Ionically Gated Small Molecule OPV: Interfacial doping of Charge collector and Transport layer
Danila S. Saranin, Abolfazl Mahmoodpoor, Pavel M. Voroshilov, Constantin R. Simovski, Anvar A. Zakhidov
aa r X i v : . [ phy s i c s . a pp - ph ] M a y Ionically Gated Small Molecule OPV:Controlled n-doping of Thick Fullereneacceptor layers
Danila S. Saranin, † Pavel M. Voroshilov, ‡ , ¶ Constantin R. Simovski, ‡ , ¶ and AnvarA. Zakhidov ∗ , † , § † National University of Science and Technology MISiS, Moscow 119049, Russia ‡ ITMO University, Kronverkskiy pr. 49, 197101 St. Petersburg, Russia ¶ Aalto University, School of Electrical Engineering, Department of Electronics andNanoengineering, P.O. Box 15500, 00076 Aalto, Finland § The University of Texas at Dallas, Physics Department and The NanoTech Institute,Richardson 75080, USA
E-mail: [email protected]
Abstract
We demonstrate the controlled n-doping in small molecule organic photovoltaic(OPV) systems by ionic gating of multi-wall carbon nanotube (MWCNT) coated fullerenes:C and C . Such electric double layer charging (EDLC) doping, achieved by ionicliquid (IL) charging, allows tuning the electronic concentration in the acceptor layers,increasing it by orders of magnitude. This leads to decreasing both the series and shuntresistances of OPV and allows to use thick (up to 200 nm) electron transport layers,increasing the durability and stability of OPV. Two stages of OPV enhancement aredescribed, upon increase of gating bias: at small (or even zero) V g the interface between orous transparent MWCNT charge collector with fullerene is improved, becoming anohmic contact. This changes the S-shaped I-V curve and improves the electrons col-lection by a MWCNT turning it into a good cathode. The effect further enhances athigher V g due to raising of Fermi level and lowering of MWCNT work function. Atnext qualitative stage, the acceptor layer becomes n-doped by electron injection fromMWCNT and ions penetration into fullerene. At this step the internal built-in field iscreated within OPV, that helps exciton dissociation and charge separation/transport,increasing further the I sc and the F F (Filling factor). Overall power conversion ef-ficiency (PCE) increases nearly 50 times in classical CuPc/fullerene OPV with bulkheterojunction photoactive layer and MWCNT cathode. Ionic gating of MWCNT-fullerene part of OPV opens a new way to tune the properties of organic devices, basedon controllable and reversible doping and modulation of work function.
Keywords
Carbon nanotubes, Electrodes, Small molecules, Organic solar cells, Fullerene, Doping, Ul-tracapacitors
Introduction
Organic photovoltaics is one of the competing technologies in the modern renewable energysector, which is capable to partially meet the needs of power generation due to many ad-vantages such as low cost, extreme flexibility, lightweight and large-area manufacturing. Since the pioneering work by Ching W. Tang on two-layer OPV cell based on moleculardonor-acceptor structure of CuPc and perylene tetracarboxylic derivative with a PCE ofabout 1%, the significant progress has been achieved in understanding and improvement ofOPV systems, leading to a PCE of over 11% in single-junction and over 13% in tandemOPV small molecule-based devices that are not so far from the commercialization threshold.2 rather quick increase in performance of OPV solar cells has happened mainly due tothe synthesis of better organic materials with higher charge mobility. However, it is notthe only way, and further improvement is still possible since those OPV cells are usuallyundoped bulk heterojunction (BHJ) type structures. If the doping of transport layers can beeasily achieved then additional enhancement of efficiency will be a straightforward success.It is well known that p-i-n OPV solar cells have better performance and are more stabledue to the thicker p-doped hole transport layer (HTL) and the n-doped electron transportlayer (ETL) with low series resistance. In a series of papers, p- and n-type electronicdoping of organic donor (D) and acceptor (A) transport layers has been shown to increase theperformance of OPV cells, and 8-10% efficiency has been achieved in tandems, demonstratinggreat promise of p-i-n organic structures. One of the outcomes of creating true p-i-n structures as opposed to intrinsic, undopedD-A structures (which are commonly and mistakenly called sometimes as p-n diodes), is theadvantage of using thick transport layers with very low series resistances. So C layers of n-doped A(n) ETL as thick as 100 nm has been used, as opposed to usual thickness of intrinsicA(i) layer of 7-10 nm. Such thicker layers allow achieving more durable OPV structureswithout pin-holes. We do not discuss here the obvious advantages of p-i-n structures, suchas ohmic contacts with electrodes (and non-sensitivity to work function of an electrode),better charge separation by built-in electric fields, and many others (well described, e.g., inreviews of Karl Leo team ).Indeed, truly doped p-i-n geometry improves the separation of positive and negativecharges by built-in potentials, and it also decreases series resistance, enhancing PCE. Themost significant progress here is obtained by charge-transfer (CT) type doping the donorlayers (e.g. of phthalocyanine by active acceptor molecules, such as F − TCNQ ). SuchCT-doping as opposed to substitutional doping (as P and B atoms substitution in Si crystaland other inorganic semiconductors) is usually achieved by intercalation of strong CT dopantbetween molecules of a host: so most recently and successfully obtained by co-evaporation3f strong organic molecular donors (such as acrylic orange, or usual Li atoms) into C layers ). However, such CT doping is usually done in low molecular OPV systems using airsensitive dopants in a high vacuum process, which is very sophisticated and expensive,and cannot be used for liquid-based air processing of small molecule BHJs OPV cells, whilethe recent progress with PCE of ∼
10% is namely due to solution processing.Doping with the optimum concentration of electrons or holes can modify the physicalproperties of both A and D layers in organic electronic devices, particularly in OPV andOLED. Therefore, improved ways to achieve carrier doping have been pursued extensively inorganic electronics arena, with metal-intercalation, (as mentioned above Li-intercalation intoC films) is one of the most important techniques for electron doping of organic/inorganicsolids, and has produced not only efficient p-i-n OPV, but also C and FeSe superconduc-tors from insulators and metallic solids. The most successful examples here are metal-intercalated graphite and C superconductors. Such metal or organic molecule (TCNQ,etc.) intercalation has been performed using not only vacuum co-evaporation but also byliquid solvent techniques. Strong donor (e.g., Li, Na or acrylic orange) intercalation candonate electrons to acceptor ETL layer, which shifts the Fermi level upward.Recently, the electric-double-layer charging (EDLC) has attracted significant attention asa new way to control the carrier density at the interface of nanoscale materials, particularlyat the contacts with carbon nanotubes (CNTs) and graphene which can be achievedreversibly and with no change of chemical composition or structure. Various novel physicalproperties such as superconductivity, metal-insulator and ferromagnetism have beenshown. We have demonstrated first that tunable polymeric OPV can be created using ILfor gating type charging by a formation of the EDLC capacitive doping of single-wall carbonnanotube (SWCNT) and MWCNT. The cathode material is another factor that hinders the performance improvement of OPVcells mostly due to its low stability in air conditions and special processing requirements.Therefore, Al or Ag electrodes are less attractive for use in OPV cells. Air-stable CNTs4ave successfully demonstrated their ability to function as a transparent electrode in smallmolecule OPV cell with doped HTL and ETL layers. However, such n-i-p OPV device withn-doped C and CNT anode on the top of p-type HTL have a certain drawback since it wasprepared in high vacuum by co-evaporation doping of C with a very expensive dopant. Thisinspired us to develop a new architecture deprived of this shortcoming. Here, we proposeanother structure that is more advantageous since n-doping of both MWCNTs and fullerenecan be done in ambient conditions with no need for any additional high-priced processing.MWCNTs have initially a value of sheet resistance R sheet exceeding 1000 Ω /sq in the undopedas-synthesized state that is very high for PV applications. The work function of CNTs in ourOPV cell can be substantially modified by EDLC under IL gating, turning them into a goodcathode. In other words, the Fermi level of MWCNTs can be raised up by n-type doping,thereby allowing better electron collection from the active layer of OPV cell, and a value of R sheet can be decreased to the acceptable level below 100 Ω /sq . In an EDLC ionic gatingmethod, the carriers can first accumulate around the extended interface of highly porousnanomaterial, such as CNT, and their concentration and work function can be controlled byan electric field of the gate V g . Such simple reversible and tunable EDLC-doping has neverbeen previously used in small molecule OPV cells.So motivation of our present paper is to create a small molecule OPV device in which n-doping can be easily achieved in a system of porous CNT electrode coated on top of fullerenefilm ETL of a most simple two-layer OPV: CuPc/fullerene with methods of reversible ionicEDLC. This requires new architecture, which combines IL supercapacitor, connected inparallel with OPV as we introduced earlier for a polymeric OPV with IL, and demonstratedits advantages in tandems. In this paper, we apply the gating in IL to classical small molecule OPV with DEME − BF IL and study the dynamics of this ionically gated system of MWCNT@Fullerene (i.e. cath-ode@ETL) to understand how it accumulates electrons, and how the ions move first inMWCNT network and then into and through fullerene molecular ETL layer of different5hicknesses. We show here that this process differs significantly from the case of polymericOPV, studied in our earlier paper, since in a thick molecular fullerene film (with moleculesbonded by Van der Waals forces) there are clear three stages of charging and n-doping ofseparately MWCNT and fullerene subsystems. Experimental
Small molecule OPV cell was prepared in the multi-source resistive glove box integrated high-vacuum system (Angstrom Engineering Inc., Canada) by the sequential thermal evaporationonto UV-ozone threated, patterned ITO-glass substrates of the following layers: a 7 nm thickcopper(II) phthalocyanine (from H.W. Sanders Corp) HTL, a 60 nm thick co-evaporatedwith a smooth gradient CuPc:fullerene mixed D:A layer and top fullerene layer of a variablethickness. The following fullerene materials were used: C or C (both >98% from Nano-C)for different devices. The base pressure of the chamber was kept around − mbar duringthe evaporation. Counter-electrode CathodeOPV cellAnode
Cover glass
Gate voltage I-V measurements
Figure 1: View of OPV device with laminated semi-transparent MWCNT cathode andMWCNT counter-electrode soaked by ionic liquid (left) and measurement setup and electricalconnection scheme for determination of solar cell parameters when gate voltage V g applied(right).MWCNT forest was produced by a chemical vapor deposition (CVD) process. Our OPVdevices contain three electrodes: an anode of the solar cell (ITO under CuPc), a cathode6first MWCNT sheet on top of fullerene ETL) and a counter-electrode (second MWCNTsheet). Both MWCNT cathode and another MWCNT electrode named "counter-electrode"or gate were deposited by manual dry lamination on the top of organic multilayer struc-ture outside the glovebox in ambient conditions. After that, the deposited MWCNTs wereimmersed in liquid hydrofluoroether (HFE) solvent for several seconds to condense tube totube interconnects and therefore improve conductivity and stability of the electrodes. Con-tacts for cathode, counter-electrode, and ITO-anode were created using silver paint. A smallamount of ionic liquid, N, N -Diethyl- N -methyl- N -(2-methoxyethyl) ammonium tetrafluo-roborate, DEME − BF (Kanto Chemical Co. Inc.), was dropped on top of both MWCNTelectrodes. A thin transparent glass cover-slip was placed over the ionic liquid. A generalview of one fabricated OPV device is shown on the left panel of Fig. 1. I on i c li qu i d I T O a nod e C u P c Fu ll e r e n e CN T ca t hod e G l ass s ub s t r a t e M i xe d l aye r CN T c oun t e r e l ec t r od e - Thin transparent coating U gate U OPV x n m Figure 2: Side view of the OPV-CNT-IL test cell. Light shines through the bottom trans-parent conductive oxide cathode (ITO) and into the CuPc/fullerene mixed active layer. Ionscan EDLC charge on a surface of CNT and move further through the porous CNT into thesmall molecule fullerene matrix of OPV, doping adjacent CNT cathode layers, which aretested for different thicknesses up to 200 nm.OPV devices were characterized with an AM 1.5G solar simulator calibrated to one sun(100 mW/cm ), and two LabVIEW controlled Keithley 2400 source measure units (SMU) ina nitrogen glove box. One of these SMUs was used to apply a gate voltage ( V g ) by connecting7 Voltage [V] -6-5-4-3-2-10123 C u rr e n t d e n s i t y [ m A / c m ] w/o ILU g = 0 VU g = 0.25 VU g = 0.5 VU g = 0.85 V C : 5 nm (a) Gate voltage [V]
I [mA/cm sc ]U [V] oc FF [%]PCE [%] (b)
Figure 3: I-V curves of the OPV solar cell with a 5 nm-thick C ETL at different values ofgate voltage under illumination (a) and the extracted parameter for the same solar cell as afunction of gate voltage (b). OPV device shows its best performance at V g = 0.85V.counter-electrode with MWCNT cathode while the second SMU measured I-V parameters ofOPV cells through connection to the anode and cathode of OPV part (see right panel of Fig.1). Application of a bias V g between the MWCNT cathode and MWCNT counter-electrodewould thus produce super-capacitive EDLC while measurements under of I , and V betweenthe ITO anode and MWCNT cathode would result in the photovoltaic I-V curve from whichwe get V oc and I sc response. Such EDLC creates an asymmetry between the anode (ITO)and cathode by the decreased work function of the MWCNTs forming built-in electric fieldthat allows better charge collection. Results and discussions
Here we show the I-V measurement results of several OPV devices differing by fullerene’stype (C or C ) and ETL thicknesses (from 5 nm to 200 nm). We have found that thesolar cell’s I-V parameters improved significantly even without any bias voltage betweenthe counter-electrode and cathode just after applying the IL to the MWCNTs. In series ofFigs. 3, 4, 6 we demonstrate that I-V curves corresponding to V g = 0V after IL application8 lass substrateITO anodeCuPcCuPc:C60C60 (x nm) MWCNT cathodeDEME-BF4
Solar light (a) -0.1 0 0.1 0.2 0.3 0.4 0.5
Voltage [V] -3-2-1012 C u rr e n t d e n s i t y [ m A / c m ] w/o ILU g = 0 VU g = 0.25 VU g = 0.5 VU g = 0.75 V U g = 1 VU g = 1.25 VU g = 1.5 VU g = 1.75 VU g = 2 V C : 40 nm (b) -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 Voltage [V] -2.5-2-1.5-1-0.500.511.522.5 C u rr e n t d e n s i t y [ m A / c m ] w/o ILU g = 0 VU g = 0.25 VU g = 0.5 VU g = 0.75 VU g = 1 V U g = 1.25 VU g = 1.5 VU g = 1.75 VU g = 2 VU g = 2.25 VU g = 2.5 V C : 100 nm (c) -0.1 0 0.1 0.2 0.3 0.4 0.5 Voltage [V] -2.5-2-1.5-1-0.500.511.522.5 C u rr e n t d e n s i t y [ m A / c m ] w/o ILU g = 0 VU g = 0.5 VU g = 0.75 VU g = 1 V U g = 1.25 VU g = 1.5 VU g = 1.75 VU g = 2 VU g = 2.25 V C : 200 nm (d) Figure 4: (a) Cross-section schematics of the OPV device with top C ETL and laminatedMWCNT cathode soaked in DEME − BF IL. (b-d) IV curves at different values of gatevoltage under illumination for 40, 100 and 200 nm thick C ETL.experience a notable shift to higher absolute values of I sc leading to increase in PCE of theOPV cell. I sc increases due to better charge separation by i-n junctions formed in fullereneby EDLC doping.To see the V g gate bias voltage dependence of the photovoltaic effect, we applied differentbias voltages between the counter-electrode and the MWCNT@C60 cathode on ETL partof OPV. In this experiment, we used the charging time of 2 mins. After applying a biasvoltage from 0 to 3V, the dependence of all three parameters of OPV, namely the I sc , V oc and F F were measured. Extracted OPV parameters for two specific devices are summarized9n Tables 1 and 2. Figs. 3, 4, 6 shows the bias voltage dependence of the I-V curves atstep-by-step increased V g for different thicknesses of fullerene layer. Gate voltage [V] I sc [ m A / c m ] x = 40 nm x = 100 nm x = 200 nm Gate voltage [V] U o c [ V ] Gate voltage [V] FF Gate voltage [V] P C E [ % ] Figure 5: Extracted parameter for the OPV solar cell with C ETL of different thicknessas a function of gate voltage.It has been observed that the device with a 5 nm-thick C layer reaches maximal valuesof all the solar cell parameters at the certain gate voltage applied ( V g = 0.85V) and starts todegrade as the increase of V g further continues beyond this threshold due to EDLC possibleformation on the CuPc layer while devices with thicker ETL layers are saturated at muchhigher V g > 2V (Figs. 4, 5). Similar behavior of I-V curves has also been observed forthe OPV devices with another type of fullerene: C . Again, extracted parameters improvewith gate voltage increasing until a certain threshold. The value of this threshold directlydepends on the ETL thickness and corresponds to the case when EDLC occupies the wholewidth of BHJ. The improvement of I-V shape is due to the formation of ohmic contacts10able 1: Output parameters for the ionically gated OPV cell with 40 nm-thick C ETL. V gate ( V ) I SC ( mA/cm ) V oc ( V ) F F P CE (%) R s (Ω · cm ) R sh (Ω · cm ) dry 0.154 0.276 0.145 0.0062 ... ...0 0.479 0.240 0.135 0.016 ... ...0.25 1.193 0.321 0.159 0.061 1222.52 178.740.5 1.805 0.404 0.186 0.136 690.49 192.180.75 2.259 0.442 0.234 0.234 437.81 250.731 2.558 0.449 0.290 0.333 207.37 300.641.25 2.704 0.449 0.328 0.398 115.53 326.041.5 2.784 0.442 0.354 0.436 93.89 341.141.75 2.881 0.442 0.386 0.492 72.13 351.88 I sc , nearlytwice increased V oc and F F and sufficiently increased PCE from the value less than 0.01%to 0.514% (more than 50 times).Lower I sc of our OPV device with MWCNTs comparing to the conventional non-transparentcathodes is caused by the reduced optical absorption in the bulk heterojunction since thickmetals reflect more unabsorbed light back than semi-transparent CNTs that was alreadyinvestigated in details by optical simulations for different transport layer thicknesses in ourprevious work. Another reason for the overall low PCE is the fact that these devices areunoptimized. Since this is a new device one should optimize the OPV thickness and ILvolume. We believe that increasing OPV thickness and decreasing IL volume will increasePCE significantly. Small deviations from the trend can be interpreted as variations causedby manual processing and non-uniformity of materials used from different batches.Let us consider the physical and photo-electrochemical processes in this IL-OPV devicein more detail. In small molecule OPV structure, the photon absorbed in CuPc (D-part) ofBHJ creates an exciton, which dissociates at fullerene interface (acceptor part) by electroninjection to LUMO of fullerene. This electron is further collected via the i-n build-in field11 lass substrateITO anodeCuPcCuPc:C70C70 (x nm)
MWCNT cathodeDEME-BF4
Solar light (a) -0.1 0 0.1 0.2 0.3 0.4 0.5
Voltage [V] -3.5-3-2.5-2-1.5-1-0.500.511.52 C u rr e n t d e n s i t y [ m A / c m ] w/o ILU g = 0 VU g = 0.25 VU g = 0.5 VU g = 0.75 VU g = 1 V U g = 1.25 VU g = 1.5 VU g = 1.75 VU g = 2 VU g = 2.25 VU g = 2.5 V C : 0 nm (b) -0.1 0 0.1 0.2 0.3 0.4 0.5 Voltage [V] -3-2-10123 C u rr e n t d e n s i t y [ m A / c m ] w/o ILU g = 0 VU g = 0.25 VU g = 0.5 VU g = 0.75 VU g = 1 V U g = 1.25 VU g = 1.5 VU g = 1.75 VU g = 2 VU g = 2.25 VU g = 2.5 V C : 100 nm (c) -0.1 0 0.1 0.2 0.3 0.4 0.5 Voltage [V] -2.5-2-1.5-1-0.500.511.522.5 C u rr e n t d e n s i t y [ m A / c m ] w/o ILU g = 0 VU g = 0.25 VU g = 0.5 VU g = 0.75 VU g = 1 V U g = 1.25 VU g = 1.5 VU g = 1.75 VU g = 2 VU g = 2.25 VU g = 2.5 V C : 140 nm (d) Figure 6: (a) Cross-section schematics of the OPV device with top C ETL and laminatedMWCNT cathode soaked in DEME − BF IL. (b-d) IV curves at different values of gatevoltage under illumination for 50, 100 and 140 nm thick C ETL.fullerene(i)-fullerene(n) at the porous MWCNT network cathode of OPV. Positive chargesin the CuPc molecules are collected by ITO anode which charges it positively, generatinginitial V oc . Positively charged ions from ionic chamber redistribute around the negativelycharged MWCNT cathode, creating EDLC on each nanotube or nanotube bundle and sta-bilizing the desired n-doping of the cathode. This doping raises the Fermi level in MWCNTcathode. Moreover, some ions distribute further between the molecules of fullerene filmswhich stabilizes the photogenerated electrons, creating a small n-doped layer around CNTand forming a desired ohmic contact at the interface with the MWCNT cathode. Namely12 Gate voltage [V] I sc [ m A / c m ] x = 50 nm x = 100 nm x = 140 nm Gate voltage [V] U o c [ V ] Gate voltage [V] FF Gate voltage [V] P C E [ % ] Figure 7: Extracted parameter for the OPV solar cell with C ETL of different thicknessas a function of gate voltage.Table 2: Output parameters for the ionically gated OPV cell with 50 nm-thick C ETL. V gate ( V ) I SC ( mA/cm ) C oc ( V ) F F P CE (%) R s (Ω · cm ) R sh (Ω · cm ) dry 0.733 0.254 0.199 0.037 ... ...0 1.614 0.344 0.184 0.102 ... ...0.25 2.071 0.420 0.217 0.188 1197.70 214.110.5 2.467 0.457 0.267 0.301 252.06 311.710.75 2.653 0.457 0.310 0.376 125.41 304.591 2.708 0.457 0.317 0.392 111.18 300.241.25 2.764 0.457 0.323 0.408 107.65 300.401.5 2.783 0.457 0.324 0.412 105.60 288.811.75 2.852 0.457 0.326 0.425 98.97 308.632 2.877 0.457 0.330 0.434 96.74 304.592.25 2.915 0.457 0.335 0.446 94.21 298.882.5 2.969 0.457 0.339 0.461 90.94 313.06 V oc region.The improvement in the solar cell can be observed with the improvement of the current-voltage characteristics. The initial S-shaped I-V curve with low solar cell parameters im-proves into one with significantly increased I sc , F F and V oc , as it was mentioned earlier fordifferent fullerene thicknesses. The device can be viewed as an OPV and supercapacitor con-nected in parallel via a common ion-porous MWCNT cathode. To maintain overall chargeneutrality in the electrolyte, the negative ions (BF ) move towards a MWCNT counter-electrode that is placed at the opposite side of the IL chamber, forming a second EDLC.The formation of the EDLC on the MWCNT counter-electrode is the charging voltage inthe supercapacitor component of the structure. Formation of EDLC is expected to result inthe n-doping of both the MWCNT electrode and the fullerene semiconductor, leading to thecreation of ohmic contacts and n-i build-in junctions within the BHJ of CuPc/fullerene.Summarizing, we have introduced here a concept of the "thick n-doped ETL based"ionically gated small molecule OPV. An electrolyte or ionic liquid in a microchamber isadded on top of the cathode of OPV. Essential for this design is the existence of counter-electrode, which allows ions to diffuse into the ETL molecular layer and also it can betransparent for light (but not necessarily). In our proof of concept studies, we have usedMWCNT as such a counter-electrode. In addition to the filling the earlier criteria, MWCNTcan be easily doped by electrochemical double layer charging, which occurs naturally duringthe operation of the ionic-OPV structure. Ions from the ionic microchamber provide counterions for EDLC charging, thus permitting both n-type doping (stabilized by in EDLC bypositive ions of DEME) or p-type doping of counter-electrode (stabilized by negative ionsin second EDLC). The organic layers adjacent to MWCNT cathode also are n-doped athigher V g in the properly built and operated device, and this doping provides not only abetter ohmic contact between the doped MWCNT cathode and n-doped fullerene organiclayer. This process reconfigures the OPV from undoped D-A type BHJ to n-i-i, or more14orrectly heterojunction of the: A(n)-A(i)-D(i) type OPV with ohmic contacts to electrodesand internal built-in fields, increasing the performance. Conclusions
We have demonstrated an effective tunable small molecule OPV structure with up to 200 nm-thick stable n-doped fullerenes (C and C ) ETL with a MWCNT cathode. This structureshows an improvement of all PV parameters V oc , I sc and F F leading to more than 50-timeincrease in PCE upon optimal ionic gating. Such operation is usually achieved with verythin intrinsic fullerene layer of 7 to 10 nm.
Acknowledgement
This work was partially supported by the Ministry of Education and Science of the RussianFederation (Grant 14.Y26.31.0010) .
References (1) Riede, M.; Mueller, T.; Tress, W.; Schueppel, R.; Leo, K.
Nanotechnology , ,424001.(2) Tang, C. W. Applied Physics Letters , , 183–185.(3) Deng, D.; Zhang, Y.; Zhang, J.; Wang, Z.; Zhu, L.; Fang, J.; Xia, B.; Wang, Z.; Lu, K.;Ma, W.; Wei, Z. Nature communications , , 13740.(4) Heliatek sets new Organic Photovoltaic world record efficiency of 13.2%. 2016; .(5) Sun, S.-S.; Sariciftci, N. S. Organic photovoltaics: mechanisms, materials, and devices ;CRC press, 2005. 156) Brédas, J.-L.; Norton, J. E.; Cornil, J.; Coropceanu, V.
Accounts of chemical research , , 1691–1699.(7) Fujii, A.; Zakhidov, A. A.; Borovkov, V. V.; Ohmori, Y.; Yoshino, K. Japanese journalof applied physics , , L1438.(8) Drechsel, J.; Männig, B.; Kozlowski, F.; Gebeyehu, D.; Werner, A.; Koch, M.; Leo, K.;Pfeiffer, M. Thin Solid Films , , 515–517.(9) Maennig, B. et al. Applied Physics A: Materials Science & Processing , , 1–14.(10) Zakhidov, A.; Yoshino, K. Synthetic metals , , 1299–1302.(11) Yoshino, K.; Tada, K.; Fujii, A.; Conwell, E. M.; Zakhidov, A. A. IEEE transactionson electron devices , , 1315–1324.(12) Siebert-Henze, E.; Lyssenko, V.; Fischer, J.; Tietze, M.; Brueckner, R.; Menke, T.;Leo, K.; Riede, M. Organic Electronics , , 563–568.(13) Meerheim, R.; Körner, C.; Leo, K. Applied Physics Letters , , 125_1.(14) Lüssem, B.; Riede, M.; Leo, K. physica status solidi (a) , , 9–43.(15) Li, F.; Pfeiffer, M.; Werner, A.; Harada, K.; Leo, K.; Hayashi, N.; Seki, K.; Liu, X.;Dang, X.-D. Journal of applied physics , , 023716.(16) others„ et al. Scientific reports , , 18931.(17) Emmenegger, C.; Mauron, P.; Sudan, P.; Wenger, P.; Hermann, V.; Gallay, R.; Züt-tel, A. Journal of Power Sources , , 321–329.(18) Uesugi, E.; Goto, H.; Eguchi, R.; Fujiwara, A.; Kubozono, Y. Scientific reports , . 1619) Ueno, K.; Nakamura, S.; Shimotani, H.; Ohtomo, A.; Kimura, N.; Nojima, T.; Aoki, H.;Iwasa, Y.; Kawasaki, M. Nature materials , , 855–858.(20) Ye, J.; Inoue, S.; Kobayashi, K.; Kasahara, Y.; Yuan, H.; Shimotani, H.; Iwasa, Y. Nature materials , , 125–128.(21) Bollinger, A. T.; Dubuis, G.; Yoon, J.; Pavuna, D.; Misewich, J.; Božović, I. Nature , , 458–460.(22) Yuan, H.; Shimotani, H.; Tsukazaki, A.; Ohtomo, A.; Kawasaki, M.; Iwasa, Y. AdvancedFunctional Materials , , 1046–1053.(23) Yamada, Y.; Ueno, K.; Fukumura, T.; Yuan, H.; Shimotani, H.; Iwasa, Y.; Gu, L.;Tsukimoto, S.; Ikuhara, Y.; Kawasaki, M. Science , , 1065–1067.(24) Cook, A. B.; Yuen, J. D.; Micheli, J. W.; Nasibulin, A. G.; Zakhidov, A. Journal ofvisualized experiments: JoVE ,(25) Cook, A. B.; Yuen, J. D.; Zakhidov, A.
Applied Physics Letters , , 194_1.(26) Kim, Y. H.; Müller-Meskamp, L.; Zakhidov, A. A.; Sachse, C.; Meiss, J.; Bikova, J.;Cook, A.; Zakhidov, A. A.; Leo, K. Solar Energy Materials and Solar Cells , ,244–250.(27) Saranin, D.; Ishteev, A.; Cook, A. B.; Yuen, J. D.; Kuznetsov, D.; Orlova, M.; Di-denko, S.; Zakhidov, A. Journal of Renewable and Sustainable Energy ,9