Charge Transfer and Collection in Dilute Organic Donor-Acceptor

2 days ago - Experimental and theoretical approaches are used to understand the role of nanomorphology on exciton dissociation and charge collection a...
0 downloads 5 Views 962KB Size
Subscriber access provided by UNIV OF ALABAMA BIRMINGHAM

Charge Transfer and Collection in Dilute Organic Donor-Acceptor Heterojunction Blends Kan Ding, Xiao Liu, and Stephen R. Forrest Nano Lett., Just Accepted Manuscript • DOI: 10.1021/acs.nanolett.8b00851 • Publication Date (Web): 23 Apr 2018 Downloaded from http://pubs.acs.org on April 24, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 21 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

Charge Transfer and Collection in Dilute Organic Donor-Acceptor Heterojunction Blends Kan Ding1, Xiao Liu2, Stephen R. Forrest*1,2,3 1 2

Department of Physics, University of Michigan

Department of Electrical Engineering and Computer Science, University of Michigan 3

Departments of Material Science and Engineering, University of Michigan Ann Arbor, MI 48109, USA Abstract

Experimental and theoretical approaches are used to understand the role of nanomorphology on exciton dissociation and charge collection at dilute donor-acceptor (D-A) organic heterojunctions. Specifically, two charge transfer (CT) states in D-A mixed HJs comprising nanocrystalline domains of tetraphenyldibenzoperiflanthene (DBP) as the donor and C70 as the acceptor are unambiguously related to the nanomorphology of the mixed layer. Alternating DBP:C70 multilayer stacks are used to identify and control the optical properties of the CT states, as well as to simulate the dilute mixed heterojunctions. A kinetic Monte Carlo model along with photoluminescence spectroscopy and scanning transmission electron microscopy are used to quantitatively evaluate the layer morphology under various growth conditions. As a result, we are able to understand the counter-intuitive observation of high charge extraction efficiency and device performance of DBP:C70 mixed layer photovoltaics at surprisingly low (~ 10%) donor concentrations. *Corresponding author email: [email protected] Keywords: multilayers; quantum confinement; percolation; kinetic Monte Carlo; bulk heterojunction

1

ACS Paragon Plus Environment

Nano Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The power conversion efficiencies of organic photovoltaic (OPV) cells1 exceed 13% which are well below their thermodynamically predicted limit of 22-25%.2 Among the many small molecular weight OPV structures investigated, mixed donor (D)-acceptor (A) heterojunction (HJ) active regions have the advantage of circumventing the trade-off between a short exciton diffusion length (LD) and a considerably longer optical absorption length (1/α, where α is the absorption coefficient) characteristic of most organics.3-6 These mixtures are optimized when there is a balance between exciton dissociation and charge extraction, which generally is found when the D and A concentrations are approximately equal. However, the performance of OPVs comprising tetraphenyldibenzoperiflanthene (DBP) and C70 is maximized at a surprisingly low DBP concentration of ~10% in C70.7, 8 It is unclear how the hole polarons in DBP created on the dissociation of excitons at D-A interfaces are extracted from such a dilute system, yet the high OPV efficiencies achieved suggests that charge extraction is unimpeded. We address this question by studying charge transfer (CT) states that serve as the intermediates formed at D-A interfaces after exciton dissociation, and hence play a central role in the photogeneration process.9 A CT state at the DBP:C70 HJ comprises an electron on the C70 acceptor and a hole on the DBP donor. The two charge carriers are Coulombically bound with an energy of a few hundred meV.10, 11 Previously,10 we found two distinct CT emission peaks (CT1 at 1.24 ± 0.07 eV and CT2 at 1.37 ± 0.13 eV) in both the electroluminescence (EL) and photoluminescence (PL) spectra of DBP:C70 blends. The CT2 peak energy red shifts with DBP concentration due to increased quantum confinement of the electron wavefunction in nanocrystalline C70 domains, whereas CT1 is attributed to a state confined to a mixed region where both donor and acceptor aggregates are small.10 In this work, we intentionally control the spatial confinement of CT states by growing alternating, ultrathin layers of C70 and DBP with the

2

ACS Paragon Plus Environment

Page 2 of 21

Page 3 of 21 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

degree of CT localization determined by the individual layer thicknesses. These structures confirm the previously proposed morphological origins of both CT states. To better understand the mechanisms governing charge extraction, a kinetic Monte Carlo (KMC) simulation model is applied to quantitatively simulate the nanomorphology of the active dilute donor layers. We find that even in the most dilute blends, planar DBP molecules stack into continuous percolating paths that enable efficient hole extraction. In previous work, the shift in CT2 peak energy with DBP concentration was attributed to reduced C70 crystallite diameters that confine the CT states. Moreover, the DBP concentration also affects the absorption spectrum. To understand the origin of the spectral energy shifts, we controlled the degree of excited state localization by growing alternating multilayer stacks of DBP and C70 to gain insights into the mechanisms affecting states in the blends of similar ratios of the two constituents. Figures 1 (a) and (b) show bright field cross-sectional scanning transmission electron microscopy (STEM) images of multilayer stacks with alternating layer thicknesses of 2 nm and 5 nm. To increase the imaging contrast between C70 and DBP molecules, dibenzo([f,f′]-4,7′-di[4-bromophenyl]-4′,7-diphenyl)diindeno[1,2,3-cd:1′,2′,3′-lm]perylene (Br2DBP) was substituted for DBP to create a contrast difference with the C70 layer (see Methods). Each sample contains 5 pairs of DBP and C70 layers capped by a C70 layer with the same thickness as the multilayer stack, and a 100 nm of Ag. Individual layers as thin as 2 nm are clearly identified in the images in Fig. 1. Figure 2 (a) shows the normalized room temperature PL spectra from DBP and C70 stacks with various layer thicknesses. All samples have a total thickness of 400 nm. The spectra are compared with that of a 1:1 mixed layer sample. The mixed sample has a lowest CT2 peak

3

ACS Paragon Plus Environment

Nano Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

energy  = 1.33 ± 0.01 eV. As the alternating layer thicknesses increase from 1 nm to 5 nm, E2 blue shifts to 1.43 ± 0.02 eV. Since the volume and ratio of materials in all the samples are equal, the energy shift of the CT2 peak is attributed to quantum confinement as the individual layer thicknesses decrease. A previously developed quantum mechanical model10 is modified to calculate the binding energy between electron and hole of the CT2 states with various alternating layer thicknesses (see Methods). The experimental and simulated results are shown in Fig. 2 (b). The calculated energy shift reasonably agrees with experiment except for the sample with 1 nm thick layers. This is possibly due to incomplete layer coverage for the thinnest sample, leading to reduced confinement. Confinement of CT states partially explains the optimal OPV performance with DBP concentration of only 10%. The electron and hole wavefunction overlap increases as the volume containing the CT state decreases, resulting in a larger recombination rate and a smaller dissociation efficiency compared with that of the delocalized CT states 10. Previous studies have also shown that an increased CT state energy leads to a higher open-circuit voltage (VOC )12, 13. Thus, a low DBP concentration benefits OPV performance due to the existence of CT states with higher energies. The question remains as to how the holes are efficiently extracted from such dilute donor layers as used in the DBP:C70 blends. The answer lies in the details of the blend morphology at the nanometer scale, which we simulated via a KMC model based on the van der Waals attraction between molecules (see Methods). The simulation results provide the average nanocrystalline C70 cluster radius ⟨r⟩ (circles), which was compared with values inferred from

4

ACS Paragon Plus Environment

Page 4 of 21

Page 5 of 21 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

the Scherrer approximation of the X-ray diffraction pattern (squares) shown in Fig. 3 (a)10. As the DBP concentration increases from 5% to 50%, ⟨r⟩ decreases from 5.2 ± 0.9 nm to 1.3 ± 0.2 nm in the simulated mixed layer, which agrees with experiment. In Fig. 3 (b), ⟨r⟩ extracted from simulations is used to calculate  , (circles) using the excited state confinement model. Black squares show E2 measured for the blends. Holes resulting from exciton dissociation are extracted through contiguous, percolating paths formed by DBP molecules. However, DBP concentrations of only ~10% could potentially impede hole extraction due to the formation of discontinuous networks. To understand the hole extraction process, therefore, we analyze the aggregation of DBP molecules in mixed HJs. Figure 4 (a) shows a 10-nm-deep slice of simulated mixed active layers with 11 mol% DBP. The evaporated molecules are incident from the top and travel normal to the surface of the film with the anode at the base. The 80-nm-thick film has a rough surface and a modest density of voids. Here, the C70 regions are shown in blue, and the green areas are DBP molecules that form into percolating paths to the anode. The red areas, which are in a substantial minority, are isolated islands of DBP disconnected from the anode. Figure 4 (b) shows only those percolating paths in DBP that extend from the anode into the active layer of a 50-nm-deep by 80-nm-thick slice. During deposition the planar DBP molecules form continuous π-π stacks that exclude C70 molecules from their interstices, thus minimizing the total energy of the blend. The stacks increase the probability of the formation of percolating paths to the anode even at very high dilutions. We define the 2D density of paths using:  =  /,

5

ACS Paragon Plus Environment

(1)

Nano Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 21

where  is the number of DBP molecules forming into continuous regions in a single layer at a distance z from the anode, and A is the anode contact area. Figure 5 shows the 2D density of percolating paths in the substrate plane in a blend with 10% DBP versus the distance from the anode. The simulation follows:  =    / ,

(2)

shown by the line in Fig. 4 (b). Here,  is the 2D density of percolating paths originating at the anode, and from the fit, ζ = 33 ± 5 nm is the characteristic percolation length, which is comparable to the typical active layer thickness using this material combination. At d =50 nm, which is the typical active layer thickness of an OPV cell, then  ≈ 0.05 nm-2,7, 8 corresponding to a mean distance between percolating paths of 4.5 nm. This is well below the C70 exciton diffusion length (~8 nm),14 suggesting that the holes can be extracted throughout the active layer with a DBP concentration of only 10 mol.%. Figure 4 (a) also shows that a fraction of DBP clusters form isolated islands where holes can be trapped following exciton dissociation. Dissociation on islands is therefore limited by the recombination rate of the previously dissociated charges within a particular island. Due to the spatial separation and energetic barriers of the trapped holes and free electrons delocalized among C70 molecules, the hole recombination rate is expected to be low compared with the charge extraction rate. In steady state, therefore, the exciton dissociation rate on isolated DBP islands is smaller than within the percolating paths. This is supported by the high peak internal quantum efficiencies of devices at donor concentrations of ~10%. In Fig. 6 we show the reverse-biased external quantum efficiencies (EQE) of OPVs to determine their absorption spectral dependence on the blended HJ composition. The OPV

6

ACS Paragon Plus Environment

Page 7 of 21 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

structure is ITO/MoO3(10 nm)/DBP:C70(54 nm)/BPhen(8 nm)/Al (100 nm). The DBP absorption coefficient is α = 3.4×105 cm-1 at a wavelength of 560 nm, and 4.3×105 cm-1 at 610 nm. This is compared to the broader but weaker absorption of C70, with α = 6 ± 3 ×104 cm-1 from 400 nm to 650 nm. The higher absorption of DBP also explains why such small concentrations can lead to a high OPV efficiency. Under reverse bias, all absorbed excitons efficiently dissociate into charges and are collected, thus EQE =ηA , the device absorption efficiency at wavelength, . Higher DBP concentrations result in increased absorption from 550 nm to 630 nm at the cost of decreased absorption from 400 nm to 500 nm. We calculate the total absorption efficiency, A as  =     /   ,

(3)

where F(λ) is an AM 1.5 G simulated solar spectrum. Then for devices with DBP concentrations between 10 and 50%, we find that  = 0.39 ± 0.01. Thus, the total absorption shows only a weak dependence on DBP concentration due to its much higher absorption coefficient compared with C70. This, again, contributes to the observed high efficiency of DBP:C70 solar cell using even dilute donor concentrations in the blends. With these analyses, we can summarize the performance of DBP:C70 mixed HJs at low DBP concentrations. Photogeneration is the product of four steps: exciton generation through light absorption with efficiency  , exciton diffusion to donor-acceptor interfaces with efficiency  , exciton dissociation into charge carriers with efficiency  , and charge collection at electrodes with efficiency !! . Reverse-biased EQE measurements suggest that  has only a weak dependence on DBP concentration. Significant quenching of bulk exciton emission in the PL spectrum indicates that  ≈ 1 even at low DBP concentrations. On the other hand, formation of CT2 at low DBP concentrations results in a larger  as well as a higher VOC. The

7

ACS Paragon Plus Environment

Nano Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

KMC simulations show that !! is also high due to the formation of DBP percolating paths at concentrations as low as 10%. Taken together, the unique stacking properties of DBP lead to its high absorption and charge collection properties, that persist even when the fraction of DBP molecules is small compared with the C70 acceptor in the blends. In conclusion, we have established a direct correlation between nanomorphology of the mixed active layers and the photogeneration processes in DBP:C70 blends using a combination of experimental and computational methods. Excitons generated in mixed regions experience spatial confinement leading to efficient dissociation into CT states with a high recombination efficiency. Excitons that dissociate at the boundary of nanocrystalline C70 regions efficiently form delocalized CT states. At low donor concentrations, the planar and high aspect ratio DBP molecules readily form continuous and extended percolating paths, allowing for the extraction of photogenerated charges at the opposing electrodes that are positioned exceptionally large distances from their point of origin. Combined with the larger absorption coefficient of DBP compared with C70 explains why optimized OPV efficiencies are obtained in such a dilute donor system.

Methods Sample Preparation. Samples for photoluminescence (PL) measurements are grown on solvent-cleaned (100) Si substrates by vacuum thermal evaporation (VTE) in a chamber with a base pressure of 10-7 torr. Prior to growth, both C70 and DBP are purified once via vacuum thermal gradient sublimation.15 During the growth of the alternating multilayer samples, temperatures of both sources are adjusted to give a stable growth rate of 0.1 Å/s while two

8

ACS Paragon Plus Environment

Page 8 of 21

Page 9 of 21 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

independently controlled source shutters are used to switch between DBP and C70 after each layer growth is completed. The PL spectra of the samples were measured in vacuum using a continuous wave He-Cd laser excitation source at a wavelength of 442 nm. Spectra were collected through a 550 nm long-pass filter and coupled into a monochromator using a fiber (Princeton Instruments SP-2300i). The signal was detected using a Si charge-coupled device array (PIXIS:400). Samples for scanning transmission electron microscope (STEM) measurements were grown using the same procedure as PL samples, but were capped with a C70 layer with a thickness equal to the total multilayer stack thickness, and a 100 nm thick Ag protection layer. The sample was then cut into thin slices using focused ion beam (FIB) milling (FEI Nova 200 Nanolab SEM/FIB). The sample thicknesses were estimated to be tens of nanometers. The samples were then examined using a JEOL 2100F high-resolution transmission electron microscope at an accelerating voltage of 200 kV. The Br2- DBP molecules used in STEM samples were synthesized by Luminescence Technology Corp.16 Measurement of the EQE of the photovoltaic cells grown similarly to the PL and STEM samples used focused monochromated light from a Xe arc-discharge lamp chopped at 200 Hz. A transconductance amplifier (Keithley 480) was used to provide a voltage bias and amplify the photocurrent. The amplified photocurrent was measured with a lock-in amplifier (Stanford Research Systems SR830). Sufficiently high reverse bias is applied until the photocurrent saturated. Under such conditions CT exciton dissociation and charge collection efficiencies approach unity, leading to EQE = ηA in the blends.

9

ACS Paragon Plus Environment

c

Nano Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 21

Quantum Confinement Calculation. The model used to calculate the CT peak energies of multilayer stacks is identical to that used in previous work

10

except that the hole is placed at

the center of the ultrathin DBP layer to account for their relative delocalization. Numerical Simulation. Following Peumans, et al.

17

and Yang, et al.18, molecules fill

sites starting at the substrate surface in a face-centered cubic (fcc) lattice with a lattice constant of 1.06 nm, consistent with the crystal structure of C70. Each DBP molecule has a length of 2.15 nm and can contact between 16 to 22 C70 molecules. Given its large size compared with C70, each DBP molecule is allowed to occupy two adjacent fcc sites, resulting in a coordination number of 18. The intermolecular binding energies for molecules in various configurations are evaluated using the Forcite molecular dynamics module in Materials Studio®. To simulate the deposition dynamics, a two-step process is employed: single molecule deposition followed by full sample annealing. In the first step, a molecule normally incident to the substrate lands at a random location at the growth surface, and attempts to move to adjacent vacant sites. An attempt is accepted if the target site has a lower total energy than its initial site, otherwise the attempt is accepted with a reduced probability of #Δ = exp − ∆ ⁄*+ , where ∆ is the difference between the total binding energies of the target and initial sites, k is Boltzmann’s constant and T = 300K is the temperature. The number of attempts a molecule is allowed to make is used as a parameter to control the average sizes of the molecular aggregates. For DBP molecules, there is an extra rotational attempt where one end of the DBP molecule stays in its original site while the other end attempts to move to an adjacent vacant site, or exchange with an adjacent C70. To account for the many-body dynamics during deposition, full sample annealing occurs after every 1,000 molecules are deposited. During this step, the 1,000 molecules are allowed to move to adjacent sites. The attempts follow the same rules as in the single molecule deposition process.

10

ACS Paragon Plus Environment

Page 11 of 21 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

Full sample annealing significantly reduces the computational intensity compared with real-time simulations in which the dynamics of all molecules are simultaneously considered. This enables the simulation of systems with a very large number of molecules (~ 106). The morphology of the simulated blend is evaluated by determining the C70 aggregate size. Spheres with increasing diameters are compared with the C70 cluster shape. The diameter of the largest sphere that can fit within the C70 cluster with a volume shape mismatch below 5% is used to characterize the C70 aggregate size. From X-ray diffraction data, the diffracted intensity from a spherical cluster containing N molecules is proportional to N2. Therefore, the average radius of a simulated blend is derived using an average with weight of r6, and the standard deviation is given by the error bars.

Acknowledgements We gratefully thank Dr. Kai Sun and Dr. Haiping Sun for help on STEM measurements, Dr. Allen Hunter for help on FIB, and Dr. Yongxi Li, Yue Qu, Quinn Burlingame, Anurag Panda, Dejiu Fan, Xiaozhou Che, Jongchan Kim and Caleb Coburn for helpful discussions. This work was partially supported by the Office of Naval Research (K. D. simulations, experiment and analysis, X. L. experiments and analysis, S. R. F., analysis) and the University of Michigan College of Engineering. This work also received technical support from the Michigan Center for Materials Characterization.

11

ACS Paragon Plus Environment

Nano Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

References 1.

Zhao, W.; Li, S.; Yao, H.; Zhang, S.; Zhang, Y.; Yang, B.; Hou, J. Journal of the

American Chemical Society 2017, 139, (21), 7148-7151. 2.

Shockley, W.; Queisser, H. J. Journal of Applied Physics 1961, 32, (3), 510-519.

3.

Chen, G.; Sasabe, H.; Wang, Z.; Wang, X.-F.; Hong, Z.; Yang, Y.; Kido, J. Advanced

Materials 2012, 24, (20), 2768-2773. 4.

Pandey, R.; Zou, Y.; Holmes, R. J. Applied Physics Letters 2012, 101, (3), 033308.

5.

Sun, Y.; Welch, G. C.; Leong, W. L.; Takacs, C. J.; Bazan, G. C.; Heeger, A. J. Nat

Mater 2012, 11, (1), 44-48. 6.

van der Poll, T. S.; Love, J. A.; Nguyen, T.-Q.; Bazan, G. C. Advanced Materials 2012,

24, (27), 3646-3649. 7.

Xiao, X.; Bergemann, K. J.; Zimmerman, J. D.; Lee, K.; Forrest, S. R. Advanced Energy

Materials 2014, 4, (7), 1301557-n/a. 8.

Xiao, X.; Zimmerman, J. D.; Lassiter, B. E.; Bergemann, K. J.; Forrest, S. R. Applied

Physics Letters 2013, 102, (7), 073302. 9.

Deibel, C.; Strobel, T.; Dyakonov, V. Advanced Materials 2010, 22, (37), 4097-4111.

10.

Liu, X.; Ding, K.; Panda, A.; Forrest, S. R. ACS Nano 2016, 10, (8), 7619-7626.

11.

Liu, Y.; Zojer, K.; Lassen, B.; Kjelstrup-Hansen, J.; Rubahn, H.-G.; Madsen, M. The

Journal of Physical Chemistry C 2015, 119, (47), 26588-26597. 12.

Guan, Z.; Li, H.-W.; Cheng, Y.; Yang, Q.; Lo, M.-F.; Ng, T.-W.; Tsang, S.-W.; Lee, C.-S.

The Journal of Physical Chemistry C 2016, 120, (26), 14059-14068. 13.

Lee, C.-C.; Su, W.-C.; Chang, W.; Huang, B.-Y.; Liu, S.-W. 2014, 16.

14.

Lassiter, B. E.; Zimmerman, J. D.; Panda, A.; Xiao, X.; Forrest, S. R. Applied Physics

Letters 2012, 101, (6), 063303. 15.

Forrest, S. R. Chemical Reviews 1997, 97, (6), 1793-1896.

16.

Fujita, T., U.S. Patent 20,080,241,592, 02 Oct 2008: 2008.

17.

Peumans, P.; Uchida, S.; Forrest, S. R. Nature 2003, 425, (6954), 158-162.

18.

Yang, F.; Forrest, S. R. ACS Nano 2008, 2, (5), 1022-1032.

12

ACS Paragon Plus Environment

Page 12 of 21

Page 13 of 21 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

Figure Captions Figure 1. (a) Bright field cross-sectional scanning transmission electron microscopic image of 5 pairs of 2-nm-thick alternating layers capped by a 20-nm-thick C70 layer and a 150-nm-thick Ag protection layer. (b) Image of 5 pairs of 5-nm-thick alternating layers capped by a 20-nm-thick C70 layer and a 50-nm-thick Ag layer. Figure 2. (a) Room temperature photoluminescence (PL) spectra of multilayer stacks consisting of alternating layers of DBP and C70 and a 1:1 mixed sample. To achieve a total thickness of 40 nm, the multilayer stacks with alternating layer thicknesses of 1 2, 3 and 5 nm containing 20, 10, 7, 4 pairs of DBP and C70 layers, respectively. The spectra are normalized to the same integrated intensity, and are offset for clarity. (b) Photoluminescence peak energies of CT2 states, E2, and calculated Coulombic binding energies vs. layer thickness. Figure 3. (a) Average C70 cluster radius vs. DBP concentration in DBP:C70 blends. The radii obtained from X-ray diffraction (squares, Ref. 10) and KMC simulations (circles) are indicated. Error bars for the X-ray diffraction data arise from uncertainties in linewidths at half maxima. Error bars are standard deviations of the simulated cluster size distributions. (b) Measured (squares) and calculated (circles) E2 based on the simulated nanocrystallite radius, , vs. DBP concentration. The error bars for the PL data arise from Gaussian fits to the CT2 peak while the error bars for the simulated results are standard deviations of the cluster size distributions. Figure 4. (a) Three-dimensional simulation of a 10-nm-deep by 80-nm-thick slice of a 1:8 DBP:C70 blended HJ. Growth is simulated by molecules arriving at top of the surface starting at the anode lying at the bottom. Blue areas are C70 molecules. Green (red) areas are DBP

13

ACS Paragon Plus Environment

Nano Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

molecules that form (do not form) percolating paths to the anode. Note the voids in the thin film. The simulation comprises 600,000 molecules. (b) Three-dimensional simulation of the DBP percolating paths of a 50-nm-deep and 80-nm-thick vertical slice of a blend with 10 mol% DBP in C70. The C70 molecules are not shown. Percolating paths are identified by connecting adjacent DBP molecules with green lines. Figure 5. Two-dimensional density of DBP percolating paths, , in the substrate plane as a function of distance from the anode, z (triangles). An exponential fit to the data is shown by the line. Figure 6. Reverse-biased external quantum efficiency (EQE) spectra of the organic photovoltaic cell with various DBP concentrations in C70, along with DBP and C70 absorption spectra.

14

ACS Paragon Plus Environment

Page 14 of 21

Page 15 of 21 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

TOC Graphic

15

ACS Paragon Plus Environment

Nano Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41

(a)

Page 16 of 21

(b)

C70 DBP

C70

DBP

C70

Fig.ACS1Paragon Plus Environment

(b) 1.47

1.0 5 nm 3 nm 2 nm 1 nm 1:1 Mix

0.9 0.8 0.7 0.6 0.3 0.2 0.1 0.0 600

700

Calculation

1.44

0.4

800

900

1000

-0.03

Experiment

CT2 Peak Energy E2 (eV)

(a) PL Intensity (a.u.)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41

Nano Letters

-0.06

1.41

-0.09

1.38

-0.12

1.35

-0.15

1.32

-0.18

1.29

-0.21

1.26

Wavelength (nm)

1

2

3

4

5

-0.24

Alternating Layer Thickness (nm)

Fig. 2 ACS Paragon Plus Environment

Coulombic Binding Energy(eV)

Page 17 of 21

CT peak energy (eV)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41

(nm)

Nano Letters

6 5 4 3 2 1 1.44 1.40 1.36 1.32 1.28 0%

Page 18 of 21

X-ray diffraction Simulation

(a) 10

20

30

40

50

CT2 peak Simulation

(b) 10%

20%

30%

DBP concentration

Fig. 3 ACS Paragon Plus Environment

40%

50%

Page 19 of 21 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41

(nm)

Nano Letters

(a)

(nm)

80

80

60

60

40

40

20

20

0

10

20

40

60

80

0 100 (nm)

Fig. 4

(b)

0

ACS Paragon Plus Environment

50 0 20 40 60 80 100 (nm)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41

(nm -2)

Nano Letters

0.30 0.25 0.20 0.15 0.10 0.05 0.00 0 10 20 30 40 50 60 70 80

z (nm) Fig. 5 ACS Paragon Plus Environment

Page 20 of 21

Page 21 of 21

10% 20% 6 33% 50% DBP 4 C70

0.6 0.4

2

0.2 0.0 400

500

600

700

Wavelength (nm) Fig. 6

ACS Paragon Plus Environment

0 800

Absorption Coefficient (105 cm-1)

0.8

Reverse-Biased EQE

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41

Nano Letters