Effects of thermal treatment and depth profiling analysis of solution processed bulk-heterojunction organic photovoltaic cells

Journal of Colloid and Interface Science 436 (2014) 9–15

Contents lists available at ScienceDirect

Journal of Colloid and Interface Science www.elsevier.com/locate/jcis

Effects of thermal treatment and depth profiling analysis of solution processed bulk-heterojunction organic photovoltaic cells Pontsho S. Mbule, Hendrik C. Swart, Odireleng M. Ntwaeaborwa ⇑ Department of Physics, University of the Free State, Bloemfontein ZA9300, South Africa

a r t i c l e

i n f o

Article history: Received 7 July 2014 Accepted 3 September 2014 Available online 16 September 2014 Keywords: Organic solar cells Zinc oxide Bulk-heterojunction TOF-SIMS Buffer layer Interface

a b s t r a c t We report the use of solution processed zinc oxide (ZnO) nanoparticles as a buffer layer inserted between the top metal electrode and the photo-active layer in bulk-heterojunction (BHJ) organic solar cell (OSC) devices. The photovoltaic properties were compared for devices annealed before (Device A) or after (Device B) the deposition of the Al top electrode. The post-annealing treatment was shown to improve the power conversion efficiency up to 2.93% and the fill factor (FF) up to 63% under AM1.5 (100 mW/ cm2) illumination. We performed the depth profile/interface analysis and elemental mapping using the time-of-flight secondary ion mass spectrometry (TOF-SIMS). Signals arising from 27Al, 16O, 12C, 32S, 64 Zn, 28Si, 120Sn and 115In give an indication of successive deposition of Al, ZnO, P3HT:PCBM and PEDOT:PSS layers on ITO coated glass substrates. Furthermore, we discuss the surface imaging and visualize the chemical information on the surface of the devices. Ó 2014 Elsevier Inc. All rights reserved.

1. Introduction An enhancement of power conversion efficiency (PCE) of BHJ–OSC devices can be achieved by tailoring the morphology of the photoactive layer. The processing of the photo-active materials and post-production treatment can cause changes on the morphology and phase separation of the materials in the photo-active layer and therefore influence the general performance of the OSC devices [1]. Controlling the processing conditions such as thermal treatment can help to improve the interfacial contact and charge carrier mobility thereby increasing the quantum efficiency of the BHJ–OSC devices [2]. The greatest challenge is to optimize the interfacial area of the donor/acceptor materials. Charge mobility within the photoactive layer and the crystallinity of P3HT polymer within the blend and the contact between the photo-active layer and the electrode is one of the most critical interface issue in organic solar cells. In addition, operation or storage of BHJ–OSC devices in air, even with encapsulation, leads to material degradation due to penetration of oxygen or water molecules through the top electrode. Inserting metal oxide nanoparticles buffer layer into the interface between the active layer and the cathode electrode is regarded as one of the effective strategies in interface engineering to improve the device performance [3]. Zinc-oxide (ZnO) ⇑ Fax: +27 051 401 3507. E-mail address: [email protected] (O.M. Ntwaeaborwa). http://dx.doi.org/10.1016/j.jcis.2014.09.005 0021-9797/Ó 2014 Elsevier Inc. All rights reserved.

nanoparticles buffer layers are of extensive use in OSCs, as hole blocking layer as well as the electron extracting layer to facilitate electron transport from the photo-active layer (P3HT:PCBM blend) to the top metal electrode [4–9]. The buffer layer should have minimal absorption losses and should be capable of driving out photogenerated carriers with minimum recombination losses and electrical resistance and the beneficial effects of the buffer layer ranges from modifying the absorber surface to protecting the sensitive interface during the deposition of metal electrode [10]. ZnO nanoparticle layer can prevent the diffusion of Al atoms into the photoactive layer that could lower the power conversion efficiency of the device by acting as recombination centers. Furthermore, ZnO layer has the potential to improve the environmental stability of the OSC device by absorbing UV photons that could lead to bond breaking or photo-oxidation in the presence of water and oxygen molecules [11]. In many multi-layered electronic devices, the interfaces between their layers govern their properties and changes such as oxidation or the formation of polar layers affect their functioning. Hence, to understand the BHJ structure the study of these interfaces is important for the construction of organic solar cell devices with improved power conversion efficiency [10–12]. In this study, we show the effects of thermal treatment on the photovoltaic (PV) properties of OSC devices. The devices PV properties are compared for devices annealed before and after the deposition of the top metal electrode. Furthermore, depth profile analysis was performed to examine inter-diffusion of chemical species from the

10

P.S. Mbule et al. / Journal of Colloid and Interface Science 436 (2014) 9–15

surface to the bulk and vice versa, and distribution of the ions on the surface was analyzed by the 3D-imaging TOF-SIMS. 2. Experimental section 2.1. ZnO nanoparticles synthesis ZnO nanoparticles were synthesized by hydroxylation of zinc acetate dihydrate (Zn(Ac)2H2O, 99%, Sigma–Aldrich) by tetramethylammonium hydroxide (TMAH, 25WT%, Sigma–Aldrich). In a typical preparation, TMAH dissolved in 30 ml of ethanol (Absolute P99.8%, Sigma–Aldrich) was added dropwise to 0.1 M zinc acetate dihydrate dissolved in 30 ml of dimethylsulfoxide (DMSO, Anhydrous P99.9%, Sigma–Aldrich) followed by vigorous stirring for 1 h at room temperature. The precipitate was separated by centrifugation and was washed at least three times by a mixture of heptane (Anhydrous, P99%, Aldrich) and ethanol in the volume ratio of 2:1 and then in heptane only. The ZnO nanoparticles were either dispersed in ethanol solvent or dried in an oven kept at 110 °C. UV–Vis spectrometer Perkin–Elmer Lambda 35 UV–Vis– NIR was used to record the reflectance spectra. 2.2. Device fabrication The device geometry was ITO/PEDOT:PSS/P3HT: PCBM/ZnO nanoparticles/Al with the device area of 0.12 cm2. Before depositing different layers on a glass substrate precoated with ITO, the substrate was first cleaned ultrasonically using isopropanol (P99.5%, Sigma–Aldrich) and acetone (ACS reagent, P99.5%, Sigma–Aldrich) consecutively for 10 min. A layer of poly(3,4-ethylenedioxythiophene):polystyrenesulfonic acid (PEDOT:PSS, CLEVIOS™ AI 4083) was spin coated followed by drying in an oven/hot plate at 110 °C for 10 min. The photoactive layer of a poly (3-hexylthiophene):[6,6] phenyl butyric acid methyl ester (P3HT:PCBM) blend with a weight ratio of 1:0.6, dissolved in chlorobenzene (Anhydrous, 99.8%) was then spin-coated at the speed of 1000 rpm for 15 s. This was followed by the deposition of a layer of ethanol solution of ZnO nanoparticles at 4000 rpm for 35 s. Finally, the top Al metal electrode (100 nm) was thermally evaporated at a pressure of 1  106 Torr through a shadow mask defining the device area. Two devices are compared in this study, namely device A that was annealed at 155 °C for 10 min before the evaporation of the Al electrode and this is referred to as pre-deposition or just pre-annealed device, and device B that was annealed at 155 °C for 10 min after the deposition of the Al top Al metal electrode and this is referred to as post-deposition or just post-annealed device. 2.3. Photovoltaic measurements Current density versus voltage (J–V) characteristics were measured using a Keithley 2400 source meter and an Oriel xenon lamp (150 W) coupled with an AM1.5 filter to simulate sunlight. The light intensity was calibrated with a silicon reference cell with a KG2 filter following standard solar cell testing procedures. All J–V measurements were conducted using the light intensity of 100 mW/cm2. The external quantum efficiencies (EQE) as a function of wavelength were measured using an incident photon-tocurrent efficiency (IPCE) measurement system (PV measurement, Inc.). The wavelength of the bias light was controlled with optical filters (Andover Corporation). 2.4. SIMS depth profiling and imaging The depth profile analysis was performed using Iontof TOFSIMS5. Two depth profiling ion beams operating in the dual beam

mode were used. While a primary current of 70 nA of Cs+ ions was sputtering the crater, the 0.3 pA Bi++ 3 ion was progressively analyzing the crater bottom, with impact energies of 1 keV (Cs+) and 30 keV (Bi++ 3 ) and the sputter area of 300  300 lm with the sputter time of 10 s/scan. Positive and negative secondary ion polarities of SIMS measurements were performed directly on the Al cathode. The elemental mapping was performed with a Cs+ 0.3 pA primary beam, and was rastered over an area of 100  100 lm on the surface. The surface images were recorded with a resolution of 512  512 pixels. The measurements were performed in a chamber maintained at the base pressure of 1.5  109 mbar. 3. Results and discussion The schematic diagram showing the conventional geometry together with the energy level alignment of OSC devices constructed in this study is shown in Fig. 1(a and b). While P3HT and PCBM form a donor–acceptor heterojunction that facilitates the dissociation of photo-generated excitons (bound electron–hole pairs), the lower conduction band edge of ZnO as compared to the lowest unoccupied molecular orbital (LUMO) of P3HT may also lead to dissociation of excitons in P3HT via rapid electron transfer to ZnO. The electron affinities of ZnO and PCBM suggest that there is a negligible barrier height for electron transport from PCBM toward the Al cathode. Furthermore, the deep valence band of ZnO creates a large barrier height to block hole injection from the P3HT: PCBM active layer to the top Al electrode. Fig. 2(a) shows the UV–Vis reflectance spectra of the ZnO nanoparticles before and after annealing for 1, 2 and 3 h. It shows the maximum reflectance from the as-prepared ZnO nanoparticles and the reflectance in the visible-region is greatly reduced for samples annealed for 2 h, indicating a very high absorption for this sample corresponding to the transition from valence band to conduction band, suggesting that annealing has got an effect on the bandgap energy of the ZnO nanoparticles as shown in Fig. 2(b). The direct bandgap energy of the prepared ZnO nanoparticles was estimated using Tauc’s relation [13] and Fig. 2(b) shows the plot of (aht)2 vs the photon energy (ht). Extrapolation of the linear portion of the plot to photon energy (ht) axis gives the bandgap of ZnO nanoparticles and was found to be 3.30 eV for the as prepared ZnO nanoparticles and it was 3.24 eV, 3.12 eV and 3.16 eV after annealing the nanoparticles for 1, 2 and 3 h respectively. The insert shows the variation of the bandgap energy with annealing time. These results suggest that annealing has got the potential to realign the energy levels of ZnO nanoparticles that could lead to the difference in the VOC values. The primary function of a buffer layer is to form a junction with the absorber layer while absorbing a maximum amount of light to the junction region and absorber layer as well as driving out photo-generated carriers with minimal losses [14]. Candidates for buffer layer should hold wider bandgap for efficient light absorption and the process for deposition should provide capability to passivate the surface states of the absorber layer and should provide an alignment of conduction band with the absorber to yield better efficiency. The particle sizes were estimated using the following equation [15]:

" EðrÞ ¼

E2g

þ

2hEgðp=rÞ2

#1=2

m

where E(r) is the nanocrystal bandgap as a function of particle radius r, Eg is bulk crystal bandgap, m⁄ = 0.24me (ZnO) is the electron effective mass, ⁄ = 1.054  1034 J s [15]. Using E(r) = 3.30 eV (as-prepared ZnO), 3.24 eV (ZnO annealed for 1 h), 3.12 eV (ZnO annealed for 2 h) and 3.16 eV (ZnO annealed for 3 h), the average

P.S. Mbule et al. / Journal of Colloid and Interface Science 436 (2014) 9–15

11

e(a)

(b)

Al

e-

ZnO P3HT:PCBM PEDOT:PSS ITO Glass Fig. 1. (a) Conventional schematic device structure and (b) Energy level diagram of OSC. The energies are referenced to the vacuum level.

Fig. 2. (a) Optical reflectance spectra of the ZnO nanoparticles before and after annealing for 1 h, 2 h and 3 h; and (b) Bandgap energy estimation of ZnO nanoparticles.

particle diameter was estimated to be 4.46 nm, 3.72 nm, 2.84 nm and 2.78 nm, respectively. The photovoltaic responses of devices A and B are shown in Fig. 3. The device performance is greatly improved by the thermal treatment after the deposition of the Al cathode (post-annealing). The PCE is significantly increased from 1.57% to 2.93%. In addition, the Voc of device B increased marginally from 0.43 V to 0.65 V. A summary of the PV characteristics is presented in Table 1. The contact between the polymer active layer and the electrode is one of

the most critical interfaces in the OSC devices. The improvement of the PV properties can be attributed to combined effects of the ZnO buffer layer and post-annealing treatment. As reported previously, ZnO nanoparticles have high electron mobility (6.6  102 cm2 V1 s1), they can strongly absorb the high energy near ultraviolet photon which may cause photo-oxidation [16], they can block holes from the photo-active layer to the top electrode, and they can also prevent Al metal ions from diffusing into the photoactive layers. On the other hand Ntwaeaborwa et al. [17]

Fig. 3. (a) J–V characteristics and (b) external quantum efficiencies of the devices.

12

P.S. Mbule et al. / Journal of Colloid and Interface Science 436 (2014) 9–15

Table 1 Photovoltaic characteristics of devices A and B. Device A

Device B

Jsc (mA/cm2)

Voc (V)

FF (%)

PCE (%)

Jsc (mA/cm2)

Voc (V)

FF (%)

PCE (%)

7.02

0.433

51.69

1.57

7.10

0.654

63.07

2.93

and Kim et al. [18] reported that the post-annealed devices showed an improved performance. The increase of Voc in device B may be attributed to a decrease in ohmic contact/series resistance at the interface between ZnO layer and Al electrode. The change in ohmic contact at the interface may be a result of energy level re-alignment due to post-annealing as stated by Zhao et al. [19]. The change of ohmic contact in organic solar cells is considered to have a strong impact on the Voc and FF of the cell. Fig. 3(b) shows the external quantum efficiency versus wavelength spectra of devices A and B. The maximum photocurrent contribution is observed at 501 nm and an improvement of device performance of about 10% is obtained from the post-annealed device. SIMS depth profiling results of device A and B are presented in Fig. 4(a and b). Although slight difference was observed in intensity, signals arising from 27Al, 16O, 12C, 32S, 64Zn, 28Si, 120Sn and 115 In ions clearly identifies successive layers of Al/ZnO/ P3HT:PCBM/PEDOT:PSS and ITO/glass, respectively. There was a slight intensity increase of the carbon signal on both devices. The formation of aluminum oxide was possible and this has been

reported before by Bulle-Lieuwma et al. [20]. The Al signal that decreases and diffuses into the active layer (P3HT:PCBM) originates from the Al metal. The interface between Al and P3HT:PCBM is found to be oxygen-rich, as shown by the high 16O signal. For conventional geometry devices, using aluminum as a back electrode, both molecular oxygen and water will diffuse through the aluminum electrode [21]. Molecular oxygen will preferentially diffuse through microscopic pinholes in the electrode while water will preferentially diffuse in between the aluminum grains. As a result, they will diffuse through all the layers in the device to the ITO interface. All organic layers will then react with molecular oxygen and water in varying degree causing oxidation and thus degradation. In addition, the 16O may also be due to ZnO layer or aluminum oxide (or hydroxide) that could be formed during Al evaporation as a result of a chemical reaction with residual gases in the vacuum chamber [22,23], and water present on the sample surface. We therefore assume that the low Voc of 0.43 V and FF of 51.6% obtained in device A maybe due to the diffusion of materials in the device. Distribution of sulfur (32S), which originates from P3HT, PSS and PEDOT and carbon (16C) from P3HT:PCBM is stable at the Al region, and then simultaneously increases toward the P3HT:PCBM region. While 16C is uniformly distributed within the active layer then significantly decreases toward the PEDOT:PSS layer, the 32S first increased slightly, decreased slowly and is therefore distributed uniformly throughout the PEDOT:PSS layer and finally decreased toward the ITO/glass region. The diffusion of tin (120Sn) and indium (115In) into the active layer from ITO/glass region has been recognized as one of the main factors detrimental

Fig. 4. TOF-SIMS depth profiles (intensity as function of a sputter time) obtained directly on the Al cathode for Devices A (a and b) Positive SIMS and (c and d) Negative SIMS.

13

P.S. Mbule et al. / Journal of Colloid and Interface Science 436 (2014) 9–15

Fig. 5. TOF-SIMS depth profiles (intensity as function of sputter time) of (a)

Carbon

Oxygen

Silicon

Sulfur

27

Al and (b)

16

C.

Aluminum

Zinc

Overlay

Fig. 6. Positive ions, 3D elemental mapping for device A, after 180 scans.

to the lifetime of organic solar cells. The diffusion of 115In and 120Sn depends on the nature of the layers present. In the case of PCBM/ PEDOT:PSS/ITO, 120Sn and 115In signals are found within the PEDOT:PSS until they reach the P3HT:PCBM/PEDOT:PSS interface, where their signals drop. de Jong et al. [24] explained using Rutherford backscattering spectroscopy (RBS) technique, that the diffusion of 115In into the PEDOT:PSS layer strongly increased upon exposure to air and they explained that water present in the atmospheric air and taken up by the PEDOT:PSS liberates the acidic protons of PSS. Acidic etching then result in indium ions dissolving in the PEDOT:PSS. This may possibly result in both indium and tin diffusing to the P3HT:PCBM layer. Because our samples were exposed to air or were kept at ambient laboratory conditions, our findings may be in accordance with the results of de jong. The silicon

(28Si) signal originating from the underlying glass substrate used is low throughout the layers and increases at the ITO/glass region. C4HS and SO2 ions originating from P3HT and PEDOT:PSS were detected as shown in Fig. 4(c and d). The chemical distribution followed the same pattern as the ones presented in Fig. 4(a and b). C2O3H3, SO2, C4HS, C11H17S and C8H7SO3 secondary ions were also detected not only in the PEDOT:PSS/ITO layers but also in the inner region of the P3HT:PCBM layer. These results suggest that the active layer region is P3HT and PSS enriched. Fig. 5(a and b), shows that the concentration of 27Al and 16C for device A is slightly intense as compared to the device B. This could be due to the heat treatment effect on devices. The 3D imaging of positive and negative secondary ions is presented in Figs. 6 and 7, respectively. The distribution of signals

14

P.S. Mbule et al. / Journal of Colloid and Interface Science 436 (2014) 9–15

SO2-

C2H3O3-

C4HS-

C11H17S-

C8H7SO3-

Overlay

Fig. 7. Negative ions, 3D elemental mapping for device A, after 180 scans.

100

(a)

10

80

8

60

6

40

100

(b)

24

80

20 16

60

4

40

2

20

12 8

20

0

0

μm 0

ZnMC:

40

80

4

0

ZnMC:

11; TC: 3.047e+004

100

100

(c) 80

160

120

40

80

40

80

27; TC: 2.418e+005

(d)

200

60

0

μm 0

600 80 500 60

400 300

40

200 20

40

0

0

μm 0

40

OMC: 222; TC: 1.684e+006

80

20

100

0

μm 0

0 40

OMC: 697; TC: 2.865e+007

80

Fig. 8. Ion mapping of Zn and O for Device A, after 60 scans (a–c) and after 180 scans (b–d).

shows high intensity/concentration for 16O, 32S while the intensity of 28Si, 64Zn and 12C and 27Al is relatively moderate. This observation is in correlation with the results observed in the depth profile shown in Fig. 4. Similar behavior is observed with the signals of C11H17S and C4HS showing most intense distribution as compared to SO2, C2H3O3 and C8H7SO3. Their respective

overlay images are also shown. However, due to their low concentration on the surface, 120Sn and 115In shows poor quality of surface imaging, hence they are not shown in this work. Fig. 8(a–c) shows the surface imaging of device A after 60 scans. Images show the Zn and O elemental or chemical distribution across the surface observed within 300 lm  300 lm dimension. These images

P.S. Mbule et al. / Journal of Colloid and Interface Science 436 (2014) 9–15

reveal the secondary ion intensities as a function of the location on the sample surface and after the first 60 scans, low concentrated Zn and O ions are homogenously distributed on the surface as indicated by a side bar scale. However, there are clear large particles on the surface of O indicating the inhomogeneous distribution. After 180 scans (Fig. 8(b–d)), the distribution of secondary ions detected intensified and showed more inhomogeneity across the surface. 4. Conclusion For a better understanding of how to improve photovoltaic properties of organic solar cells, we have studied the effect of thermal treatment before and after the deposition of an Al electrode. The improvement in photovoltaic properties was observed from the post annealed device (device B) and this is attributed to the combined effects of the ZnO buffer layer and post annealing treatment. We carried out the composition depth profile analysis of the organic solar cells. The results indicated that the BHJ layer was P3HT enriched and PSS penetrated into the BHJ layer. The interface between Al and P3HT:PCBM was found to be oxygen-rich suggesting species like AlO or AlO2 may have formed at the interface. This was supported by a clear increase in intensity of the 16O signal. Surface imaging showed both homogeneous and some inhomogeneous distribution of chemical species. Acknowledgments The authors would like to thank the University of the Free State (UFS), South African National Research Foundation (NRF) and Photonic Initiative of South Africa (PISA) for financial support. This research was partially supported by the KRCF (Korea Research Council of Fundamental Science and Technology) and the KIST (Korea Institute of Science and Technology) for ‘‘NAP (National Agenda Project) program’’. Dr. BongSoo Kim and Mr. Tae-Hee Kim, for their contribution in the fabrication of the devices and interpretation of data.

15

References [1] J.K.J. van Duren, X. Yang, J. Loos, C.W.T. Bulle-Lieuwma, A.B. Sieval, J.C. Hummelen, R.A.J. Janssen, Adv. Funct. Mater. 14 (2004) 425–434. [2] L. Chang, H.W.A. Lademann, J.-B. Bonekamp, K. Meerholz, A.J. Moule, Adv. Funct. Mater. 21 (2011) 1779–1787. [3] J. Huang, Z. Yin, Q. Zheng, Energy Environ. Sci. 4 (2011) 3861–3877. [4] L. Qian, J. Yang, R. Zhou, A. Tang, Y. Zheng, T.-K. Tseng, D. Bera, J. Xue, P.H. Holloway, J. Mater. Chem. 21 (2011) 3814–3817. [5] H.-J. Park, K.-H. Lee, B. Kumar, K.-S. Shin, S.-W. Jeong, S.-W. Kim, J. Nanoelectron. Optoelectron. 5 (2010) 1–4. [6] S.R. Ferreira, P. Lu, Y.-J. Lee, R.J. Davis, J.W.P. Hsu, J. Phys. Chem. 115 (2011) 13471–13475. [7] S. Kim, C.-H. Kim, S.K. Lee, J.-H. Jeong, J. Lee, S.-H. Lee, S.-H. Jin, W.S. Shin, C.E. Song, J.-H. Choi, J.-R. Jeong, Chem. Commun. 46 (2013) 6033–6035. [8] Y.-J. Noh, S.-I. Na, S.-S. Kim, Sol. Energy Mater. Sol. Cells 117 (2013) 139–144. [9] R.L.Z. Hoye, D. Muñus-Rojas, D.C. Iza, K.P. Musselman, J.L. MacManus-Driscoll, Sol. Energy Mater. Sol. Cells 116 (2013) 197–202. [10] G.G. Andersson, W.J.H. van Gennip, J.W. Niemantsverdriet, H.H. Brongersma, Chem. Phys. 278 (2002) 159–167. [11] P.S. Mbule, T.-H. Kim, B.-S. Kim, H.C. Swart, O.M. Ntwaeaborwa, Sol. Energy Mater. Sol. Cells 112 (2013) 6–12. [12] M.A. Baldo, S.R. Forrest, Phys. Rev. B 64 (2001) 085201-1–085201-17. [13] H. Oh, J. Krantz, I. Litzov, T. Stubhan, L. Pinna, C.J. Brabec, Sol. Energy Mater. Sol. Cells 95 (2011) 2194–2199. [14] B.E. McCandless, S.S. Hegedus, Proc. of the 22th IEEE Photovoltaic Specialists Conference, (1991) 967–972. [15] O.M. Ntwaeaborwa, R.E. Kroon, V. Kumar, T. Dubroca, J.-P. Park, H.C. Swart, J. Phys. Chem. Sol. 70 (2009) 1438–1442. [16] B. Masenelli, E. Tutis, M.N. Bussac, L. Zuppiroli, Synth. Metals 122 (2001) 141– 144. [17] O.M. Ntwaeaborwa, R. Zhou, L. Qian, S.S. Pitale, J. Xue, H.C. Swart, P.H. Holloway, Phys. B: Condens. Matter 407 (2012) 1631–1633. [18] H.J. Kim, J.H. Park, H.H. Lee, D.R. Lee, J.J. Kim, Org. Electrons 10 (2009) 1505– 1510. [19] Y. Zhao, Z. Xie, Y. Qu, Y. Geng, L. Wang, Synth. Metals 158 (2008) 908–911. [20] C.W.T. Bulle-Lieuwma, W.J.H. Van Gen nip, J.K.J. VanDuren, P. Jonkheijm, R.A.J. Janssen, J. Niemantsverdriet, Appl. Surf. Sci. 203–204 (2003) 547–550. [21] K. Norrman, V.M. Madsen, A.S. Gevorgyan, F.C. Krebs, J. Am. Chem. Soc. 132 (2010) 16883–16892. [22] H. Heil, J. Steiger, S. Karg, M. Gastel, H. Ortner, H. von Seggern, M. Stössel, J. Appl. Phys. 89 (2001) 420–424. [23] R.I.R. Blyth, S.A. Sardar, F.P. Netzer, M.G. Ramsey, Appl. Phys. Lett. 77 (2000) 1212–1214. [24] M.P. de Jong, L.J. van Ijzendoorn, M.J.A. de Voigt, Appl. Phys. Lett. 77 (2000) 2255–2257.