New architecture for accurate characterization of the behavior of individual sub-cells within a tandem organic solar cell

Abstract

The great challenge to improve the performance of a tandem organic solar cell is how we can measure the true behavior of its individual sub-cells. In this work, we develop a new architecture of the tandem solar cell to directly and accurately measure I–V and quantum efficiency of the individual sub-cells by using a very thin, semi-transparent metal interlayer with different surface modification as anode and cathode. Separate measurements of reference cells are used in the conventional method to evaluate the contribution of every individual sub-cell in a tandem structure. The discrepancy between the summations of two individual sub-cells in a tandem device and the sum of conventional two reference cells is significant. This great discrepancy is mainly ascribed to the light trapping effect from the rough silver interlayer connection. This work demonstrates that direct measuring of individual sub-cells in a tandem structure is the only way to evaluate the true contribution of each sub-cell, thus having important impact on the study of fundamentals and on the improvement of tandem solar cells.

---

1. Introduction

In organic solar cells, exciton dissociation at the donor–acceptor (D–A) interface is believed to be the main source of photocurrent.^1,2 Due to the short exciton diffusion length of organic materials, a very thin (a few tens of nanometers) active layer is used in organic solar cells to fully utilize the excitons produced by light excitation. The absorption of light in multiple thin layers of an organic solar cell, especially the absorption within the exciton diffusion length away from the D–A interface, can be enhanced due to the interference effect.^2,3 To further improve the power conversion efficiency of an organic solar cell, more photons absorbed within the exciton diffusion length away from D–A interface are desired. One way to achieve this goal is to improve the spectral coverage by blending active materials with a different absorption range in one single cell.⁴ The tandem organic solar cell,^5–15 stacking more than one single-cell in series or parallel, on the other hand, acts as a spectrum splitting device to expand the spectral coverage by the preferential absorption of light in its front sub-cell (upon which the light is incident) and back sub-cell. The characterization of tandem organic solar cells is generally performed by measuring the reference cells with the same structure of each sub-cell and then comparing the summation with the tandem cell. Such a method assumes that the connection of the sub-cells would not change their individual performance, which is considered to be an improper assumption in amorphous silicon tandem cells.¹⁶ Measuring the actual performance of reference cells in a tandem device also assumes that light trapping, interference effect and other effects can be ignored. This was proven to be incorrect in this paper. It is obvious that measuring the individual contribution of each sub-cell within the tandem devices is better than measuring the reference cells. To do so, the connection between sub-cell components must act as a cathode or anode for the front sub-cell and anode or cathode for the back sub-cell for the measurement. The widely used metal cluster^6–14 as the recombination and connection interlayer can not be used for characterizing the sub-cells separately within the tandem organic solar cells. A quite thick connection interlayer other than a metal cluster is thus required when characterizing sub-cells within the tandem structure. Thick gold or conducting oxide connection interlayer has been tried in several studies, but there are no detailed reports of separate characterizations of the sub-cells within the same device.^17–21 A tandem structure with almost absorption-free inter-connection layers is reported for measuring the EQE of individual sub-cell,⁵ however the measurement using 530-nm blue light and 730-nm red light to selectively excite the front and back sub-cell, respectively, is tedious and cannot be used to measure the whole tandem cell without extra light bias. In addition, the approach is only demonstrated with serial connections, possibly due to its difficulty in parallel connection. To study the fundamentals and improve the overall device performance, it is of great importance to explore a simple and accurate method for studying the contribution from each individual sub-cell and also from the whole cell in a tandem solar device.

In this paper, a novel architecture of the tandem solar cell was constructed, in which a 15 nm silver layer was used as a connection interlayer to physically separate the front and back sub-cells. The current–voltage (I–V) characterizations of individual sub-cells and tandem reference cells were conducted. Quantum efficiency measurements were also carried out to examine whether the tandem organic solar cell was a simple summation of two reference cells. Light trapping and interference effect are evident in organic solar cell devices, and surface plasma enhanced absorption might also play a role in the device performance. Through separate characterization of sub-cells within the tandem structure, this work explores a simple approach to accurately evaluate the contribution from each individual sub-cell in tandem solar cells.

2. Results and discussions

2.1 Silver connection electrode

Fig. 1 shows the structure of the studied tandem cells. A 15 nm silver interlayer is used to physically separate the front and back sub-cells. The sheet resistance of 15 nm silver on glass is ∼1 Ω/□, which is good enough for an electrical connection. In our study, the 15 nm silver layer is connected to a pre-patterned indium tin oxide (ITO) bar to make the electric measurements more reliable. For a serial connection, the 15 nm silver interlayer acts as a cathode for the front sub-cell and at the same time as an anode for the back sub-cell. When connected in parallel, the 15 nm silver interlayer acts as cathode for both front and back sub-cells.

Fig. 1 Schematic structure of the tandem cell. Two heterojunction cells (front and back cell) are stacked in series or parallel.

Fig. 2a and b shows the surface morphologies of the organic active layer, copper phthalocyanine (CuPc)\fullerene (C₆₀), on ITO and 15 nm silver layer on its top, respectively. The root-mean-square roughness (rms) of 1.9 nm is observed on the organic layer. A 15 nm silver layer deposition on its top makes the rms increase to 2.2 nm while the island-like feature remains the same. This may suggest that after the thermal deposition of 15 nm Ag the surface is much rougher than that of the organic layer. Since the Ag interlayer is very thin, it should be formed by interconnected Ag clusters. Fig. 2c schematically shows the Ag film on the organic layer. Because of the electromagnetic field coupling with the surface plasmons of the protruded silver clusters, the optical properties of the silver layer might differ significantly from its bulk or thin-film.²² The incident light resonances with the surface plasmons of Ag clusters at the Ag\dielectric interface, increasing the absorption or scattering of the organic layer at interface. Additionally, because of the rough morphology and the very low reflective index, the reflection of this thin Ag film would be quite different from bulk materials, especially when the light incident angle is not normal.²³

Fig. 2 AFM images of the surface of (a) organic layer and (b) 15 nm Ag on top of it, (c) scheme of 15 nm Ag film on the organic layer.

Fig. 3 shows the light intensity after ITO\Ag (15 nm) and ITO\CuPc (130 Å)\C₆₀ (220 Å)\Alq₃ (35 Å)\Ag (150 Å) respectively. The intensity monotonously decreases for Ag on the ITO surface from 400 to 800 nm and the averaged 27% light can go through it. The highly absorbed organic layer under the Ag layer does not decrease the transmission, and even increases around 550 nm. The transmission spectrum of Ag on the organic active layer thus shows considerable high transmission in the whole visible range. An average transmission of 50% between 400 and 800 nm makes the thin silver layer electrode on the organic active layer semitransparent, which is much higher than that of the 15 nm Ag layer on ITO. The rough, thin silver layer also shows different reflection properties from the ITO side and the Ag side, respectively (see the digital photos of ITO\CuPc (130 Å)\C₆₀ (220 Å)\Alq₃ (35 Å)\Ag (150 Å) in Fig. S1, ESI).† Light magenta (blue and red light) color is observed from the silver side, just the complement one from the ITO side (dark green). The higher transmission of ITO\CuPc (130 Å)\C₆₀ (220 Å)\Alq₃ (35 Å)\Ag (150 Å) around 550 and 800 nm means that the blue and red light can effectively pass through this structure. The light magenta color from the silver side indicates that when the blue and red light, which permeates through the front sub-cell, is reflected by the thick silver electrode, it would be reflected or scattered back into the back sub-cell. A proper organic stack between two silver electrodes can be used to enhance the output of light in OLEDs.²⁴ Transfer matrix calculation also reveals that the reflections from the ITO side and from the silver side are quite different in the ITO\organic film\Ag (15 nm) structure (see Fig. S2 in the ESI).† The reflection from the silver side of this structure is much larger than that from the ITO side. Transfer matrix calculation is based on flattened layers and is not applicable for our case with 15 nm silver layer on the organic material. Considering the rough morphology of the Ag layer and the high reflective index of the material (C₆₀) beneath it, the light scattered back into the back sub-cell would be even larger.²³ This indicates that the light (mostly blue and red) transmitted into the back sub-cell would be trapped there.

Fig. 3 Light intensity after ITO\Ag (15 nm), ITO\CuPc (130Å)\C60 (220Å)\Alq3 (35Å)\Ag (150Å) and the transmission spectrum of 15 nm Ag on the organic layer.

2.2 Characterizations of reference cells

Since the characterization of the individual sub-cells conducted within a tandem device in this work is different from the convention method by measuring reference cells as reported in literature,^5,7–9 the 15 nm silver layer must have the ability to function as an anode or a cathode. Surface modification is used in this study, in which the cathode is Alq₃ modified Ag and the anode is C₆₀ modified Ag. A number of single cells in the structures of ITO\CuPc(130 Å)\C₆₀(220 Å)\Alq₃ (35 Å)\Ag(500 Å) (device A), ITO\Ag(150 Å)\ C₆₀(13 Å)\CuPc(190 Å)\ C₆₀ (310 Å)\Alq3(35 Å)\Ag(500 Å) (device B), ITO\Ag(150 Å)\Alq₃(23 Å)\C₆₀(310 Å)\CuPc(190 Å)\C₆₀(20 Å)\Ag(500 Å) (device C), and ITO\Ag(150Å)\CuPc(190 Å)\C₆₀ (310 Å)\ Alq₃ (35 Å)\Ag(500 Å) (device D) are constructed to test the effect of the surface modification. The validation of surface modification for changing the work function of the electrode has been proven,^24–27 and also confirmed by our tests with devices A, B, C, and D (Fig. 4a). The introduction of Alq₃ beneath silver in device D produces an observed photovoltaic effect, indicating that Alq₃ modification could lower the work function of Ag. Only 1.3 nm C₆₀ on silver in device B improves both the photocurrent and photovoltage and thus validates the C₆₀ modification of Ag for anode usage. Even in the inverted structure of device C, quite good performance is still obtained because of the surface modification of Ag. The thicknesses of CuPc and C60 in the devices were optimized in our experimental conditions to balance the photocurrent between front and back sub-cells in the serial connected tandem solar cell. In fact, the tandem devices studied in this paper are constructed by connecting devices A and B in series and devices A and C in parallel. Thus devices A, B and C are the reference cells of the studied tandem devices. Fig. 4b shows the external quantum efficiency (EQE) of devices A, B, and C. Interestingly, devices B and C show a different spectral response although the thickness of each layer of both devices is almost the same. The spectral response of device B is mainly located in the red region, while the response of device C is almost uniformly in the whole visible range. The lower EQE of device B from 400 to 650 nm than that of device A could be attributed to the less light transmission into the active layer. However, the EQE enhancement from 700 to 800 nm can only come from the light trapped in device B due to high back-reflection from the rough silver interlayer.²³ As the discussion below shows, the interference effect which plays an important role in the performance of organic solar cells can only explain well the EQE of device A. Thus it might demonstrate that the light trapping abilities in those cells are different.

Fig. 4 (a) I–V curves of devices A, B, C, and D; (b) EQE of devices A, B, and C.

2.3 I–V characterization within the tandem devices

I–V characterizations of the tandem devices are done before and after EQE measurement, and small decrease of the photocurrent after the EOE measurement indicates that the performance of the devices is relatively stable during the short time measurements, indicating harmony of I–V and EQE measurements (see Fig. S3 in the ESI).† The I–V results obtained from serial and parallel tandem devices, and the reference cells (devices A, B, and C) are also shown in Fig. 5. In the serially connected tandem device (Fig. 5a), the total output of the voltage is equal to the sum of that of two sub-cells in the tandem device, but smaller than the summation of the voltages of two reference sub-cells (device A and device B). The current output is equal to the summation of the two sub-cells in parallel, but lower than the current summation of devices A and C. The short circuit currents of the front sub-cells in both serial and parallel connections are smaller than that of device A, but the current of the back sub-cell is larger than that of device B in the serial connection and that of device C in the parallel connection. Additionally, the I–V shape of the back sub-cell in the parallel connection is very different from that of device C. The interference effect is introduced by folding the two sub-cells.²⁸ Because of a very thin connection interlayer used in the tandem structure, the interference effect in the front sub-cells is much less than that of device A, resulting in decrease of light adsorption and then drop of the photocurrent. Due to the low reflective index and roughness of the silver interlayer, the back reflection in the back sub-cell would be prominent.²³ This enhances the light trapping in the back sub-cell, resulting in a larger photocurrent in the back sub-cells compared to that of devices B or C. The thin metal layer used in the tandem device might introduce other effects such as surface plasma enhanced absorption²² on the total device performance. Table 1 summarizes the performance of the serial and parallel tandem cells and their sub-cells. The discrepancy between two sub-cells and the reference cells indicates that a tandem device is not simply piled up from reference cells, which, however, can only be studied using the method developed in this work.

Table 1 Summary of the open-circuit voltage, short-circuit current, fill factor, and power conversion efficiency of a tandem solar cell (in serial and in parallel) and their sub-cells

Device		V oc/mV	J sc/mA cm^−2	FF/%	η/%
In serial	Total	836	6.2	40.4	0.70
	Front sub-cell	483	5.5	50.2	0.44
	Back sub-cell	338	8.7	20.6	0.20
In parallel	Total	481	11.1	38.5	0.68
	Front sub-cell	475	3.9	46.9	0.29
	Back sub-cell	488	6.9	36.1	0.41

---

Fig. 5 I–V curves of the serial (top) and parallel (bottom) tandem cells.

2.4 Quantum efficiency characterization within the tandem devices

Fig. 6a shows that both sub-cells in parallel connection have relatively high EQE in the whole visible range, which of the front sub-cell is much lower than that of the reference cell in device A. This can be explained by a cathode used in the front sub-cell thinner than that of device A, resulting in less reflection and then less interference-enhanced absorption. However, EQE of the back sub-cell is much larger than that of the reference cell in device C. Several batches of devices with the same structure parameters were constructed and examined. The difference between the reference cells and their corresponding sub-cells reported here is much larger than the variation from batch to batch. The light trapping, as discussed previously, may play the most important role in the EQE enhancement. It is observed in Fig. 6a that the dip around 550 nm for the sub-cells is much shallower than that of the corresponding reference cells. As observed in Fig. 3, the transmission of light around 550 nm is higher than other wavelengths. The light trapping makes light around 550 nm transmitted into the back sub-cell and then multi-reflected for significant enhancement of absorption and EQE. With the help of light trapping in the back sub-cell, the EQE dip around 550 nm becomes shallow. The EQE of the serial connection devices and its sub-cells is shown in Fig. 6b. The highest quantum efficiency of the tandem cell is around 600 nm, where the current balance between the front and the back sub-cell is the best, which is in agreement with the observation in ref. 9. In addition, the spectral response of the front sub-cell in the serial device (Fig. 6b) decreases compared to its corresponding reference cell (device A), and decreases more in the red range compared to the front sub-cell in the parallel connection. The EQE decrease of the front sub-cell in the serial connection can also be ascribed to the thin silver connection layer for less interference effect. The response of the back sub-cell is mainly located in the red range of the serial device, similar to device B. However, the response of device B is stronger in the red range and even greater in the blue range in comparison with the back sub-cell. The EQE enhancement of the back sub-cell in the red range and the EQE difference between the back sub-cell in the serial connection and device B are most likely resulted from the light trapping and the different trapping ability, respectively.

Fig. 6 External quantum efficiency of the parallel (top) and serial (bottom) tandem cells.

2.5 Theoretical calculation of the photocurrent spectrum of organic solar cells

Theoretical calculation of the photocurrent was carried out to simulate the experimental data, based on the interference effect in multilayer thin devices and on the assumption that all absorbed photons contributed to the measured short circuit current in the cell. Fig. 7 shows the measured and calculated photocurrent of device A, in which the calculated photocurrent spectrum is in good agreement with the experimental result. This indicates that the interference effect can be used to well explain the observed EQE response. However, the calculated photocurrent spectra of the tandem structures (in series and in parallel), devices B and C have large discrepancy from the measured photocurrent (not shown here). This may suggest that the theoretical calculation of the photocurrent spectrum of tandem structures needs to take account of not only the interference effect, but also of the light trapping discussed above and even other effects, such as surface plasma caused by an interconnected silver interlayer.

Fig. 7 Normalized theoretical calculated and measured photocurrent of the front sub-cell.

3. Experimental

The cells were constructed on pre-cleaned glass substrates coated with transparent conducting ITO. The sheet resistance of ITO is about 12 Ω/□. After solvent-cleaning, the substrate was placed in a chamber for oxygen plasma pretreatment for 5 min. Then the substrate was transferred to a high vacuum chamber (∼1 × 10⁻⁶ Pa) for organic deposition and to another high vacuum chamber (∼1 × 10⁻⁶ Pa) for metal deposition. Since the three chambers were connected, the transportation of the devices after the oxygen plasma pretreatment and mask deposition of the organic films was conducted under high vacuum conditions (∼3 × 10⁻⁵ Pa). The organic solar cells are of 4 mm² area, with 2 mm wide ITO strips over-crossed by 2 mm -wide metal bars. The donor and acceptor materials are CuPc and C₆₀, both in the front and back sub-cells.

The I–V characteristics were measured by Keithley 2420 in the dark or under illumination of an Oriel solar simulator with 300 mW cm⁻² AM1.5 G spectrum. The devices were illuminated through the transparent ITO electrode. EQE measurements were performed without bias illumination with respect to a calibrated silicon diode. The monochromic light was supplied by xenon light illuminating through a Cornerstone monochromator. A chopper was placed after the monochromator and the signal was collected by Merlin lock-in radiometry after amplification by the current preamplifier. All measurements were carried out in air at room temperature without encapsulation. The surface morphology measurements were done by atomic force microscopy (Veeco Metrology Group, Dimension 3100 SPM) using tapping mode.

4. Conclusions

In summary, a simple novel method is developed in this work to directly characterize individual sub-cells within a tandem structure by using a very thin, semi-transparent metal interlayer with different surface modification as anode and cathode, with which the voltage and current outputs of the tandem cell equal the summation of the front and back sub-cell in serial and parallel connection, respectively. Using the conventional characterization approach based on two reference cells, it was observed that the simple summation of measured voltage and photocurrent from two reference cells is larger than that of the tandem devices. Quantum efficiency measurements also show that the tandem organic solar cell is not a simple summation of two reference cells. The discrepancy is caused by light trapping, interference effect and other effects in multiple thin layers with introduction of a thin silver interlayer, thus demonstrating that the measurement of reference cells cannot evaluate the true contribution of each sub-cell in a tandem structure. Clearly, the method demonstrated by our work can directly evaluate the true performance of sub-cells within a tandem solar cell. This could provide a powerful tool for exploring the fundamentals and the improvement of a tandem solar cell from accurate data of its sub-cells.

Acknowledgements

This work is financially supported by Singapore A*STAR under Grant No: 052 117 0031.

References

1. P. Peumans, S. Uchida and S. R. Forrest, Nature, 2003, 425, 158.
2. Q. L. Song, C. M. Li, M. B. Chan-Park, M. Lu, H. Yang and X. Y. Hou, Phys. Rev. Lett., 2007, 98, 176403.
3. Q. L. Song, H. Yang, H. R. Wu and F. Y. Li, J. Lumin., 2006, 119, 142.
4. M. Y. Chan, S. L. Lai, M. K. Fung, C. S. Lee and S. T. Lee, Appl. Phys. Lett., 2007, 90 023504.
5. J. Y. Kim, K. Lee, N. E. Coates, D. Moses, T. Q. Nguyen, M. Dante and A. J. Heeger, Science, 2007, 317, 222.
6. A. Yakimov and S. R. Forrest, Appl. Phys. Lett., 2002, 80, 1667.
7. J. G. Xue, S. Uchida, B. P. Rand and S. R. Forrest, Appl. Phys. Lett., 2004, 85, 5757.
8. J. Drechsel, B. Mannig, F. Kozlowski, M. Pfeiffer, K. Leo and H. Hoppe, Appl. Phys. Lett., 2005, 86, 244102.
9. G. Dennler, H. J. Prall, R. Koeppe, M. Egginger, R. Autengruber and N. S. Sariciftci, Appl. Phys. Lett., 2006, 89 073502.
10. M. Hiramoto, M. Suezaki and M. Yokoyama, Chem. Lett., 1990, 3, 327.
11. V. Shrotriya, E. H. E. Wu, G. Li, Y. Yao and Y. Yang, Appl. Phys. Lett., 2006, 88 064104.
12. K. Triyana, T. Yasuda, K. Fujita and T. Tsutsui, Thin Solid Films, 2005, 477, 198.
13. K. Triyana, T. Yasuda, K. Fujita and T. Tsutsui, Jpn. J. Appl. Phys., 2004, 43, 2352.
14. D. Cheyns, H. Gommans, M. Odijk, J. Poortmans and P. Heremans, Sol. Energy Mater. Sol. Cells, 2007, 91, 399.
15. N. K. Persson and O. Inganas, Sol. Energy Mater. Sol. Cells, 2006, 90, 3491.
16. J. Burdick and T. Glatfelter, Solar Cells, 1986, 18, 301.
17. A. Hadipour, B. de Boer, J. Wildeman, F. B. Kooistra, J. C. Hummelen, M. G. R. Turbiez, M. M. Wienk, R. A. J. Janssen and P. W. M. Blom, Adv. Funct. Mater., 2006, 16, 897.
18. A. Colsmann, J. Junge, C. Kayser and U. Lemmer, Appl. Phys. Lett., 2006, 89, 203506.
19. A. G. F. Janssen, T. Riedl, S. Hamwi, H. H. Johannes and W. Kowalsky, Appl. Phys. Lett., 2007, 91 073519.
20. K. Kawano, N. Ito, T. Nishimori and J. Sakai, Appl. Phys. Lett., 2006, 88 073514.
21. J. Gilot, M. M. Wienk and R. A. J. Janssen, Appl. Phys. Lett., 2007, 90, 143512.
22. B. P. Rand, P. Peumans and S. R. Forrest, J. Appl. Phys., 2004, 96, 7519.
23. J. Krc, F. Smole and M. Topic, Sol. Energy Mater. Sol. Cells, 2005, 86, 537.
24. Y. Q. Li, J. X. Tang, Z. Y. Xie, L. S. Hung and S. S. Lau, Chem. Phys. Lett., 2004, 386, 128.
25. Q. L. Song, F. Y. Li, H. Yang, H. R. Wu, X. Z. Wang, W. Zhou, J. M. Zhao, X. M. Ding, C. H. Huang and X. Y. Hou, Chem. Phys. Lett., 2005, 416, 42.
26. Q. L. Song, C. M. Li, M. L. Wang, X. Y. Sun and X. Y. Hou, Appl. Phys. Lett., 2007, 90 071109.
27. M. L. Wang, Q. L. Song, H. R. Wu, B. F. Ding, X. D. Gao, X. Y. Sun, X. M. Ding and X. Y. Hou, Org. Electron., 2007, 8, 445.
28. K. Tvingstedt, V. Andersson, F. Zhang and O. Inganas, Appl. Phys. Lett., 2007, 91, 123514.

---

Footnote

† Electronic supplementary information (ESI) available: Digital photos of ITO\CuPc (130 Å)\C₆₀ (220 Å)\Alq₃ (35 Å)\Ag (150 Å) from a) Ag and b) ITO side on black surface and on white paper (Fig. S1); transfer matrix calculation of reflectivity of ITO\CuPc (130 Å)\C₆₀ (220 Å)\Alq₃ (35 Å)\Ag (150 Å) from Ag and ITO side, respectively (Fig. S2); I–V curves before and after EQE measurements (Fig. S3). See DOI: 10.1039/b805140b

---