Design of charge transporting grids for efficient ITO-free flexible up-scaled organic photovoltaics

Abstract

Successfully transferring the device efficiency of small area organic solar cells (SA-OSCs) to a large scale area is a tough challenge. The charge collecting and transporting grids are demonstrated to be effective at addressing this issue, and are widely used in commercial silicon solar cells. However, appreciable shadow loss (5–10%) can be caused with these grids. Thus, a rational design of the grid structure to reduce this significant shadow loss is highly desired. Here, we show that the significant energy loss on scaling up the OSC area stems from the accumulated current density along the charge transport direction. Accordingly, a rational pattern of shorter and triangular Ag grids is designed to accommodate the accumulated current density, leading to a high efficiency of 6.93% for up-scaled OSCs of 4 cm².

---

Introduction

Converting solar irradiance into electrical power using organic based materials has drawn tremendous attention due to the advantages of low cost, flexibility, and roll to roll processing integrated ease of large area manufacturing with roll to roll processing.^1–7 Recent progress on lab-scale small area organic solar cells (SA-OSCs, device area ∼0.1 cm²)^8–16 has greatly encouraged this potential for future application,^17,18 with power conversion efficiencies (PCEs) over 10%^9,19–21 for both single and double junction structures. However, practical application can only be possible in the case where high efficiency is also obtained on a large scale,^22–24 which unfortunately still lags behind due to the significant energy loss on scaling up the device area.^25–27 To date, large OSCs are built in several or tens of square centimeters. Thus far, according to previous reports, the best monolithic large area OSC (LA-OSC) without charge transporting (CT)²⁸ grids or charge collecting (CC)^29,30 grids (device area >2 cm²) exhibited a PCE of only 3.90%³¹ until a more recent work reported a boost to over 7% on a 4 cm² device sheet with the design of a conductance gradient transparent electrode (TE).³² A module geometry based on serially interconnecting small area sub-cells with finite widths is another approach to scale up OSCs without suffering significant energy loss.^33,34 However, area loss as so-called shadow loss is inevitable in the modules. To the best of our knowledge, the highest efficiency modules reported were 4.15 cm² with a PCE of 7.5%.³⁰ Furthermore, a solar cell efficiency table reported the best LA-OSC device performances of 9.5% and 8.7% on 25 cm² and 807 cm² sheets, respectively (without detailed device information).³⁵

It is worth noting that for commercially available silicon based solar cells, the integration of the CT or CC-grids is the most important strategy to reduce energy loss while enlarging the photovoltaic device area. Thereafter, besides the gradient conductance TE design, depositing the CT or CC-grids^29,36,37 can also effectively translate the small area device efficiency into the up-scaled device with great potential for practical applications. In these structures, the up-scaled device sheet is divided into several narrow strips by the grids and the photo current transporting distance is shortened, and thus the energy loss in the up-scaled device is alleviated. As a result, up-scaled OSCs on a 25 cm² square have shown the highest PCE of 4.8% with CT-grids.²⁹ However, a negative effect caused by the grid structure is the appreciable shadow induced optical loss. Therefore, rational design of the grid structures to better balance the resistive energy loss and the shadow loss has become an urgent issue to further promote the device performance of LA-OSCs.

To realize a high OSC device performance on an up-scaled area, deep insight into the energy loss mechanisms is required.^25,32 The basic energy loss mechanisms on up-scaled OSCs have been reviewed in previous reports,^{22,29,38–40} where the thermal dissipation energy loss conforming to Joule's law and the bias voltage drop-induced energy loss were shown to be the main origins.^22,29,39 These two main energy loss sources are intimately relating to the resistance of TE³⁹ and the quality of the active layers or interfaces. Most importantly, a basic phenomenon with scaling up the device area of OSCs is the uneven energy loss density distribution, which becomes more and more severe with increasing device area.³² With regard to this, we here studied the device area and charge transporting distance effects on the device performance of up-scaled OSCs. The results showed that photo current accumulation along the flow direction on the TE sheet accounts for the origin of severe energy loss on up-scaling the device area, and that the energy loss is negligible within the first 0.5 or 1 cm, but grows rapidly beyond this scale. This trend implies that the deposition of CT-grids on the initial ∼0.5 cm is not necessary, and a conductance gradient grid structure design, e.g. a triangular geometry as previously used in inorganic solar cells,⁴¹ would be more rational to better balance the resistance-induced energy loss and the shadow loss. Therefore, for the first time, incomplete crossing, triangular CT-grids (Fig. 1) are designed. Compared to traditional full crossing, rectangular CT-grids, such a design of CT-grids can significantly reduce the shadow loss, but deliver a similar effect in reducing the resistance associated energy loss. By integrating the grid pattern onto a top illumination micro-cavity device structure with ultra-thin Ag as the TE,^9,13 an up-scaled device efficiency up to 6.73% is demonstrated on an ITO-free, flexible texture, which shows an improvement of 7% in device performance and a reduction of 65% in shadow loss compared to traditional CT-grids.

Fig. 1 Schematic diagram of the designed up-scaled organic solar cell architectures with CT and CC-grids, and the chemical structures of the photovoltaic materials.

Results and discussion

Fig. 1 shows the device architecture, in which a top-illuminated OSC device with MoO₃ capped ultra-thin Ag as the TE is designed for ITO-free use.⁴² The capping MoO₃ layer is deposited to suppress the reflection energy loss induced by the high reflectivity of the ultra-thin Ag film. Notably, strong micro-cavities can form in the device chamber for efficient light trapping.⁴³ The popular polythieno[3,4-b]thiophene-alt-benzodithiophene co-polymer (PTB7⁴⁴ or PTB7-Th²¹):[6,6]-phenyl-C₇₁-butyric acid methyl ester (PC₇₁BM) blend was used as the active layer, sandwiched between a poly[(9,9-bis(3-(N,N-dimethylamino)-propyl)-2,7-fluorene)-alt-2,7-(9,9-dioctylflorene)] (PFN)⁴⁵ electron transporting layer and a MoO₃ hole transporting layer. The molecular structures of PTB7, PTB7-Th, PC₇₁BM, and PFN are shown in Fig. 1.

The 12 nm ultra-thin Ag based single junction micro-cavity OSCs were fabricated and compared with the ITO based counterparts, denoted as structure B and A, respectively ((A) glass/ITO/ZnO/active layer/MoO₃/Ag (100 nm); (B) glass/Ag (100 nm)/PFN/active layer/MoO₃/Ag (12 nm)/MoO₃ (40 nm)). Fig. 2a shows the I–V characteristic curves with the A and B device structures for the SA-OSCs (0.052 cm²). The corresponding device parameters are summarized in Table 1. As can be seen, device A shows the highest PCE of 8.21%, with a V_OC of 0.74 V, FF of 0.70, and J_SC of 15.75 mA cm⁻², which is similar to literature reports.⁴⁵ We notice that device B exhibits a marginal drop in PCE even with the ITO-free texture compared to device A, with a best PCE of 8.09%, V_OC of 0.74 V, FF of 0.73, and J_SC of 14.96 mA cm⁻². The comparable J_SCs are consistent with the optical simulation, where the simulated J_SC is 18.62 mA cm⁻² for device structure A and 18.24 mA cm⁻² for structure B. This confirms the efficacy of the optical micro-cavity strategy with ultra-thin Ag TE for light trapping. Fig. 2b shows the external quantum efficiency (EQE) or the incident photon-to-electron (IPCE) spectra of structures A and B. Although the PTB7:PC₇₁BM active layer absorbing profiles of devices A and B are similar, device B shows a significant enhancement in the band tail absorbing region compared to device A. This is consistent with the simulated light absorbing capacity of the PTB7:PC₇₁BM active layer in each device, or the simulated EQE spectra shown in Fig. S1 (ESI†) (assuming 100% internal quantum efficiency). As the absorption coefficient of the PTB7:PC₇₁BM film is weak at the band tails, the significant enhancement in the band tail photon to electron response of device B is attributed to the strong micro-cavity effect, which is confirmed in the optical field intensity distributions shown in Fig. 2c and d for devices A and B, respectively.^13,31,43 These results suggest the ultra-thin Ag based optical micro-cavity structure has potential for efficient light trapping and practical applications. Notably, device B shows advantages over device A in terms of its ITO-free and top-illumination structure, and shows greater potential for low cost production. Thereafter, based on device B, we enlarged the device area to 4 cm².

Fig. 2 (a) I–V characteristics of ITO and ultra-thin Ag based organic solar cells, corresponding to device structure A (glass/ITO/ZnO/PTB7:PC₇₁BM/MoO₃/Ag) and B (glass/Ag (100 nm)/PFN/PTB7:PC₇₁BM/MoO₃/Ag (12 nm)/MoO₃ (40 nm)), respectively. (b) EQE spectra of the PTB7:PC₇₁BM active layer-based devices A and B. Simulated optical field intensity distribution in (c) device A and (d) device B.

Table 1 Device parameters of the studied organic solar cells with different TEs (ITO or ultra-thin Ag), different device areas, and different grid structures

TE	Area (cm^2)	Grids	J SC (mA cm^−2)	V OC (V)	FF	Efficiency (%)
						Ave.	Best
a With bottom charge collecting grid. b With PTB7-Th:PC71BM as active layer. c With PTB7-Th:PC71BM as active layer on flexible PET substrate.
ITO	0.052	—	15.75	0.74	0.70	7.97 ± 0.09	8.21
Ag-12 nm	0.052	—	14.96	0.74	0.73	7.82 ± 0.11	8.09
	4	None	10.95	0.73	0.43	3.11 ± 0.16	3.44
		b-G^a	11.54	0.73	0.46	3.52 ± 0.31	3.95
		G-1	12.87	0.73	0.58	4.98 ± 0.40	5.56
		G-2	13.12	0.73	0.59	5.08 ± 0.29	5.63
		G-3	13.19	0.73	0.63	5.42 ± 0.35	6.04
		G-4	13.95	0.73	0.62	5.74 ± 0.28	6.27
		G-5	14.21	0.73	0.62	6.02 ± 0.16	6.41
		G-5^b	14.62	0.78	0.61	6.71 ± 0.16	6.93
		G-5^c	14.21	0.78	0.61	6.09 ± 0.36	6.73

---

Fig. 3a shows the I–V characteristic curves of the corresponding up-scaled device based on a PTB7:PC₇₁BM blend, and the device parameters are summarized in Table 1. It is observed that directly enlarging the device area gives a significant device performance drop, with the PCE decreasing from 7.82% to 3.11%, FF from 0.73 to 0.39, and J_SC from 14.96 to 10.95 mA cm⁻². The device performance of the up-scaled OSC retains 40% of its original small area device efficiency, and causes a significant power loss of 60%. The drop in device performance on the device area increasing from 0.052 to 4 cm² is consistent with previous literature reports, and can be attributed to the thermal dissipation energy loss³⁹ and the bias voltage drop-induced excessive charge recombination.^9,22,40 By integrating the bottom CC-grid onto the up-scaled device, the device performance recovers to 3.95% due to the variation in the photo current flow direction on the ultra-thin Ag TE (Fig. S2 in ESI†) and reduced energy loss density at the charge collection region for improved photo-current collection. Nevertheless, the energy loss of the up-scaled OSCs with CC-grids is still quite severe compared with the small area device.

Fig. 3 (a) I–V characteristic curves of device area dependent up-scaled organic solar cells, inset shows the authentic device, (b) charge transporting distance dependent I–V characteristic curves of organic solar cells, inset shows the authentic device picture.

In order to further understand the detailed energy loss mechanism on scaling up the device area, the device area and illumination position (or charge transporting distance) dependent device performance was studied. The device geometries of the OSC with device areas of 1, 2, 3, 4 cm² are shown in the inset of Fig. 3a (the authentic device pictures), and the current flow distance was 0.5, 1, 1.5, and 2 cm for the 1, 2, 3, 4 cm² devices, respectively. The device with a current flow distance of 0.5 cm shows the best device performance of 7.80%, which is comparable to that of the SA-OSC, indicating marginal energy loss. With a distance of 1 cm, the device performance shows some drop to 7.21%. However, with distances of 1.5 and 2 cm, the device efficiency is severely reduced to 5.42% and 3.95%, respectively. To gain insight into the energy loss mechanisms, we calculated the energy loss of thermal dissipation and bias-voltage drop on each pixel of the large sheet according to our previous work.³² As shown, the average energy losses are of 0.15, 0.61, 1.37, and 2.43 mW cm⁻², for the up-scaled devices with charge transporting distances of 0.5 cm, 1 cm, 1.5 cm, and 2 cm, respectively. The bias voltage drop along the photo current flow direction was also calculated, with a maximum bias voltage drop of 0.03, 0.14, 0.30, and 0.54 V for the OSCs with charge transporting distances of 0.5, 1, 1.5, and 2 cm, respectively. The trend in the thermal dissipation energy loss or the bias-voltage drop fit well with the device performance loss on elongating the photo current transporting distance, as shown in Fig. S3 (ESI†).

In addition, we fabricated devices with different illumination positions where the charge transporting distance was varied, but there was no accumulated gradient photo-current flow formation. The device area is defined by the cross area of the anode and cathode, which is 0.2 cm² (dimensions of 2 cm in width and 0.1 cm in length), and the authentic device picture is shown in the inset of Fig. 3b. As the device moves away from the anode CT-grid, the photo current transporting distance on the Ag TE gradually increases from 0 to 2 cm. Fig. 3b shows the illumination position dependent I–V characteristics of the OSCs. It is interesting to find that the device performance shows little difference with the illumination or the device position moving far away from the hole collecting grid, despite the photo current transporting distance significantly increasing from 0 to 2 cm. Fig. S3 (ESI†) shows the energy loss calculation results, where the maximum energy loss density is only 0.36 mW cm⁻², and the maximum bias voltage drop is 0.027 V, even with the longest charge transporting distance of 2 cm. This marginal energy loss density is consistent with the nearly identical device performance. Therefore, we conclude that the origin of the sharp drop in device performance with device area up-scaling is the accumulated photo-current flow density, instead of the elongated charge transporting distance.

An effective method to avoid sharp photo-current flow accumulation is to divide the large solar cell sheet into several narrow strips using the CT-grids. Initially, “traditional” CT-grids were deposited onto the ultra-thin Ag TE as outlined in previous reports.^29,37 Three grid patterns differing in location and number were designed to obtain the optimized conditions.

Grid 1: one bottom CC-grid and two side CT-grids;

Grid 2: one bottom CC-grid and one middle CT-grid;

Grid 3: one bottom CC-grid and two middle CT-grids.

The grid is composed of 800 nm thick Ag with a width of ∼1 mm as used previously.²⁹ Considering that the length of the up-scaled OSC is 20 mm, one single grid causes a 5% shadow loss on the whole TE sheet. The I–V characteristics of the up-scaled devices with different grid structures are shown in Fig. 4a, and the corresponding device parameters are summarized in Table 1. As can be seen, the up-scaled OSC with grid 3 shows the best device performance of 6.04%, with a V_OC of 0.73 V, J_SC of 13.19 mA cm⁻², and FF of 0.63. Compared with the up-scaled OSCs without grids, the up-scaled OSC with grid 3 shows a significant improvement, which can be attributed to the reduced resistive energy loss. Detailed analysis regarding the effect of the three grid patterns on the up-scaled OSC device performance can be found in the ESI† (caption of Fig. S2).

Fig. 4 (a) I–V characteristic curves of the up-scaled organic solar cells with different grid structures, PTB7:PC₇₁BM or PTB7-Th:PC₇₁BM active layers and on different substrates, e.g. glass or PET, and the inset showed the authentic photos of large area organic solar cells with grid pattern 4 and 5. (b) Calculated energy loss density on the charge transporting grids with different geometries.

Furthermore, we calculated the energy loss mechanism with grid 3 according to our previous work. It is observed that the shadow loss of the grid is around 10%, which can be responsible for the drop in J_SC compared to the SA-OSCs. The average energy loss density with the photo current transporting to the grids is calculated to be around 0.29 mW cm⁻², and average energy loss density on the grid is calculated to be around 0.034 mW cm⁻². (The boundary conditions for energy loss calculation are as follow. The R_sheet of the CT-grid is estimated to be ∼0.018 Ohmic per square, the R_sheet of the 12 nm Ag is ∼10 ± 3 Ohmic per square, and the J_max of the OSC device is 12.0 mA cm⁻².) The calculated energy loss density distribution profile on the CT-grid of grid 3 is depicted in Fig. 4b (Fig. S4 shows the model used for calculation, ESI†). Similar to the gradient current flow density on the transparent sheet without grids, the energy loss density distribution also shows a gradient distribution due to the accumulated current flow density along the grids. By compromising the shadow loss and charge transporting loss, we proposed a triangle structured grid design (grid 4, schematic diagram is shown in Fig. S2 (ESI†), and the authentic device picture is shown in the inset of Fig. 4a). The energy loss density distribution on the CT-grids of grid 4 along the x axis is also shown in Fig. 4b. As can be seen, the energy loss density distribution profiles are altered slightly on replacing the rectangles with triangle-structured grid, which should have a marginal effect on the overall device performance. More importantly, the shadow loss with the triangle grid 4 is significantly reduced by 50% compared to the rectangle structured grid design. Notably, the energy loss density in the initial 0.6 cm is marginal according to Fig. 3a and Fig. S3 (ESI†). Accordingly, we varied the grid geometry design by shortening the grid length (grid 5, structure shown in Fig. S2 (ESI†), and the authentic device picture is shown in the inset of Fig. 4a) to better deal with the compromise between the energy loss and light trapping. With this design, the shadow loss is further reduced by 30% compared to grid 4.

Finally, the up-scaled OSCs with grid structure 4 and 5 were designed and fabricated, with the I–V characteristic curves and device parameters presented and summarized in Fig. 4a and Table 1, respectively. Remarkably, the up-scaled OSCs with grid structures 4 and 5 show maximum device performances of 6.27% and 6.41%, respectively. The appreciable improvement in device performance can be mainly attributed to the reduced shadow loss of 70% compared to the traditional grid structure design (grid 3), while the charge transporting property remains the same. This result verified the efficacy of the designed grid pattern to better balance the charge transport and suppress the shadow loss. By incorporating the PTB7-Th:PC₇₁BM active layer, the up-scaled device with grid structure 5 shows the highest efficiency up to 6.93%. Furthermore, we also built the designed up-scaled OSC with grid structure 5 onto a flexible substrate of polyethylene terephthalate (PET). The PTB7-Th:PC₇₁BM based 4 cm² OSC device with grid structure 5 on PET shows a highest efficiency of 6.73%, retaining ∼97% of that on glass substrates, mainly due to the top-illumination structure that spares the optical loss caused by the bottom substrate. Notably, the PCE value of our up-scaled OSC on a PET substrate is among the highest ever reported in the literature for flexible up-scaled OSCs (device area >4 cm²) without ITO. Due to the simplicity and effectiveness of the proposed grid structure design, this work is of great significance for the practical application of LA-OSCs in the future.

We note that the energy loss will become more severe as the device area is further enlarged due to the increased resistive energy loss across the longer grids. But this loss can be reduced by using girds with a lower sheet resistance and the designed conductance gradient grid structure still exhibits advantages over the traditional rectangular structured grid. We simulated the effectiveness of the designed conductance gradient grid structure on organic solar cells with a larger area of a 10 × 10 cm² square. The device grid structure patterns are shown in Fig. S5 (ESI†). With the calculation assuming an R_sheet of the CT-grid of 0.0017 Ohmic per square and J_max of 12 mA cm⁻², the up-scaled OSC with a traditional rectangular grid pattern shows a total energy loss of 11% (optical loss 10%, resistive energy loss 1%), while the device with a triangular grid pattern shows a total energy loss of 6.5% (optical loss 5%, resistive energy loss 1.5%). Therefore, the design rule of the charge transporting grids conforming to the energy loss gradient also holds true even for large scale devices. In addition, by combining the design principles to popular hexagonal grids, we proposed a novel design, as shown in Fig. S6 (ESI†), where an improved balance between the optical loss and the resistive energy loss can be expected.

Conclusions

In summary, we have demonstrated highly efficient up-scaled organic solar cells with a best PCE of 6.93% on a 4 cm² flexible sheet by carefully designing the charge transporting grids. The high efficiency is mainly attributed to three mechanisms. First, ultra-thin Ag is applied as the TE for ITO-free use and a top-illumination device with a strong micro-cavity effect for light trapping is designed. Second, this top-illumination structure enables ease of integration of the charge transporting grids, which is shown to have a dramatic influence on the device performance of the up-scaled OSCs. Finally, according to the gradient energy loss density, triangular and shorter charge transporting grids are designed to give a better compromise between the resistance-induced energy loss and the shadow loss. Our work should have an impact on the practical applications of OSCs in future.

Acknowledgements

This work was supported by the Major State Basic Research Development Program (2014CB643503), the National Natural Science Foundation of China (Grants 51261130582, 51561145001, and 91233114), and the Postdoctoral Science Foundation of China (2015M570505).

Notes and references

1. F. C. Krebs, N. Espinosa, M. Hösel, R. R. Søndergaard and M. Jørgensen, Adv. Mater., 2014, 26, 29–39.
2. J. Kettle, N. Bristow, T. K. N. Sweet, N. Jenkins, G. A. dos Reis Benatto, M. Jorgensen and F. C. Krebs, Energy Environ. Sci., 2015, 8, 3266–3273.
3. W. Li, K. H. Hendriks, A. Furlan, M. M. Wienk and R. A. Janssen, J. Am. Chem. Soc., 2015, 137, 2231–2234.
4. J. Choi, K.-H. Kim, H. Yu, C. Lee, H. Kang, I. Song, Y. Kim, J. H. Oh and B. J. Kim, Chem. Mater., 2015, 27, 5230–5237.
5. Y. Li, Acc. Chem. Res., 2012, 45, 723–733.
6. C. Cui, W.-Y. Wong and Y. Li, Energy Environ. Sci., 2014, 7, 2276–2284.
7. S. Li, W. Liu, M. Shi, J. Mai, T.-K. Lau, J. Wan, X. Lu, C.-Z. Li and H. Chen, Energy Environ. Sci., 2016, 9, 604–610.
8. S. Li, Z. He, J. Yu, A. Zhong, R. Tang, H. Wu, J. Qin and Z. Li, J. Mater. Chem., 2012, 22, 12523–12531.
9. L. Zuo, C.-Y. Chang, C.-C. Chueh, S. Zhang, H. Li, A. K. Y. Jen and H. Chen, Energy Environ. Sci., 2015, 8, 1712–1718.
10. S. Zhang, L. Ye, W. Zhao, B. Yang, Q. Wang and J. Hou, Sci. China: Chem., 2015, 58, 248–256.
11. X. Yang, W. Liu and H. Chen, Sci. China: Chem., 2015, 58, 210–220.
12. W. Cao and J. Xue, Energy Environ. Sci., 2014, 7, 2123–2144.
13. L. Zuo, C.-C. Chueh, Y.-X. Xu, K.-S. Chen, Y. Zang, C.-Z. Li, H. Chen and A. K. Y. Jen, Adv. Mater., 2014, 26, 6778–6784.
14. Y. Lin, Y. Li and X. Zhan, Chem. Soc. Rev., 2012, 41, 4245–4272.
15. F. Huang, K.-S. Chen, H.-L. Yip, S. K. Hau, O. Acton, Y. Zhang, J. Luo and A. K. Y. Jen, J. Am. Chem. Soc., 2009, 131, 13886–13887.
16. S. Liu, K. Zhang, J. Lu, J. Zhang, H.-L. Yip, F. Huang and Y. Cao, J. Am. Chem. Soc., 2013, 135, 15326–15329.
17. L. Zuo, S. Zhang, S. Dai and H. Chen, RSC Adv., 2015, 5, 49369–49375.
18. W. Lee, C. Lee, H. Yu, D. J. Kim, C. Wang, H. Y. Woo, J. H. Oh and B. J. Kim, Adv. Funct. Mater., 2016, 26, 1543–1553.
19. J. You, L. Dou, K. Yoshimura, T. Kato, K. Ohya, T. Moriarty, K. Emery, C.-C. Chen, J. Gao, G. Li and Y. Yang, Nat. Commun., 2013, 4, 1446.
20. C.-C. Chen, W.-H. Chang, K. Yoshimura, K. Ohya, J. You, J. Gao, Z. Hong and Y. Yang, Adv. Mater., 2014, 26, 5670–5677.
21. J.-D. Chen, C. Cui, Y.-Q. Li, L. Zhou, Q.-D. Ou, C. Li, Y. Li and J.-X. Tang, Adv. Mater., 2015, 27, 1035–1041.
22. Y. Galagan, E. W. C. Coenen, B. Zimmermann, L. H. Slooff, W. J. H. Verhees, S. C. Veenstra, J. M. Kroon, M. Jørgensen, F. C. Krebs and R. Andriessen, Adv. Energy Mater., 2014, 4, 1300498.
23. F. C. Krebs, R. Søndergaard and M. Jørgensen, Sol. Energy Mater. Sol. Cells, 2011, 95, 1348–1353.
24. N. Espinosa, M. Hosel, M. Jorgensen and F. C. Krebs, Energy Environ. Sci., 2014, 7, 855–866.
25. H. Jin, A. Pivrikas, K. H. Lee, M. Aljada, M. Hambsch, P. L. Burn and P. Meredith, Adv. Energy Mater., 2012, 2, 1338–1342.
26. W. Liu, S. Liu, N. K. Zawacka, T. R. Andersen, P. Cheng, L. Fu, M. Chen, W. Fu, E. Bundgaard, M. Jorgensen, X. Zhan, F. C. Krebs and H. Chen, J. Mater. Chem. A, 2014, 2, 19809–19814.
27. Y. Li, L. Mao, F. Tang, Q. Chen, Y. Wang, F. Ye, L. Chen, Y. Li, D. Wu and Z. Cui, Sol. Energy Mater. Sol. Cells, 2015, 143, 354–359.
28. S.-Y. Park, W.-I. Jeong, D.-G. Kim, J.-K. Kim, D. C. Lim, J. H. Kim, J.-J. Kim and J.-W. Kang, Appl. Phys. Lett., 2010, 96, 173301.
29. A. Armin, M. Hambsch, P. Wolfer, H. Jin, J. Li, Z. Shi, P. L. Burn and P. Meredith, Adv. Energy Mater., 2014, 5, 1401221.
30. S. Hong, H. Kang, G. Kim, S. Lee, S. Kim, J.-H. Lee, J. Lee, M. Yi, J. Kim, H. Back, J.-R. Kim and K. Lee, Nat. Commun., 2016, 7, 10279.
31. H. Jin, C. Tao, M. Velusamy, M. Aljada, Y. Zhang, M. Hambsch, P. L. Burn and P. Meredith, Adv. Mater., 2012, 24, 2572–2577.
32. L. Zuo, S. Zhang, H. Li and H. Chen, Adv. Mater., 2015, 27, 6983–6989.
33. H. Kang, S. Hong, H. Back and K. Lee, Adv. Mater., 2014, 26, 1631.
34. L. Lucera, F. Machui, P. Kubis, H. Schmidt, J. Adams, S. Strohm, T. Ahmad, K. Forberich, H.-J. Egelhaaf and C. Brabec, Energy Environ. Sci., 2016, 9, 89–94.
35. M. A. Green, K. Emery, Y. Hishikawa, W. Warta and E. D. Dunlop, Prog. Photovoltaics, 2015, 23, 1–9.
36. L. Mao, Q. Chen, Y. Li, Y. Li, J. Cai, W. Su, S. Bai, Y. Jin, C.-Q. Ma, Z. Cui and L. Chen, Nano Energy, 2014, 10, 259–267.
37. S. Choi, W. J. Potscavage and B. Kippelen, Opt. Express, 2010, 18, A458–A466.
38. A. K. Pandey and J.-M. Nunzi, Appl. Phys. Lett., 2011, 99, 093309.
39. S. Choi, W. J. Potscavage and B. Kippelen, J. Appl. Phys., 2009, 106, 054507.
40. D. Gupta, M. Bag and K. S. Narayan, Appl. Phys. Lett., 2008, 93, 163301.
41. H. B. Serreze, 13th IEEE Photovoltaic Specialists Conference (PVSC), 1978, pp. 609–614.
42. C. J. M. Emmott, A. Urbina and J. Nelson, Sol. Energy Mater. Sol. Cells, 2012, 97, 14–21.
43. N. P. Sergeant, A. Hadipour, B. Niesen, D. Cheyns, P. Heremans, P. Peumans and B. P. Rand, Adv. Mater., 2012, 24, 728–732.
44. Y. Liang, Z. Xu, J. Xia, S.-T. Tsai, Y. Wu, G. Li, C. Ray and L. Yu, Adv. Mater., 2010, 22, E135–E138.
45. Z. He, C. Zhong, S. Su, M. Xu, H. Wu and Y. Cao, Nat. Photonics, 2012, 6, 591–595.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental section, optical simulation, schematic diagram of current flow with grids, and calculated energy loss densities with grid structures. See DOI: 10.1039/c6qm00043f

---