负责-过渡金属氧化物的作用

www.afm-journal.de

www.MaterialsViews.com

F U L L P A P E

The Role of Transition Metal Oxides in Charge-Generation Layers for Stacked Organic Light-Emitting Diodes

¨ger, Thomas Winkler, Marco Witte, By Sami Hamwi, Jens Meyer, Michael Kro

Thomas Riedl,*Antoine Kahn, and Wolfgang Kowalsky

OLEDs for the consumer market can start. An elegant way to meet this requirement The mechanism of charge generation in transition metal oxide (TMO)-based

using present organic materials and devices

charge-generation layers (CGL)used in stacked organic light-emitting diodes

is to stack a number of OLEDs on top of each

(OLEDs)is reported upon. An interconnecting unit between two vertically other, so as to signiﬁcantlyreduce the stress stacked OLEDs, consisting of an abrupt heterointerface between a Cs 2CO 3-on each light-emitting unit while still

doped 4,7-diphenyl-1,10-phenanthroline layer and a WO 3ﬁlmis investigated. achieving a given luminance level.

Interconnecting units that serve as charge-Minimum thicknesses are determined for these layers to allow for

generation layers (CGL)are required when simultaneous operation of both sub-OLEDs in the stacked device.

driving OLED stacks as two-terminal

Luminance–currentdensity–voltagemeasurements, angular dependent

devices. The ﬁrststudies by Kido et al.

spectral emission characteristics, and optical device simulations lead to suggested that indium tin oxide (ITO)or minimum thicknesses of the n-type doped layer and the TMO layer of 5and tetraﬂuorotetracyanoquinodimethane(F4-TCNQ) adjacent to a hole-transport layer 2.5nm, respectively. Using data on interface energetic determined by

(HTL)may lead to the generation of holes ultraviolet photoelectron and inverse photoemission spectroscopy, it is

and electrons upon application of an electric

shown that the actual charge generation occurs between the WO 3layer and

ﬁeld.[1]Since this early work, various

its neighboring hole-transport material, 4,4’,4’’-tris(N -carbazolyl)-triphenyl concepts for CGL structures have been amine. The role of the adjacent n-type doped electron transport layer is only to published, including junctions between facilitate electron injection from the TMO into the adjacent sub-OLED.

chemically p-and n-doped charge transport

layer, [2–4]the insertion of thin metal or transparent conductive oxide (TCO)

[1,5]

layers, and the insertion of transition metal oxides (TMOs).[6–10]

1. Introduction It was recently shown that the charge-generation mechanism

when using a doped organic p–nheterojunction is based on a

Organic light-emitting diodes (OLEDs)have attracted much temperature-independent ﬁeld-inducedcharge separation sup-interest in research and development in the last two decades. A ported by a large band bending at the interface. This interpretation long operating lifetime must be ensured before mass-production of was evidenced by results from Kelvin probe measurements. [11]On

the other hand, speculation remains about the mechanism

[*]Prof. T. Riedl operating at TMO-based CGLs. Terai et al. proposed a thermal Institute of Electronic Devices stimulation model in which the charge generation is claimed to be University of Wuppertal

thermally assisted. Based on the assumption of impurity levels

Rainer-Gruenter-Str. 21

within the bandgap of V 2O 5, electrons are supposed to diffuse from D-42119Wuppertal (Germany)

the valence band of the TMO to the lowest unoccupied molecular E-mail:[email protected]

orbital (LUMO)of the adjacent n-type doped electron-transport S. Hamwi, T. Winkler, M. Witte, Prof. W. Kowalsky

Technical University of Braunschweig layer (ETL),which is regarded as the charge-generation mechan-Institute of High-Frequency Technology ism. [12]Very recently, Qi et al. suggested a speciﬁcenergy-level Schleinitzstraße22

alignment between lithium-doped 4,7-diphenyl-1,10-phenanthro-D-38106Braunschweig (Germany)

line (BCP)and MoO 3to explain the charge-generation mechan-Dr. J. Meyer, Prof. A. Kahn

ism. The assumption made is that of a thermally assisted tunneling Department of Electrical Engineering

injection of electrons into the ETL and a concomitant hole Princeton University

Princeton, NJ 08544(USA)generation within the TMO layer. [13]In that model, MoO 3is

¨ger Dr. M. Kro assumed to be a p-type semiconductor with the valence band and

InnovationLab GmbH conduction band located at 5.7and 2.3eV below vacuum level Speyerer Straße4

(E vac ), respectively. However, in view of recent reports on the

D-69115Heidelberg (Germany)

electronics structure of MoO 3this model must be revised. [14–17]Based on results obtained by ultraviolet and inverse photoemission DOI:10.1002/adfm.201000301

1762

ß2010WILEY-VCH Verlag GmbH &Co. KGaA, Weinheim Adv. Funct. Mater. 2010, 20, 1762–1766

www.MaterialsViews.com

www.afm-journal.de

spectroscopy (UPSand IPES), MoO 3exhibits a deep lying conduction band at 6.7eV and a high work function of 6.9eV and therefore shows properties of an n-type semiconductor. [14]Similar results have been reported for other TMOs like WO 3exhibiting a similarly high work function, ﬁrstmeasured by Meyer et al. via Kelvin probe technique and conﬁrmedby photoemission spectroscopy. [17,18]Consequently, the CGL model mentioned (2-phenylpyridine)iridium(Ir(ppy)3)-doped TPBi (7vol%)located between the HTL and the ETL. Finally, each OLED unit is capped by Cs 2CO 3-doped BPhen (16wt%)as an efﬁcientelectron-injecting layer (EIL).Indium tin oxide (ITO)and Al are used as bottom and top electrode, respectively. 2.1. Critical Layer Thickness

FULL PAPER

above, in which holes are claimed to be generated within the TMO and subsequently drift towards the HTL, must be revised entirely. As a ﬁrststep to study the heterointerface between both sub-To clarify the TMO-based CGL mechanism, we present a OLEDs, we varied the thickness of the WO 3layer in OLED 2(Fig.1) detailed study of the interconnecting unit placed in a stacked from 0to 5nm and simultaneously changed the thickness of the double OLED structure. By conventional vertical stacking, an neighboring HTL from 45to 40nm, keeping the overall thickness abrupt heterointerface is automatically formed between a Cs 2CO 3-of the stacked OLEDs as well as the distance between the emission doped 4,7-diphenyl-1,10-phenanthroline (BPhen)layer as the layers and the electrodes constant (Table1, series A). The current topmost ETL of the bottom light-emitting unit, and WO 3as efﬁciency(h ) versus luminance (L ) characteristics are shown in the lowermost constituent of the top OLED (Fig.1). To analyze the Fig. 2a. Without WO 3, or for a TMO thickness less than 1.5nm, the functionality of the interconnecting unit, we vary the thickness of stacked OLEDs only exhibit low current efﬁcienciesof about its components and measure the electro-optical properties of the 10cd A À1. This changes abruptly for a thickness of the TMO layer stacked OLEDs. With the help of luminance–currentdensity–of 2.5nm and beyond, for which the current efﬁcienciesjump to voltage (L–I–V) measurements, the analysis of the angular values between 50and 56cd A À1(at1000cd m À2). Furthermore, resolved spectral emission characteristics, and simultaneous we observed that only for devices with WO 3layer thickness larger optical device simulation, we unambiguously identify the than 2nm the onset voltage is less than 5V (notshown here). This minimum required thickness that makes each individual indicates an efﬁcientcharge-generation mechanism, since the constituents of the interconnecting unit fully functional. Using onset voltage is only twice that of each individual sub-OLED of the energetics determined via UPS and IPES for the interfaces similar structure. [18]In this case, no signiﬁcantpotential drop between the interconnecting unit and adjacent charge transport occurs within the CGL to generate and separate the charge carriers. layers, [19]we demonstrate that the actual charge generation The ﬁrstresult of this experiment is therefore that the CGL process takes place at the interface between the thin ﬁlmof WO 3interconnecting architecture is not fully functional in stacked and the neighboring HTL. A large interface dipole between these OLED devices as long as the WO 3layer thickness is below some two layers is found. These results prove former assumptions of a critical value. A further set of experiments leads to a similar thermally assisted charge-generation mechanism within the TMO conclusion regarding the EIL, in that the efﬁciencyof stacked to be invalid. Based on these ﬁndings,we suggest a general design OLEDs is signiﬁcantlylower than 60cd A À1when using rule for CGLs involving TMOs.

insufﬁcientlythick ﬁlmsof Cs 2CO 3-doped BPhen at the CGL heterointerface. The thickness of the n-doped ETL of OLED 1(Fig.1) is varied from 0to 10nm while simultaneously changing 2. Results and Discussion

the thickness of the neighboring layer of TPBi from 50to 40nm and keeping the thickness of WO 3at 5nm as well as the overall The twofold stacked OLEDs were prepared by stacking two green thickness of the stacked OLEDs constant (Table1, series B). light-emitting organic diodes with identical layer sequence on-top Starting at a doped BPhen thickness below 5nm, we obtained a of each other (Fig.1). Each OLED comprises a thin ﬁlmof WO 3current efﬁciencyaround 40cd A À1(Fig.2b). As the layer thickness followed by 4,4’,4’’-tris(N -carbazolyl)-triphenyl amine (TCTA)as increases to 5nm and beyond, the current efﬁciencyagain jumps the HTL and 1,3,5-tris(phenyl-2-benzimidazolyl)-benzene(TPBi)to higher values of around 55to 60cd A À1, indicating full as the ETL. The emission layer is formed by fac tris

functionality of the stacked

device.

Table 1. Layer sequence and thicknesses (givenin nm) of the two OLED series A and B.

Layer

Series A Series B

Al top electrode

OLED 2

BPhen:Cs2CO 32020TPBi

3030TPBi:Ir(ppy)310

10TCTA (45–X) 40WO 3

X 5OLED 1

BPhen:Cs

2CO 320Y

TPBi

30(50–Y) TPBi:Ir(ppy)31010TCTA 4040WO 3

5Figure 1. Layer sequence of the twofold stacked OLED.

ITO bottom electrode

Adv. Funct. Mater. 2010, 20, 1762–1766ß2010WILEY-VCH Verlag GmbH &Co. KGaA, Weinheim

1763

www.afm-journal.de

www.MaterialsViews.com

F U L L P A P E R

0.690.66

C I E Y

0.63

0.600.57

CIE X

0.650.64

C I E Y

0.63

0.620.61

CIE X

Figure 2. h versus L characteristics of the twofold stacked OLEDs upon a) variation of the WO 3layer thickness with constant thickness of the BPhen:Cs2CO 3component (seriesA) and b) variation of BPhen:Cs2CO 3thickness with constant thickness of WO 3(seriesB).

Figure 3. CIE characteristics of angular resolved EL spectra obtained by measurement and optical device simulation for a) series A and b) series B.

To clarify the nature of an incomplete interconnecting unit and its impact on the operation of the stacked OLEDs, we conducted angular-dependent measurements of the electroluminescence (EL)spectra and compared the results with optical device simulation. For a better illustration, the EL spectra are converted to CIE (CommissionInternationale de l’Eclairage)coordinates and summarized within respective sections of the CIE color space diagrams (Fig.3). While the stacked OLEDs with a WO 3layer below the critical thickness showed highly angular-dependent CIE characteristics, the situation changed with increased thickness of the TMO layer (Fig.3a). There, the CIE values only ranged from X ¼0.64to 0.62and from Y ¼0.28to 0.30for angles between 08and 708. The same characteristics are obtained with optical device simulation by assuming full operation of both light-emitting units. However, in case of stacked OLEDs with an incomplete heterointerface (withWO 3thickness below 2nm), the CIE characteristics can only be reproduced if we assume that only OLED 1emits light. As a result it is essential to note, that the low efﬁciencyof the entire stack cannot be explained by two partially functional sub-OLEDs. On the other hand, this result indicates that the contribution of sub-OLED 1to the total current efﬁciencyis less than that of sub-OLED 2operated under similar conditions, as

reﬂectedby the L–I–Vmeasurements above (Fig.2a). Thereby, the asymmetric contribution of both light-emitting units to the luminance must be attributed to the non-optimal distance of the emission layer of sub-OLED 1to the metallic top contact, which leads to a substantially lower out-coupling efﬁciencyfor this light-emitting unit. Similarly, the comparison between experiment and simulation upon variation of the thickness of the Cs 2CO 3-doped BPhen ﬁlmdemonstrates the following. Full operation of both light-emitting units is achieved for thick layers of BPhen:Cs2CO 3, whereas light emission from only sub-OLED 2is obtained for a BPhen:Cs2CO 3thickness below 5nm (illustratedby two CIE characteristics in Fig. 3b). As a ﬁrstresult of these electro-optical studies, we conclude that 5nm of n-type doped BPhen and 2.5nm of WO 3represent the critical thicknesses for full operation of both sub-OLEDs of the entire stacked device. When only one sub-OLED is actually emitting, a leakage current through the non-emitting sub-OLED must supply charge carriers for the emitting sub-OLED, leading to comparatively high operating voltages for these stacked OLEDs (notshown here). 2.2. CGL:Principle of Operation

All reports on CGL units published so far emphasize the requirement of an n-type doped organic semiconductor

layer

1764

ß2010WILEY-VCH Verlag GmbH &Co. KGaA, Weinheim Adv. Funct. Mater. 2010, 20, 1762–1766

www.MaterialsViews.com

www.afm-journal.de

adjacent to a p-type doped layer, a TCO layer, or a TMO layer. This is strong evidence of the essential role played by such a layer in the mechanisms operating in the CGL. In the following, we show that it serves as EIL, while the actual process of charge generation occurs at a different interface. We use the electronic structure of the interface between Cs 2CO 3-doped BPhen and WO 3determined via UPS and IPES. [19]As shown in previous Kelvin probe and UPS FULL PAPER

vac

E measurements, a 10-nm-thick layer of WO 3(depositedon Au-coated n þdoped Si) exhibits a high work function (WF)of E F F

6.68eV. [17,18]The electron afﬁnity(EA)and ionization energy (IE)of such a layer have been found equal to 6.45and 9.83eV via IPES and UPS measurements, respectively. The small energy difference between EA and WF indicates that the Fermi level E F is very close to the TMO conduction band minimum and thus that the TMO is a highly n-type doped semiconductor, presumably due to oxygen vacancy defects acting as donors in transparent conducting Figure 5. Energy level diagram of the BPhen:Cs2CO 3/WO3/TCTAjunction oxides. [20]For MoO 3, the formation of oxygen deﬁcientﬁlmshas determined by UPS and IPES (spectraof these measurements will be been evidenced by X-ray photoelectron spectroscopy and is published elsewhere) [19].

attributed to the decomposition and preferential evaporation of layer, an incomplete space charge region forms, the WF does not the lower vapor pressure constituent atomic species. [21,22]

reach saturation in the BPhen-covered WO 3, and the built-in During the gradual deposition of Cs 2CO 3-doped BPhen (9wt%)potential remains too low to allow for efﬁcienttunneling of onto WO 3, the WF F of the ﬁlmdecreases with almost parabolic electrons. In that case, the energy difference between the CB of characteristics, indicating the formation of a space charge region WO 3and the LUMO of BPhen:Cs2CO 3increases as the thickness within the organic ETL (Fig.4). Starting with the large value of the n-type doped layer decreases. Concomitantly, the shape of corresponding to the neat layer of WO 3, F drops and ﬁnallythe tunnel barrier for electrons changes unfavorably into a saturates at 2.5eV for thicknesses d of 13–26nm. The saturation rectangle. Taking into account the fact that the n-type doping actually occurs at approximately 7.5nm when considering the concentration used in the stacked OLEDs was higher than in the parabolic characteristics of the work function indicated by the ﬁtsamples studied by UPS/IPES,the minimum thickness of 5nm (Fig.4). Taking into account the IE (6.8eV) and EA (2.4eV) of found above is in favorable agreement with the characteristic width BPhen:Cs2CO 3measured by UPS and IPES, the energy level of the space charge region derived here. On the other hand, the 2.5-alignment at the interface with the TMO can be represented as nm minimum thickness of WO 3found in our electro-optical shown in Figure 5, which corresponds to the heterointerface studies is believed to be simply related to the required amount of between the bottom OLED 1and the top OLED 2. It is evident from deposited material necessary to form a continuous layer of TMO. this schematic that no charge generation occurs between the n-type From the study of the heterointerface between the stacked OLEDs, doped BPhen and the TMO layer. On the other hand, electrons we can unambiguously conclude that the Cs 2CO 3-doped BPhen reaching this interface can tunnel from the conduction band (CB)only acts as an EIL and is not directly involved in the charge-of WO 3through the narrow potential barrier into the LUMO of the generation process. Consequently, the actual charge-generation n-doped organic ETL. The higher tunneling probability is strongly mechanism must be attributed to the heterointerface between supported by the fact that the high WF difference of the TMO and WO 3and TCTA within OLED 2. UPS and IPES measurements BPhen:Cs2CO 3amounts to 4.2eV, leading to iso-energetic electron show an interfacial dipole D ¼1.5eV between the two materials, levels in the CB of WO 3and LUMO of the ETL. These results also complete with only 1.6nm of the organic material, as well as a small explain that below the minimum thickness of n-type doped BPhen

(0.8eV) barrier between the highest occupied molecular orbital (HOMO)of TCTA and the CB of WO 3. [19]This electronic conﬁgurationallows therefore electrons to be injected into the TMO CB, resulting in a hole in the HTL. Note that this interface molecular level alignment is entirely equivalent to that recently

found by Kro

¨ger et al. for interfaces between MoO 3or WO 3and another HTL, N,N ’-diphenyl-N,N’-bis(1-naphthyl)-1,1’-biphenyl-4,4’-diamine(a -NPD). [16,23]

Accordingly, our ﬁndingscast some doubt on the recently published hypothesis about a two-step process of a charge generation directly within the TMO and an electron injection afterwards based on a tunneling-assisted thermionic emission into the n-type doped ETL, since that explanation neglects recent results concerning the electronic structure of TMOs. [13]The resulting misconception denotes the combination of n-type doped ETL and TMO as the essential components of a CGL. According to our Figure 4. F measured by UPS on Cs 2CO 3-doped BPhen (9wt%)with study, the combination of the TMO and HTL ﬁlmscan be regarded thickness d deposited on 10nm of WO 3.

as the actual CGL, given the high work function and the deep

lying

Adv. Funct. Mater. 2010, 20, 1762–1766ß2010WILEY-VCH Verlag GmbH &Co. KGaA, Weinheim

1765

www.afm-journal.de

www.MaterialsViews.com

photodetecting unit was applied for L –I –V measurements under ambient conditions. The angular resolved EL spectra were obtained by an optical ﬁberconnected with an imaging monochromator system (Triax320,CCD4000, Jobin Yvon). Thereby, the stacked OLEDs were operated at current densities from 10to 80mA cm À2. The maximum angle of collection given by the measurement setup can be estimated to 0.18. Optical device simulation was performed by the commercial software ETFOS (Fluxim).The optical parameters of the organic materials in the device stack have been obtained byspectroscopicellipsometry(Sopra).TheUPSandIPESmeasurementswereconducted at Princeton University. Details of the experimental setup are given elsewhere [19].

CB of the TMOs on the one hand, and the energy level alignment with the HOMO of the adjacent HTL on the other. This result appears to be of general nature, as the recent results reported by ¨ger et al. show that the hole injection from MoO 3into a -NPD is Kro

based on a similar mechanism. [16]It is also worth noting that the role of TMOs in CGLs can be compared with the one of, for example, 1,4,5,8,9,11-hexaazatriphenylene-hexacarbonitrile (HAT-CN), which also exhibits a high WF ($6eV) and a deep lying LUMO level. This leads to the same electric-ﬁeld-assistedcharge-generation mechanism at the interface between HAT-CN and a hole transport material. [24,25]As a consequence, the n-type doped BPhen is only required to enable an efﬁcientelectron injection from the TMO into the bottom OLED unit. The formation of an interfacial dipole is also likely at the interface between TMO and a non-doped ETL. This could possibly explain the steep decay of the WF within the ﬁrstfew monolayers of BPhen:Cs2CO 3deposited on top of the TMO, leading to some local deviation from the ideal parabolic ﬁt(Fig.4, ﬁrstfew nanometers). However, the value of the dipole is supposed to amount between 1.5and 2eV, which is not sufﬁcientfor an electron injection from the deep lying CB of the TMO into the LUMO of a non-doped ETL like TPBi as has been shown by our electro-optical studies.

F U L L P A P E R

Acknowledgements

Work in Braunschweig was ﬁnanciallysupported by the German Federal Ministry for Education and Research (FKZ:13N8995, 13N9152). Work in Princeton was supported by the National Science Foundation (GrantNo. DMR-0705920) and the Princeton MRSEC of the NSF (GrantNo. DMR-0819860). J. M. thanks the Deutsche Forschungsgemeinschaft (DFG)for generous support within the postdoctoral fellowship program.

Received:February 12, 2010Published online:May 14, 2010

3. Conclusions

In summary, we have demonstrated that the charge-generation mechanism in TMO-based interconnecting units of stacked OLEDs occurs at the heterointerface between the TMO and the adjacent non-doped hole-transporting layer (e.g.,TCTA). Despite a substantial interfacial dipole of 1.5eV, the energetic difference between the CB of WO 3and the HOMO of TCTA only amounts to 0.8eV, allowing for an efﬁcientcharge generation and separation at this interface. Consequently, the combination of TMO and HTL states the actual CGL. This is due to the nature of TMOs like WO 3and MoO 3having a deep-lying conduction band and a high WF. Accordingly, an electric-ﬁeld-assistedcharge-generation process takes place. Consequently, the adjacent n-type doped electron transport layer is only used to facilitate the electron injection from the TMO into the adjacent sub-OLED. These results have been used to explain the experimental results on twofold stacked OLEDs with a corresponding CGL architecture. In order to ﬁndboth sub-OLEDs fully functional, a critical thickness of the doped ETL and the TMO has been determined to be 5and 2.5nm, respectively.

4. Experimental

The device preparation and characterization was carried out at the TU Braunschweig. All devices were prepared on commercial glass substrates coated with 140-nm-thick ITO with a sheet resistance of 14V sq À1(Merck).The deposition of the organic and inorganic ﬁlmswere carried out by thermal evaporation within a 10À8mbar vacuum system with separate deposition chambers for n-type doping as well as transport and emitting materials and metal contact. The deposition rate for all organic and inorganic materials was controlled by quartz-crystal monitors and kept constant within the range from 0.02to 0.1nm s À1. For the evaporation of WO 3and Cs 2CO 3we used shielded high-temperature evaporation sources (CreaTec).The doping of TPBi and BPhen with Ir(ppy)3and Cs 2CO 3, respectively, was made by thermal co-evaporation controlled via two separate quartz-crystal monitors. A Keithley 2400source meter in combination with a calibrated Advantest TQ 8221

[1]J. Kido, T. Matsumoto, T. Nakada, J. Endo, K. Mori, N. Kawamura, A. Yokoi,

SID Int. Symp. Dig. Tech. Pap. 2003, 34, 979.

[2]L. S. Liao, K. P. Klubek, C. W. Tang, Appl. Phys. Lett. 2004, 84, 167. [3]T.-Y. Cho, C.-L. Lin, C.-C. Wu, Appl. Phys. Lett. 2006, 88, 111106.

[4]X. D. Gao, J. Zhou, Z. T. Xie, B. F. Ding, Y. C. Qian, X. M. Ding, X. Y. Hou,

Appl. Phys. Lett. 2008, 93, 083304.

[5]J. X. Sun, X. L. Zhu, H. J. Peng, M. Wong, H. S. Kwok, Appl. Phys. Lett. 2005,

87, 093504.

[6]F. Guo, D. Ma, Appl. Phys. Lett. 2005, 87, 173510.

[7]C.-W. Chen, Y.-J. Lu, C.-C. Wu, E. H.-E. Wu, C.-W. Chu, Y. Yang, Appl. Phys.

Lett. 2005, 87, 241121.

[8]C.-C. Chang, J.-F. Chen, S.-W. Hwang, C. H. Chen, Appl. Phys. Lett. 2005, 87,

253501.

[9]H. Kanno, R. J. Holmes, Y. Sun, S. Kena-Cohen, S. R. Forrest, Adv. Mater.

2006, 18, 339.

[10]D.-S. Leem, J.-H. Lee, J.-J. Kim, J.-W. Kang, Appl. Phys. Lett. 2008, 87,

103304.

¨ger, S. Hamwi, J. Meyer, T. Dobbertin, T. Riedl, W. Kowalsky, [11]M. Kro

H.-H. Johannes, Phys. Rev. B 2007, 75, 235321.

[12]M. Terai, K. Fujita, T. Tsutsui, Jpn. J. Appl. Phys 2005, 44, L 1059. [13]X. Qi, N. Li, S. R. Forrest, J. Appl. Phys. 2010, 107, 014514.

¨ger, S. Hamwi, J. Meyer, T. Riedl, W. Kowalsky, A. Kahn, Org. Electron. [14]M. Kro

2009, 10, 932.

[15]D. Y. Kim, J. Subbiah, G. Sarasqueta, F. So, H. Ding, Irfan, Y. Gao, Appl.

Phys. Lett. 2009, 95, 093304.

[16]K. Kanai, K. Koizumi, S. Ouchi, Y. Tsakamoto, K. Sakanoue, Y. Ouchi,

K. Seki, Org. Electron. 2010, 11, 188.

¨ger, S. Hamwi, J. Meyer, T. Riedl, W. Kowalsky, A. Kahn, Appl. Phys. [17]M. Kro

Lett. 2009, 95, 123301.

¨low, H.-H. Johannes, T. Riedl, W. Kowalsky, Appl. [18]J. Meyer, S. Hamwi, T. Bu

Phys. Lett. 2007, 91, 113506.

¨ger, S. Hamwi, T. Riedl, W. Kowalsky, A. Kahn, unpublished. [19]J. Meyer, M. Kro

[20]S. Samson, C. G. Fonstad, J. Appl. Phys 1973, 44, 4618.

[21]T. S. Sian, G. B. Reddy, Sol. Energy Mater. Sol. Cells 2004, 82, 375.

[22]K. S. Rao, K. V. Madhuri, S. Uthanna, O. M. Hussain, C. Julien, Mat. Sci.

Eng. B 2003, 100, 79.

¨ger, A. Kahn, Appl. Phys. Lett. 2010, 96, 133308. [23]J. Meyer, A. Shu, M. Kro

[24]L. S. Liao, K. P. Klubek, Appl. Phys. Lett. 2008, 92, 223311.

[25]Y.-K. Kim, J. W. Kim, Y. Park, Appl. Phys. Lett. 2009, 94,

063305.

1766