80% of the solar cell market due to the element’s earth-abundance, and the devices’ high efficiency and maturity in fabrication at large scale. According to Shockley–Queisser theory, the efficiency of the standard crystalline silicon (c-Si) solar cell used today is limited to 31%.¹ Optical, recombination, series resistance, and thermal or quantum losses are the four main losses that lead to the reduction of Si solar cell efficiency.²

Polished Si substrate is a poor absorber that reflects approximately 35% of incident photons. To reduce optical losses, different kinds of surface texturing and antireflection layers (AR) have been carried out on a commercial scale. Such processes have achieved a reflectance of incident photons of <10%. However, UV light is still highly reflected, which accounts for ∼7% of spectral irradiance.^3,4 Moreover, plasma enhanced chemical vapor deposition (PECVD) is used to deposit AR coatings on a large area, but this process is costlier. By contrast, sub-wavelength Si nanostructures can provide remarkable diffuse reflection and are low-cost in fabrication. When the Si nanostructures are decreased to the quantum scale (<7 nm), the greater split of the Fermi levels will probably increase V_oc.⁵ Despite the benefits from light trapping, the high surface recombination velocity caused by surface defects is the main obstacle for the development of the 1-D sub-wavelength structure Si solar cell.^6,7

Crystalline silicon is an indirect-bandgap semiconductor with bandgap of 1.12 eV; absorption of high-energy photons creates hot electrons and holes that cool quickly to the band edges by sequential emission of phonons. Approximately half of the solar energy loss is related to the spectral mismatch, known as thermal or quantum losses. The utility of the whole solar spectrum can be improved by using a certain down-conversion materials, due to the extension of the operating spectral range towards the ultraviolet and emission of lower energy photons. Rare earth-doped luminescent materials have been extensively studied and have been demonstrated as having prominent down-conversion properties in recent decades.^8,9 However, the integration of such materials on Si based solar cells is still far from practical. Apart from these, inorganic semiconductor quantum dots (QDs) also possess down-conversion properties to improve the utility of full solar energy. Inorganic QDs have the possibility of harvesting effectively UV light then emitting via photon down-conversion, due to the absorption of a high-energy photon converted into two photons with lower energies.^10–13 An additional benefit of the QDs comes from the ability to slow down electronic relaxation.^14,15 The extraction of hot carriers before they cool to the band edges will allow high efficiency solar cells. Great interest in QDs has also arisen from the multiple exciton generation (MEG) phenomenon, a process in which absorption of a photon exceeds twice the bandgap energy to produce two or more electron–hole pairs. If all of the energy of the hot carriers were captured, solar-to-electric power conversion efficiencies could be increased, theoretically, to as high as 66%.^16–19 With their prominent photoelectric properties, QDs will open up a way towards breaking the single junction Shockley–Queisser limit of the first and second generation solar cells, thus moving photovoltaics toward the third generation regime.¹⁸

Here, an Si coral-like one-dimensional (1-D) nanostructure was synthesised. We investigated how these Si nanocorals (NCs) affected the photovoltaic performance of a single crystalline p–n homojunction Si solar cell with a pyramid textured surface, and its integration with PbS QDs afterwards. Firstly, the optimum length of 1-D Si NCs was explored. The 1-D light trapping structure successfully decreased the reflectance of UV-VIS-NIR light to <10% and increased the PCE by 17.5%. Subsequently, ∼3 nm zero-dimension (0-D) PbS QDs were deposited on Si NCs by spin-coating. After the Si NCs solar cell is integrated with PbS QDs, the composited solar cell showed a further enhancement of 10.7% in PCE. It means that a total 30% enhancement of PCE is achieved compared to the pyramid textured Si without Si NCs and PbS QDs. Based on the superior photoelectric performance, we demonstrate that the 0-D/1-D structure composited solar cell with the properties of superior light trapping, photo down-conversion and remarkable passivation will have a chance to improve current c-Si based solar cells.

2. Experimental

2.1 Growth of vertically aligned Si NCs

The pyramid textured single-crystalline Si wafers employed in this work were purchased from Changzhou Yijing Optoelectronics Technology Co., Ltd. The thickness of the wafers was 200 μm with a bulk p–n junction; the resistivity of the Si wafer was in the range of 1–10 Ω cm. Vertically aligned Si NCs were prepared using metal-assisted electroless etching (MAE) by two steps.^20–22 Firstly, the pyramid textured silicon substrate was etched in an aqueous solution of silver nitrate (0.002 M) and hydrofluoric acid (0.4 M) for 30 s to deposit a layer of Ag nanoparticle masks. Then, the pyramid textured silicon substrate with Ag nanoparticle masks was immersed into an aqueous solution of hydrogen peroxide (0.7 M) and hydrofluoric acid (0.4 M) for 0, 15 s, 30 s, 45 s, 60 s to grow Si NCs on the n-type surface. The length of the Si NCs was adjusted by changing the etching times. The pyramid-textured Si substrate with Si NCs was then washed with concentrated nitric acid for 5–10 min to completely remove residual silver mask and dried under air. All the MAE processes were operated at room temperature.

2.2 Synthesis of PbS QDs

The PbS QDs were prepared by wet solution phase chemical syntheses with some modifications.²³ In brief, 910 mg of lead acetate was dissolved in 8 mL of oleylamine and reacted under N₂ gas at 150 °C for 40 min to form the Pb–oleylamine complex. Afterwards, 6 mL of oleic acid solution containing 115 mg of sulfur were quickly injected into the above reaction solution. The resulting mixture was heated to 200 °C and kept for 30 min. After the solution was cooled to room temperature, hexane was added to precipitate PbS QDs followed by centrifugation. The solid product dispersed very well in toluene.

2.3 Characterization

A field emission scanning electron microscope (FESEM, Hitachi, Japan, operated at 15 kV) was used to observe the cross-section of the sample. A JEOL JEM-200CX microscope operating at 160 kV in the bright-field mode was used for transmission electron microscopy (TEM), to characterize PbS QDs. X-Ray diffraction (XRD) patterns were recorded by D/MAX-2000 (Rigaku, Japan), using Cu Kα radiation at a scan rate (2θ) of 2° min⁻¹. The absorption and reflection spectra of the specimens were measured using a CARY 500 Scan UV/VIS/NIR spectrophotometer with an integrating sphere (Labsphere) in a wavelength range of 250–1200 nm. The effective recombination minority-carrier lifetimes of the specimens were characterized by a Semilab WT-2000PVN. Photoluminescence spectroscopy was carried out with a VARIAN Cary-Eclipse 500 fluorescence spectrophotometer equipped with a 60 W xenon lamp as the excitation source; the sample was excited by a light beam at 650 nm. The absorption spectra of PbS QDs were characterized by a Beckman Coulter DU 730 UV/VIS/NIR spectrophotometer. The total cell area measured was 9 cm². Chemical vapor deposition (CVD) is employed to evaporate Al as the back-side electrode. The front-side Ag electrode is printed and dried at 120 °C, and went through rapid thermal annealing at 750 °C for 2 s. The J–V characteristics of the solar cells were investigated under illumination with AM1.5G (100 mW cm⁻²) and provided by a Zennium electrochemical workstation (model: Xpot). The light intensity was calibrated with a silicon standard cell.

3. Results and discussion

1-D Si nanocorals are grown on the pyramid textured Si surface to improve the absorptive character. Previous studies of Si sub-wavelength structures are mainly based on p-type Si. The p-type 1-D structure can grown to be micron-sized, then converted to be n-type afterwards.²⁴ However, Si NCs in this length would be harmful in our system, since the thickness of the bulk n-type Si is no more than 1 μm. Fig. 1 shows the surface and cross sectional FESEM images of Si NCs of 20 nm, 80 nm, 120 nm and 200 nm, corresponding to the electroless etching time of 15 s, 30 s, 45 s and 60 s in the aqueous solution of H₂O₂ & HF, respectively. The diameter of Si NCs increases with the increase of etching time. The absorption spectra of Si NCs are shown in Fig. 2a. It is found that the absorbance at 250–1100 nm was reinforced rapidly with the increase of the Si NCs’ length. The absorption spectrum of pyramid textured Si without NCs is displayed as line a. Even though the absorption spectrum was broadened when the polished Si was pyramid textured, the absorbance in the region of 250–550 nm is still low. Interestingly, as an effect of Si NCs, the absorbance between 250 and 550 nm is greatly improved. This is because high-energy photons have a larger absorption coefficient on the silicon surface than low-energy photons; absorption depth is given by the inverse of the absorption coefficient, or α⁻¹. Presumably, high-energy photons tend to have a shallower absorption depth.⁷ As the superficial area on Si NCs increases, more high-energy photons are absorbed, leading to the expansion of the absorption spectrum to UV region.

Fig. 1 Surface and cross sectional FESEM images of silicon nanocorals: (a) 20 nm; (b) 80 nm; (c) 120 nm; (d) 200 nm.

Fig. 2 (a) Absorption spectra of Si NCs solar cells. (b) Effect of the length of Si NCs on effective recombination minority-carrier lifetime and optical absorption spectra of Si NCs solar cells. (c) J–V characteristics of pyramid textured Si without NCs, pyramid textured Si with 80 nm NCs, and the PbS QDs/Si NCs composited solar cell. (d) Effect of the length of Si NCs on J_sc, V_oc, FF, and PCE of Si NCs solar cells.

Regardless of the light trapping benefits from the 1-D Si NCs, the surface recombination velocity will at the same time have a negative effect on J_sc. In general, recombination loss depends on the surface defects of Si NCs that restrict the cell conversion efficiency. Si NCs array with high aspect ratios have a large surface recombination velocity, leading to poor charge carrier collection efficiency. The effective minority-carrier recombination lifetime and absorbance as a function of the length of Si NCs are shown in Fig. 2b. The absorption increase is directly proportional to the length of Si NCs. Meanwhile, the recombination lifetime decreases rapidly with the length increase of Si NCs. This will probably decrease J_sc, indicating that the recombination lifetime of minority-carriers is inversely proportional to the surface recombination velocity, and J_sc is square-root proportional to the minority-carrier recombination lifetime. Due to the negative correlation displayed between light absorption and minority-carrier recombination lifetime, J_sc would peak at a specific length of Si NCs.

The photoelectric performances of Si solar cells with Si NCs of different lengths (0, 20, 80, 120, 200 nm) are investigated under the illumination of AM1.5G. The photovoltaic performance indicated by J_sc, V_oc, FF, and PCE as a function of Si NCs length is estimated and plotted in Fig. 2d. The largest J_sc value (30.07 mA cm⁻²) is achieved by an Si NCs solar cell with 80 nm long Si NCs, which is a 14.7% enhancement, compared to J_sc value of the pyramid textured surface without Si NCs (26.22 mA cm⁻²). The J–V characteristics of an Si solar cell with 80-nm NCs are shown in Fig. 2c. Even though the absorbance of 80-nm-long Si NCs is 45% stronger than that of the non Si NCs counterpart, the absorption enhancement occur mostly in the region of 250–550 nm, where the photon energy is lost quickly due to surface recombination. Thus, we believe that the optical benefit is counteracted by the enhanced recombination velocity to a certain extent. When the length of Si NCs comes to 200 nm, J_sc rapidly decreases to 23.25 mA cm⁻². This indicates that the enhanced recombination velocity dominates over the optical benefit, leading to an inferior J_sc. The open-circuit voltage (V_oc) exhibits negligible change. The Si NCs in our work have exceeded the Si Bohr radius, so they do not change the Fermi energy distribution in the p–n junction of the Si solar cell. Similarly, an negligible change of FF indicates that Si NCs on the top of the device do not affect the series resistance. In this case, the PCE would be mainly influenced by J_sc rather than V_oc or FF since the J_sc of the Si NCs is influenced directly by the interaction of the minority carrier lifetime and light absorption. As displayed by the curves in the graph, PCE has an analogous changing trend with J_sc, peaking at 80 nm. It increases from 9.49% to 11.15%, indicating that the enhanced absorbance dominates greatly over surface recombination.

Based on the considerable enhancement of J_sc and PCE as Si NCs are employed on the pyramid textured surface, the PCE still retains great potential to be improved further, which lies in the reduction of the surface recombination velocity. Thus, we studied the influence of PbS QDs integrated on the Si NCs solar cell. Lead sulfide has a large Bohr radius (18 nm) with a certain down-conversion depending on the size, making it a promising candidate for highly efficient photovoltaic devices. A TEM image (Fig. 4a) shows that the PbS QDs are uniform and mono-disperse with an average diameter of 2–3 nm. PbS QDs are spin-coated on the Si NCs (2000 rpm) and then treated thermally under 200 °C under the protection of N₂ for 7 min, to obtain a close contact between PbS & Si NCs. This process is important to prevent extra series resistance. A schematic illustration of an Si NCs/PbS QDs composited solar cell is shown in Fig. 3. The length of the Si NCs employed is 80 nm. The complex light-trapping structure with rough surface is propitious for the deposition of PbS QDs. As shown in Fig. 4b, the thickness of the Si NCs capped with a PbS layer is about 100 nm, and PbS QDs are well-distributed, indicating that a large amount of PbS QDs are packed into the gaps of the Si NCs and a ∼20 nm PbS QDs layer is on the top-side. The XRD pattern of PbS QDs integrated with Si NCs is shown in Fig. 4c. The peak at a 2θ value of 69.130° can be indexed to the (400) plane of Si. The diffraction peak of Si overwhelmed the diffraction peak of PbS because of the ultra-small amount of PbS, compared to the Si substrate. As shown in Fig. 4c (2), the peak at 2θ values of 25.963°, 30.074°, 43.058°, 50.976°, 53.411° can be indexed to the (111), (200), (220), (311), (222) planes of PbS, indicating that the pure phase of PbS was obtained and no other impurity peaks were detected.

Fig. 3 Schematic illustration of an Si NCs/PbS QDs composited solar cell, based on the pyramid textured mono-Si.

Fig. 4 (a) TEM image of PbS QDs. (b) cross section FESEM image of Si NCs (80 nm) capped with PbS QD. (c) X-Ray diffraction pattern of PbS QDs and the composited PbS QDs/Si NCs solar cell.

Fig. 2c shows the J–V characteristics of pyramid-textured Si without Si NCs, pyramid-textured Si with 80 nm NCs, and the composited solar cell. The photoelectric performance is summarized in Table 1. It is well worth noting that the V_oc displays a slight enhancement after integration with PbS QDs. This is probably due to the additional light trapping of PbS QDs on the top of the device. The generation of electron–hole pairs is increased, leading to the slight change of the Fermi energy distribution in the Si p–n junction.²⁵ Another observation of the J–V characteristic obtained from the slope of the curves indicates that the FF of the composited solar cell stays almost unchanged. It implies that PbS QDs on the topside of the cell does not affect the conductance of electrons in the n⁺ emitter layer and no extra series resistance is produced. The J_sc of PbS/Si NCs increased from 30.07 to 32.76 mA cm⁻². With both the enhancement of V_oc and J_sc, the PCE of the composited solar cell exhibits a 10.7% improvement, compared to its non PbS capped Si NCs counterpart.

Table 1 Photovoltaic performances of three different Si solar cells

	Voc/mV	Jsc/mA cm^−2	FF (%)	PCE (%)	ΔPCE
Pyramid textured Si	0.548	26.22	66.1	9.49	—
Si NCs (80 nm)	0.551	30.07	67.3	11.15	17.5%
PbS/Si NCs (80 nm)	0.566	32.76	66.6	12.34	30.0%

---

In order to understand the underlying mechanism for the further enhanced cell efficiency, we studied the effective recombination minority-carrier lifetime of Si NCs capped with PbS QDs. As shown in Fig. 5a, after integration with PbS QDs, the effective recombination minority-carrier lifetime of Si NCs is improved remarkably, which corresponds to a considerable enhancement of J_sc. This implies that the PbS QDs coating on the top-side acts as a passivation layer. In the previous reports, PbS QDs have the ability to slow down electronic relaxation,²⁶ as the quasi-continuous valence and conduction bands of the bulk semiconductor are discretized in PbS QDs, due to the confinement of charge carriers. The energy spacing between the electronic levels becomes much larger than the highest phonon frequency of the lattice, resulting in a “phonon bottleneck”. Thus, hot carrier relaxation is only possible via slower multi phonon emission.¹⁵ With the ability to slow down electronic relaxation, PbS QDs achieve a fast extraction of hot carriers before they cool to the band edges, which contributes directly to the enhancement of minority-carrier lifetime. In addition, previous reports indicated that the MEG character of PbS QDs would help to generate more electron–hole pairs.²⁷ In macroscopic circumstances, the higher density of hot carriers enhanced the effective recombination minority-carrier lifetime to a certain extent. Typically, hot electron–hole pairs generated in bulk c-Si by high energy photons relax rapidly to the conduction and valence band edge, losing the excess energy by thermalization. With MEG, the excess energy of high energy photons is converted to additional excitons, enhancing the J_sc. Moreover, it is interesting to find that the recombination lifetime difference between Si NCs and Si NCs capped with PbS QDs increases with the increase of Si NCs length. The enhancements by Si NCs capped with PbS QDs are 2.4%, 7.8%, 11.3%, 19.9%, and 41.5% for Si NCs with a length of 0, 20, 80, 120, and 200 nm, respectively. This variation tendency is probably due to the bigger recombination velocity of the longer Si NCs. Even though the recombination lifetimes of 120 nm and 200 nm Si NCs capped with PbS QDs achieved much improvement, their cell efficiencies were 11.76% and 10.52%, respectively, which are lower than that of a PbS QDs/80 nm Si NCs composited cell.

Fig. 5 (a) Effect of the Si NCs length on surface recombination lifetime of an Si NCs solar cell and the composited solar cell. (b) Absorption spectra and photoluminescence spectra of PbS QDs.

To investigate the down-conversion properties of PbS QDs, the absorption and photoluminescence (PL) spectra are shown in Fig. 5b. The absorption peak of PbS QDs is observed at 685 nm. With an bandgap of 1.81 eV, the size of PbS QDs is measured to be 2.5 nm, based on the Potential Morphing Method (PMM), which corresponds well with previous reports.²⁸ The PL spectrum peaks at 866 nm, with a Stokes shift of 181 nm. This down-conversion coating brings a positive effect in two aspects. On the one hand, the higher energy photons tend to produce electron–hole pairs near the surface of the device and the photo generated carriers disappear easily through recombination on Si surface with mass of surface defects. Nevertheless, as the PbS QDs down-conversion layer is on the frontside, the emitted lower energy photons are absorbed in the deeper region, leading to more photons absorbed in the depletion region. With the help of a built-in-electric field, photo-generated electron–hole pairs will be separated immediately and contribute to the J_sc enhancement. On the other, in general, one photon just excites a single electron–hole pair in the bulk c-Si substrate; the photon energy higher than 1.12 eV cools quickly to the band edges by sequential emission of phonons. With the higher energy photons converted into lower ones, less energy is lost. Moreover, since the PbS has absorption in the visible region, which corresponds with the response spectrum of c-Si, the more-utilized visible region will probably provide further contribution to the enhanced J_sc.

4. Conclusion

In summary, Si NCs and PbS QDs have been composited together and have achieved 30% PCE improvement, compared to the non Si NCs & non PbS QDs capped counterpart, with the down-conversion properties of PbS QDs, as well as the enhanced light absorption and minority-carrier lifetime. We demonstrate that this large improvement contains two steps: firstly, the light trapping of Si NCs; and secondly, the benefits of surface passivation and photon down-conversion from PbS QDs. We believe that the combination of an appropriate 1-D Si nanostructure and 0-D QDs will find a promising way to obtain high efficiency c-Si solar cells.

Acknowledgements

This work is supported by the Shanghai Science & Technology Committee (12521102501, 11ZR1426500), the first-class discipline construction planning in Shanghai University, PCSIRT (RT1269), the Program of Shanghai Normal University (DZL124) and the Key Laboratory of Resource Chemistry of the Ministry of Education of China.

References

1. W. Shockley and H. J. Queisser, J. Appl. Phys., 1961, 32, 510–519.
2. M. A. Green, Solar cells: Operating Principles, Technology and System Applications, The University of New South Wales, Sydney, Australia, 1998.
3. C. Y. Huang, D. Y. Wang, C. H. Wang, Y. T. Chen, Y. T. Wang, Y. T. Jiang, Y. J. Yang, C. C. Chen and Y. F. Chen, ACS Nano, 2010, 4, 5849–5854.
4. X. Y. Guo, H. Li, B. Y. Ahn, E. B. Duoss, K. J. Hsia, J. A. Lewis and R. G. Nuzzo, Proc. Natl. Acad. Sci. U. S. A., 2009, 106, 20149–20154.
5. S. Park, E. Cho, D. Song, G. Conibeer and M. A. Green, Sol. Energy Mater. Sol. Cells, 2009, 93, 684–690.
6. E. Garnett and P. D. Yang, Nano Lett., 2010, 10, 1082–1087.
7. K. Q. Peng and S. Lee, Adv. Mater., 2011, 23, 198–215.
8. L. Aarts, B. M. van der Ende and A. Meijerink, J. Appl. Phys., 2009, 106, 023522.
9. R. T. Wegh, H. Donker, E. V. D. van Loef and K. D. Oskam, J. Lumin., 2000, 87–89, 1017–1019.
10. H. G. Zhao, M. Chaker and D. L. Ma, J. Phys. Chem. C, 2009, 113, 6497–6504.
11. H. Choi, J. H. Ko, Y. H. Kim and S. Jeong, J. Am. Chem. Soc., 2013, 135, 5278–5281.
12. S. H. Jin, D. H. Kim, G. H. Jun, S. H. Hong and S. Jeon, ACS Nano, 2013, 7, 1239–1245.
13. O. Kojima, H. Nakatani, T. Kita, O. Wada, K. Akahane and M. Tsuchiya, J. Appl. Phys., 2008, 103, 113504.
14. W. A. Tisdale, K. J. Williams, B. A. Timp, D. J. Norris, E. S. Aydil and X. Y. Zhu, Science, 2010, 328, 1543–1547.
15. A. Pandey and P. Guyot-Sionnest, Science, 2008, 322, 929–932.
16. R. T. Ross and A. J. Nozik, J. Appl. Phys., 1982, 53, 3813.
17. M. C. Beard, J. Phys. Chem. Lett., 2011, 2, 1282–1288.
18. O. E. Semonin, J. M. Luther, S. Choi, H. Y. Chen, J. B. Gao, A. J. Nozik and M. C. Beard, Science, 2011, 334, 1530–1533.
19. D. N. Congreve, J. Y. Lee, N. J. Thompson, E. Hontz, S. R. Yost, P. D. Reusswig, M. E. Bahlke, S. Reineke, T. Van Voorhis and M. A. Baldo, Science, 2013, 340, 334–337.
20. K. Balasundaram, P. K. Mohseni, Y. C. Shuai, D. Y. Zhao, W. D. Zhou and X. L. Li, Appl. Phys. Lett., 2013, 103, 214103.
21. M. Z. Tymieniecki and Z. A. K. Durrani, Appl. Phys. Lett., 2011, 98, 102113.
22. X. Li and P. W. Bohn, Appl. Phys. Lett., 2000, 77, 2572.
23. H. G. Zhao, D. F. Wang, M. Chaker and D. L. Ma, J. Phys. Chem. C, 2009, 113, 6497–6504.
24. K. Q. Peng, Y. Xu, Y. Wu, Y. J. Yan, S. Lee and J. Zhu, Small, 2005, 1, 1062–1067.
25. C. Y. Huang, G. C. Lin, Y. J. Wu, T. Y. Lin, Y. J. Yang and Y. F. Chen, J. Phys. Chem. C, 2011, 115, 13083–13087.
26. E. Istrate, S. Hoogland, V. Sukhovatkin, L. Levina, S. Myrskog, P. Smith and E. H. Sargent, J. Phys. Chem. B, 2008, 112, 2757–2760.
27. Y. Yang, W. Rodriguez-Cordoba and T. Q. Lian, Nano Lett., 2012, 12, 4235–4241.
28. S. Baskoutas, A. F. Terzis and W. Schommers, J. Comput. Theor. Nanosci., 2006, 3, 269–271.

---