interfacial charge trapping in the polymer solar cells and its elimination by solvent annealing

Interfacial charge trapping in the polymer solar cells and its elimination by solvent annealing A... Iwamoto2 1Technical Physics Division, Bhabha Atomic Research Centre, Trombay, Mumbai

Interfacial charge trapping in the polymer solar cells and its elimination by solvent annealing

A K Chauhan, Abhay Gusain, P Jha, P Veerender, S P Koiry, C Sridevi, D K Aswal, S K Gupta, D Taguchi, T Manaka, and M Iwamoto

Citation: AIP Advances 6, 095012 (2016); doi: 10.1063/1.4963014

View online: http://dx.doi.org/10.1063/1.4963014

View Table of Contents: http://aip.scitation.org/toc/adv/6/9

Published by the American Institute of Physics

Interfacial charge trapping in the polymer solar cells

and its elimination by solvent annealing

A K Chauhan,1, aAbhay Gusain,1P Jha,1P Veerender,1S P Koiry,1

C Sridevi,1D K Aswal,1S K Gupta,1D Taguchi,2T Manaka,2

and M Iwamoto2

1Technical Physics Division, Bhabha Atomic Research Centre,

Trombay, Mumbai 400085, India

2Department of Physical Electronics, Tokyo Institute of Technology, O-okayama,

Megro-ku, Tokyo 152-8552, Japan

(Received 8 May 2016; accepted 6 September 2016; published online 14 September 2016)

The PCDTBT:PCBM solar cells were fabricated adopting a tandem layer approach

to investigate the critical issues of charge trapping, radiation absorption, and effi-ciency in polymer solar cells This layered structure was found to be a source of charge trapping which was identified and confirmed by impedance spectroscopy The low efficiency in multilayered structures was related to trapping of photo-generated carriers and low carrier mobility, and thus an increased recombination Solvent annealing of the structures in tetrahydrofuran vapors was found benefi-cial in homogenizing the active layer, dissolving additional interfaces, and elim-ination of charge traps which improved the carrier mobilities and eventually the

device efficiencies © 2016 Author(s) All article content, except where other-wise noted, is licensed under a Creative Commons Attribution (CC BY) license ( http://creativecommons.org/licenses/by/4.0/ ) [http://dx.doi.org/10.1063/1.4963014]

INTRODUCTION

Bulk hetero junction (BHJ) polymer solar cells are of great interest and provide an attractive alternative to produce renewable energy owing to their easy fabrication process, low cost, light weight, flexibility, and short pay-back time.1 3The performance of a BHJ solar cell depends upon two crucial parameters i.e absorption of incident radiation and collection of separated charge carriers Although, the absorption coefficient is material specific but the thickness of active layer can be increased for

a better absorption of light.4The usual method of layer deposition is spin coating which normally renders imperfect morphologies with aggregations for higher thickness That is a major reason for the non-uniform absorption of light, charge trapping, Langevin recombination, poor carrier mobility

of the extracted charges, which result in reduced number of photo-generated charge carriers and fill factor and hence, overall reduced device efficiency Therefore, lower thickness values for active layers have been preferred and optimized.5,6 In view of this, the mechanisms of charge generation and charge trapping are a topic of intense research in most of the third-generation solar cell materials Thus a study was required for the charge trappings at imperfect interfaces in solar cells and their possible remedies

Recently, poly[N-9’-heptadecanyl-2,7-carbazole-alt-5,5(4’,7’,dia-2-thienyl-2’,1’,3’-benzothia-diazole)] (PCDTBT) has emerged as promising donor material for the fabrication of bulk hetero-junction solar cells owing to its low lying highest occupied molecular orbital (HOMO) position which imparts high open circuit potential (VOC) as well as air-stability to devices.7 By the use of PCDTBT and appropriate acceptor phenyl C71-butyricmethyl ester (PCBM), an efficiency > 6% has been achieved8 and further improvement in the efficiency towards theoretical limit is desirable.9 In contrast to P3HT:PCBM based devices, further improvement on PCDTBT based BHJ devices by

a Corresponding Author: Email: akchau@barc.gov.in ; akc.barc@gmail.com Phone: +91-22-25593911

2158-3226/2016/6(9)/095012/9 6, 095012-1 © Author(s) 2016

095012-2 Chauhan et al. AIP Advances 6, 095012 (2016)

either thermal annealing or employing thicker active layer has not been successful.10,11This is due

to amorphous nature of PCDTBT owing to weak π− π interaction, which does not get affected by thermal annealing.12 , 13In addition, this amorphous nature of PCDTBT give rise to intercalated struc-ture when blended with PCBM, which occupy available open spaces between the side chains of the polymer before forming the pure electron transporting PCBM phase.14 This intercalated structure results in reduced π− π interaction in PCDTBT molecules and hinders the phase separation.15 In order to address the issue, solvent additives have been introduced in blending solutions to control the blend intermixing and morphology during the spin-coating process This resulted in reduced number of interfaces for recombination of charges and thereby structural ordering in polymer dom-inating regions, leading to fast charge transport and extraction.16In addition to this, mixed-solvent vapor annealing has also been adopted to inhibit intercalation thereby increasing the probability of charged-transfer state dissociation, as well as lifetime and mobility of the carriers.17

Therefore, an intuitive approach for the layer by layer growth in PCDTBT:PCBM polymer solar cells followed by solvent vapor annealing was attempted for a better understanding of charge trapping and its implications on device efficiency In this paper, we are reporting the results of this approach to understand the charge trapping, radiation absorption, and efficiency of PCDTBT:PCBM solar cells

by a tandem layered deposition of the photo active blend followed by solvent annealing On the basis

of RC equivalent circuit model, this layered structure was shown to be a source of charge trapping which was identified and confirmed by impedance spectroscopy

EXPERIMENTAL

For fabrication of devices, ITO coated glass having 10 Ohm/square resistivity was taken and patterned (into three linear strips size 3 mm x 25 mm separated by ∼4 mm) using chemical treatment

by masking with tape and etching the exposed area in an optimized solution of (HCl+HNO3+H2O) for

30 minutes at room temperature.18The etched substrate was cleaned using doubly distilled deionized water, acetone, iso-propanol and dried by blowing pressurized nitrogen The substrate was then treated with UV-Ozone for 20 minutes prior to deposition of Hole Transport Layer (HTL) 50 nm layer of poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS) (Ossila-PH-1000) was spin casted on the substrate at 2500 rpm followed by its annealing for 2 hrs at 120◦C After annealing the PEDOT:PSS, deposited substrate were transferred inside a glove box (MB 20G, MBraun Inc Germany) where oxygen and moisture were controlled within 0.1 ppm level Each batch of the active layer blend was prepared by mixing of PCDTBT and of PCBM in 1:2 weight ratio in chloroform solvent The mixture was stirred at room temperature using magnetic stirrer for 48 hrs under dark, then temperature was raised to 60◦C for 6 hrs and again stirred for 48 hrs The blend so prepared was filtered using a 0.45 micron filter

The active layer blend was spin casted on already deposited PEDOT:PSS layer at a spinning speed of 2000 rpm in glove box The chloroform was preferred as a solvent over dichloroben-zene due to its high vapor pressure and low boiling point,19 the deposition procedure was quick and the solvent almost entirely evaporated during the spin casting process The films were left for about a minute before another layer was deposited with the same spinning speed and solvent concentration Upto 5 layers were deposited one after another on the substrate Finally, 100 nm Alu-minum electrodes were deposited by thermal evaporation (vacuum: 1x 10-6 Torr) using a shadow mask to yield 3 x 3 array of devices on a single substrate The final device configuration was ITO/PEDOT:PSS/PCDTBT:PC71BM/Al and the area of each cell was 6 mm2 The typical device structure is shown in Fig1b Prior to transferring the device outside glove box, it was sealed using a special epoxy which was cured by UV treatment

The UV–Vis spectra were recorded in the range of 300-1000 nm using a JASCO spectropho-tometer The thickness for each layer was estimated using absorption data and re-confirmed by stylus profilometer In order to investigate the effect of solvent annealing on surface morphologies, AFM images were recorded for single layer, five layer, before and after solvent annealing The morphology

of the blend films was imaged using Atomic Force Microscope (AFM) (Nanonics model MV-4000) For measurement of charge carrier mobilities, the devices were fabricated in field effect transistor (FET) bottom gate top contact geometry For this purpose, commercially available Si substrate having

FIG 1 (a) Chemical structure of PCDTBT and PCBM (b) Typical device structure (c) Bottom Gate Top Contact FET configuration of mobility measurements (d) solvent annealing setup.

500 nm of thermally grown SiO2were taken and cleaned Then 1-5 layers of active material blend were deposited following method as described above The typical schematic is shown in Fig1c The transfer characteristics of the fabricated devices were measured using standard FET measurement setup and are presented in Fig.2 Field effect hole mobility values (in saturation region) were calculated from the transfer characteristics using the following expression:20,21

µ = 2L

WC i

∂√I DS

∂V GS

!2

where Ci is the specific capacitance, W is the channel width and L is the channel length and

∂√IDS/∂VGSis the slope of the

√

IDSvs VGSplot

Impedance spectroscopy is a very useful tool to analyze imperfect interfaces where an small

ac signal is applied to a device and an equivalent circuit is evaluated after analyzing the recorded impedance spectra.22,23 This technique can distinguish between interfacial and bulk contributions

FIG 2 FET Transfer characteristics of single layer and 5-layer PCDTBT:PCBM films.

095012-4 Chauhan et al. AIP Advances 6, 095012 (2016)

because the equivalent circuit is drawn on the basis of device physics.23For the impedance spec-troscopy measurements, the solar cells were kept at open circuit dc bias potential and AC ampli-tude of 10 mV was applied using PARSTAT impedance analyzer The spectra were recorded in

100 mHz to 500 kHz range and analyzed using an equivalent circuit model fitted using ZSimpWin 3.22 software Measurements at open circuit potential are preferred as at this dc bias no current flows

through the device, the slope (∂J/∂V) is maximum, and the JV remain linear for the applied small

perturbation

Some of the films were also subjected to solvent annealing using a set-up shown in Fig.1d The set-up consists of a sample holder inside a lid-covered beaker The devices were kept in a sample holder and exposed to vapors of 1 mL tetrahydrofuran (THF) at room temperature for 30 sec

RESULTS & DISCUSSION

In order to verify the layer by layer deposition of PCDTBT:PCBM blend and thickness estimation, UV-Vis spectra were recorded and presented in Fig.3 The typical spectrum consists of four peaks and

no significant change is observed in any of the peak positions The broad band observed at 400 and

550 nm is attributable to the π–π∗ transition of carbazole moiety and the intra-molecular charge transfer interaction from the carbazole unit to the benzo-thiadiazole unit in PCDTBT, respectively.24

On the other hand, maxima observed at 380 and 476 nm is assigned to PCBM These bands are overlapped in the PCDTBT:PCBM blend Nevertheless, the evolution of 476 nm band may be related

to PCBM rich regions whereas that of 550 nm is related to PCDTBT As expected, the absorbance

is increasing linearly and proportionately with the thickness of active layer, suggesting multilayer deposition of blend films resulted in thicker films and therefore absorbs more amount of light as compared with single layer films The estimated thickness data is presented in Table-Iwhich was also confirmed independently by stylus profilometer The first layer of active material, which was deposited over the hole transport layer of PEDOT:PSS, was found to have ∼70 nm thickness The

FIG 3 Optical Absorption spectra of single and multilayer films.

TABLE I Thickness of different layers estimated using optical absorption and profilometry.

FIG 4 J-V characteristics of (a) different layers devices under 1 sun (b) for single layer device before and after solvent annealing (c) for multilayer device before and after solvent annealing (red plots for solvent annealed films).

thicknesses for next successive layers were observed to in the range of 52-62 nm The total thickness for a five layer device was measured to be ∼290 nm The larger thickness of first layer can be attributed to the surface energy difference of the PEDOT:PSS and PCDTBT:PCBM blend However, the thickness of next layers have been effected due to the deposition using same solvent, as this might tried to dissolve some part of earlier deposited layer

In order to investigate whether the increment in the thickness indeed aids in light harvesting

in BHJ solar cell, the J-V characteristics of various devices were recorded under illumination of 1 sun intensity using a solar simulator The recorded curves are presented in Fig.4and photovoltaic parameters of these devices are summarized in Table-II-a The best efficiency of ∼3.55% obtained for the devices using single layer is comparable to the efficiencies reported for PCDTBT:PCBM BHJ solar cells fabricated under identical processing conditions.25 – 27 It is also inferred from the table that open circuit voltage (VOC) increases and, short circuit current density (JSC) and fill factor (FF) decreases for devices having higher thickness as compared with single layer devices The over-all effect of this is a monotonous decrease in the efficiency of the device with higher thickness

Though JSCis expected to increase with blend layer thickness, we observed a significant decrease

in the JSCfor higher thickness films On the other hand, the VOCis enhanced when the thickness is increased The open circuit voltage is determined by the generation rate of the excitons as well density

of the defect states and there exist a direct relationship between them28: Thus the increased thickness

of these film leads to generation of more charge carriers, which may be one of the reason behind improved VOC In addition, the enhancement in the VOCmay be related to enhanced recombination into the triplet states, resulting in increment in the energy of the charge transfer states29 and/or increment of the chemical capacitance

The one possible reason for reduction in short circuit current for higher thickness devices could

be the obstruction in the movement of charges by the imperfect interfaces which would act as shallow trap for these photo-generated carriers Solvent annealing experiments were performed to study this charge trapping and THF solvent was used for this purpose The photovoltaic parameters of all solvent annealed films are presented in Table II-b The J-V characteristics for two extreme cases

TABLE II Summary of photovoltaic parameters of BHJ solar cells with different layers under input intensity of 100 mW/cm 2

(a) Before Solvent Annealing (b) After Solvent Annealing

J sc V oc FF Efficiency J sc V oc FF Efficiency Thickness (±10 nm) Composition (mA/cm2) (V) (%) (%) (mA/cm2) (V) (%) (%)

095012-6 Chauhan et al. AIP Advances 6, 095012 (2016)

FIG 5 (a) Impedance spectra of BHJ solar cells for 1- 5 layer devices Also (b) having active layer of different thickness and (e) having thick active layer after solvent annealing under illumination of 1 sun intensity Spectra were fitted (solid lines) using equivalent circuit (c) and (d) for thin and thick active layer, respectively.

(1 and 5 layers) are presented in Figs.4b-c It is evident that when the films of higher thickness were solvent annealed in THF vapors, an improvement in all photovoltaic parameters was observed An enhanced device efficiency ∼3.9% of devices from 1.53% is observed for 290 nm thicker device This improvement may be attributed to improved interior morphology of the blend which reduces the recombination and therefore resulting an enhanced JSCand FF and therefore, an overall improvement

in the efficiency

To further understand the improvement of efficiency in higher thickness devices after solvent annealing, we have carried out impedance spectroscopy at 1 sun illumination The Nyquist plots

of devices under illuminated conditions are shown in Fig.5 Devices exhibited a single semi-circle

in the complex plane for single active layer, whereas a second partial arc gradually appeared at high frequencies in devices with higher number of active layers In addition, the intersection of the semicircle with Z’ axis occurs at increased resistance at high frequency and approximately similar resistance value ∼ 3.43 ohms at high frequency The gradual appearance of second semicircle corre-sponding to additional time constant (capacitance) emphasized the charge storing capacities of the multilayer devices at the interface layers We have modeled these data using a simple equivalent circuit, consisting of a resistor (R1) in series with a resistor R2and capacitor CPE1 in parallel, and various parameters derived from this model are represented in Table III In the equivalent circuit, CPE (constant phase element) is used to describe a non-ideal capacitor with a capacitive contribution associated with a quality factor (n), which varies between 0 and 1 As the diameter of semicircle rep-resents shunt resistance of the device,30it is clear from the Table that the five layered devices show a large shunt resistance after solvent annealing in comparison to as deposited 5-layer device, which is a significant improvement The value of resistance R1represents the resistive losses occurring in ITO, PEDOT:PSS whereas the resistance R2in parallel with constant phase element CPE1 is attributable

to recombination process occurring at the interface of donor and acceptor Correspondingly, R2 is

TABLE III Impedance Spectroscopy of different layered devices.

attributable to the bulk resistance of the active layer i.e recombination resistance and is a measure

of whether the photo-generated charge carriers will be efficiently collected at the electrodes or be lost to bimolecular recombination31and the value of CPE, on the other hand related to a chemical capacitance owing to photo generated charge carriers accumulated in the fullerene phase within the active layer and provides information about DOS in fullerene LUMO states (EFn)32:

Cµ= e2 ∂n

∂E Fn = e2 n c

where Cµ, ncand EFnis the chemical capacitance per unit volume, electron concentration, and quasi-Fermi level due to the rise of the electron quasi-Fermi level in the acceptor LUMO Since, quasi-Fermi level of donor HOMO (EFp) remains almost unchanged under 100mW/cm2 illumination, Cµinfluences the final value of VOC in the device This could be another reason for the enhancement of Voc in the multilayer devices

R2 is a resistance to transport of carriers observed at low frequency and could be related to resistive losses occurring at various interfaces As evident by the table, the total resistance value

is increased when the thickness of the active layer is increased, suggesting more resistive losses in thicker active layer Owing to this enhanced bulk resistance, the effective number of collected photo-generated charge carriers are reduced and thereby contributing to decreased short circuit current On the other hand, the value of CPE1 is increased due to more occupancy of photo-generated charge carriers, which effectively raises the Fermi level of PCBM and therefore results in enhanced VOC Other than this, an additional feature that consists of a small arc appears when thicker active layers are used for device fabrication and therefore, two additional elements R3and CPE2 are required to fit these spectra The high frequency resistor R3represents an absolute loss of the photo-generated power owing to this additional resistance, reciprocal of ∂J/∂V increases and therefore FF reduces.33The extra component of capacitance CPE2 has emerged due to trapping (holding) of the photo-generated carriers at the interfaces This fact has been proven as after solvent annealing this extra capacitance has been removed

In order to investigate the effect of solvent annealing on surface morphologies, AFM images were also recorded for single layer, five layer and their respective solvent annealed devices The recorded images are presented in Fig6 (a-d) The RMS roughness of these active layer surfaces

FIG 6 AFM images of PCDTBT:PCBM (a) single layer, (b) single layer after THF treatment, (c) Five layer and (d) Five layer after THF treatment The shown line profile is in nm and the measured RMS roughness for (a) 0.68 nm (b) 0.72 nm for (c) 0.82 nm and for (d) 0.91 nm.

095012-8 Chauhan et al. AIP Advances 6, 095012 (2016)

TABLE IV Hole mobilties in single and five layer device.

(measured area 500 nm x 500 nm) were found to be 0.68 nm and 0.82 nm respectively The roughness was marginally increased after the solvent treatment, the surface was homogenized and number of pinholes have reduced

The inconsistent carrier trapping at the interfaces is expected to have an adverse effect on the carrier mobilites Therefore, to confirm charge trappings, mobility measurements were planned and performed on the single and multilayer devices The estimated mobilities from transfer characteristics

of single and five layer are presented in TableIV It is clear that as number of layer increases from one to five, the hole mobilities were reduced from 1.28x10-4 cm2/Vs to 7.36x10-5 cm2/Vs That may be related to increased capacitance values of multilayer devices which reflects that more charge holding capacities and multiple scatterings at the interfaces When a multiple layer device was given identical THF treatment, a significant enhancement was observed and the numerical values were approximately restore back from from 7.36x10-5cm2/Vs to 1.19x10-4cm2/Vs However, for single layer devices the mobility value was marginally reduced after the treatment Thus, the THF treatment has enhanced the mobilties which confirms that extra capacitance is being observed due to charge holding at the interfaces

CONCLUSION

The PCDTBT:PCBM solar cells were fabricated adopting a layer-by-layer approach to under-stand the core issues of charge trapping, radiation absorption, and efficiency in these solar cells An overall improvement in the efficiency of polymer solar cells was demonstrated by adopting the mixed approach of using thick active layer and solvent annealing Although, using a thicker active layer

of 290 nm as compared to 75 nm thickness allows enhanced light harvesting but it also resulted in

an increased recombination due to presence of traps The impedance spectroscopy confirmed that major problems associated with lower efficiency for higher thickness devices could be charge trap-ping at the interfaces, which may be related to roughening of interior morphology This has been successfully proved by a deliberate creation of imperfect interfaces between successive layers The additional capacitive component observed at low frequencies in impedance spectroscopy could be related to the stored charge carrier and hence enhanced chemical capacitance It has been inferred that

J scis not limited due to external fields but a pronounced recombination process due to restriction in charge movement is hindering a significant fraction of the separated charges to reach respective elec-trodes The solvent annealing of the thick active layers by tetrahydrofuran vapors results elimination

of imperfections and thereby leading to improvement in all photovoltaic parameters and therefore, results in overall efficiency improvement of 3.9% as compared to 1.53% before solvent annealing The efficiency improvement has also been attributed to the suppression of the resistive losses occur-ring at various interfaces of the active layer as supported by impedance spectroscopy and mobility measurements

This work is supported by DST-JSPS Indo-Japan collaborative research project (DST/INT/JSPS/ P-198/2015) granted to Prof S.K Gupta and Prof M Iwamoto

1 M A Brady, G M Su, and M L Chabinyc, Soft Matter7(23), 11065–11077 (2011).

2 A J Heeger, Advanced Materials26(1), 10–28 (2014).

3 G Li, R Zhu, and Y Yang, Nature Photonics6(3), 153–161 (2012).

4 G Namkoong, J Kong, M Samson, I.-W Hwang, and K Lee, Organic Electronics14(1), 74–79 (2013).

5 F Nickel, C Sprau, M F Klein, P Kapetana, N Christ, X Liu, S Klinkhammer, U Lemmer, and A Colsmann,

Solar Energy Materials and Solar Cells104, 18–22 (2012).

6 Y Sun, C J Takacs, S R Cowan, J H Seo, X Gong, A Roy, and A J Heeger, Advanced Materials23(19), 2226–2230 (2011).

7 S Beaupr´e and M Leclerc, Journal of Materials Chemistry A1(37), 11097–11105 (2013).

8 N Blouin, A Michaud, D Gendron, S Wakim, E Blair, R Neagu-Plesu, M Belletete, G Durocher, Y Tao, and M Leclerc,

Journal of the American Chemical Society130(2), 732–742 (2008).

9 M C Scharber, D M¨uhlbacher, M Koppe, P Denk, C Waldauf, A J Heeger, and C J Brabec, Advanced Materials18(6), 789–794 (2006).

10 T M Clarke, J Peet, A Nattestad, N Drolet, G Dennler, C Lungenschmied, M Leclerc, and A J Mozer,

Organic Electronics13(11), 2639–2646 (2012).

11 A Gusain, V Saxena, P Veerender, P Jha, S Koiry, A Chauhan, D Aswal, and S Gupta, AIP Conf Proc (2013).

12 P A Staniec, A J Parnell, A D Dunbar, H Yi, A J Pearson, T Wang, P E Hopkinson, C Kinane, R M Dalgliesh, and

A M Donald, Advanced Energy Materials1(4), 499–504 (2011).

13 D H Wang, K H Park, J H Seo, J Seifter, J H Jeon, J K Kim, J H Park, O O Park, and A J Heeger,

Advanced Energy Materials1(5), 766–770 (2011).

14 N C Cates, R Gysel, J E Dahl, A Sellinger, and M D McGehee, Chemistry of Materials22(11), 3543–3548 (2010).

15 H Yi, S Al-Faifi, A Iraqi, D C Watters, J Kingsley, and D G Lidzey, Journal of Materials Chemistry21(35), 13649–13656 (2011).

16 G Fang, J Liu, Y Fu, B Meng, B Zhang, Z Xie, and L Wang, Organic Electronics13(11), 2733–2740 (2012).

17 J Liu, L Chen, B Gao, X Cao, Y Han, Z Xie, and L Wang, Journal of Materials Chemistry A1(20), 6216–6225 (2013).

18 A Chauhan, A Gusain, P Jha, S Koiry, V Saxena, P Veerender, D Aswal, and S Gupta, Applied Physics Letters104(13),

133901 (2014).

19 S W Tsang, S Chen, and F So, Advanced Materials25(17), 2434–2439 (2013).

20 H Klauk, Chemical Society Reviews39(7), 2643–2666 (2010).

21 P Jha, S Koiry, V Saxena, P Veerender, A Gusain, A Chauhan, A Debnath, D Aswal, and S Gupta, Organic Electronics

14(10), 2635–2644 (2013).

22J R Macdonald and W R Kenan, Impedance Spectroscopy: Emphasizing Solid Materials and Systems (Wiley, 1987).

23 X Chen, D Taguchi, T Shino, T Manaka, and M Iwamoto, Journal of Applied Physics110(7), 074509 (2011).

24 C Scharsich, F S Fischer, K Wilma, R Hildner, S Ludwigs, and A K¨ohler, Journal of Polymer Science Part B: Polymer Physics53(20), 1416–1430 (2015).

25 F Etzold, I A Howard, R Mauer, M Meister, T.-D Kim, K.-S Lee, N S Baek, and F Laquai, Journal of the American Chemical Society133(24), 9469–9479 (2011).

26 B Kim, H R Yeom, M H Yun, J Y Kim, and C Yang, Macromolecules45(21), 8658–8664 (2012).

27 E A Parlak, T A Tumay, N Tore, S¸ Sarıo˘glan, P Kavak and F T¨urksoy, Solar Energy Materials and Solar Cells110, 58–62 (2013).

28 D Di Nuzzo, A Aguirre, M Shahid, V S Gevaerts, S C Meskers, and R A Janssen, Advanced Materials22(38), 4321–4324 (2010).

29 J Bisquert, The Journal of Physical Chemistry B106(2), 325–333 (2002).

30 D Chi, S Qu, Z Wang, and J Wang, Journal of Materials Chemistry C2(22), 4383–4387 (2014).

31 J Bisquert, Physical Chemistry Chemical Physics5(24), 5360–5364 (2003).

32 G Garcia-Belmonte, P P Boix, J Bisquert, M Sessolo, and H J Bolink, Solar Energy Materials and Solar Cells94(2), 366–375 (2010).

33 A Guerrero, S Loser, G Garcia-Belmonte, C J Bruns, J Smith, H Miyauchi, S I Stupp, J Bisquert, and T J Marks,

Physical Chemistry Chemical Physics15(39), 16456–16462 (2013).