Band‐like Charge Photogeneration at a Crystalline ... · as the potential for low-cost fabrication on flexible substrates, but they lag behind their inorganic counterparts in performance

www.advenergymat.de

Full paper

1701494 (1 of 10) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Band-like Charge Photogeneration at a Crystalline Organic Donor/Acceptor Interface

Michael A. Fusella, Alyssa N. Brigeman, Matthew Welborn, Geoffrey E. Purdum, Yixin Yan, Richard D. Schaller, YunHui L. Lin, Yueh-Lin Loo, Troy Van Voorhis, Noel C. Giebink, and Barry P. Rand*

DOI: 10.1002/aenm.201701494

inorganic solar cells. This discrepancy is due in large part to differences in the way that charges are generated in OPVs compared with their inorganic counter-parts. Whereas photon absorption in an inorganic semiconductor leads directly to free charge formation, optical excitation in organic semiconductors results in a tightly bound electron–hole pair, known as an exciton, owing to a low dielectric constant and weak intermolecular elec-tronic coupling. A heterojunction between donor (D) and acceptor (A) molecules is therefore needed to drive exciton disso-ciation and yield separated charges. How-ever, even after electron transfer occurs across the D/A interface, the electron and hole are still Coulombically bound to one another, forming an intermediate charge transfer (CT) state. Overcoming the CT state binding energy, typically on the order of 0.1–0.5 eV[3] results in energy losses[4–6] that fundamentally reduce

the thermodynamic limiting efficiency of OPVs below the Shockley–Queisser limit.[7,8] Despite the seemingly large dif-ferences between organic and inorganic optoelectronic proper-ties, similarities have been shown, particularly when structural order of the molecules is enhanced. Charge transport in dis-ordered organic semiconductors relies on a phonon-assisted

Organic photovoltaic cells possess desirable practical characteristics, such as the potential for low-cost fabrication on flexible substrates, but they lag behind their inorganic counterparts in performance due in part to funda-mental energy loss mechanisms, such as overcoming the charge transfer (CT) state binding energy when photogenerated charge is transferred across the donor/acceptor interface. However, recent work has suggested that crystalline interfaces can reduce this binding energy due to enhanced CT state delocali-zation. Solar cells based on rubrene and C60 are investigated as an archetypal system because it allows the degree of crystallinity to be moldulated from a highly disordered to highly ordered system. Using a postdeposition annealing method to transform as-deposited amorphous rubrene thin films into ones that are highly crystalline, it is shown that the CT state of a highly crystalline rubrene/C60 heterojunction undergoes extreme delocalization parallel to the interface leading to a band-like state that exhibits a linear Stark effect. This state parallels the direct charge formation of inorganic solar cells and reduces energetic losses by 220 meV compared with 12 other archetypal heterojunc-tions reported in the literature.

M. A. Fusella, Y. H. L. Lin, Prof. B. P. RandDepartment of Electrical EngineeringPrinceton UniversityPrinceton, NJ 08544, USA E-mail: [email protected]. N. Brigeman, Y. Yan, Prof. N. C. GiebinkDepartment of Electrical EngineeringThe Pennsylvania State UniversityUniversity Park, PA 16802, USADr. M. Welborn, Prof. T. V. VoorhisDepartment of ChemistryMassachusetts Institute of TechnologyCambridge, MA 02139, USA

Organic Solar Cells

1. Introduction

Organic photovoltaic cells (OPVs) are promising for low-cost energy conversion,[1] and while their power-conversion efficien-cies have steadily climbed, recently reaching 11%,[2] there still remains a wide efficiency gap between OPVs and conventional

The ORCID identification number(s) for the author(s) of this article can be found under https://doi.org/10.1002/aenm.201701494.

G. E. Purdum, Prof. Y.-L. LooDepartment of Chemical and Biological EngineeringPrinceton UniversityPrinceton, NJ 08544, USAProf. R. D. SchallerCenter for Nanoscale MaterialsArgonne National LaboratoryArgonne, IL 60439, USAProf. R. D. SchallerDepartment of ChemistryNorthwestern UniversityEvanston, IL 60208, USAProf. Y.-L. Loo, Prof. B. P. RandAndlinger Center for Energy and the EnvironmentPrinceton UniversityPrinceton, NJ 08544, USA

Adv. Energy Mater. 2018, 8, 1701494

www.advenergymat.dewww.advancedsciencenews.com

© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim1701494 (2 of 10)

hopping mechanism. However, studies conducted on certain highly ordered organic single crystals show band transport of both holes[9] and electrons,[10] mimicking the transport mecha-nisms in inorganic materials. Further, extremely long exciton diffusion lengths of up to 8 µm have been measured in rubrene single crystals.[11] These results suggest that highly ordered organic systems can be key to unlocking other advantages typi-cally only seen with inorganics, such as the absence of CT state energy losses.

Further, it has recently been shown that enhancing the degree of crystallinity in certain organic systems can enable CT states to delocalize over multiple molecules.[12] Much debate exists about the role of CT state delocalization in charge gen-eration and whether or not higher-order “hot” CT states—believed to be more delocalized—are necessary precursors to the separated charge state.[13–15] In this work, we investigate a highly crystalline system of rubrene and C60 and observe for the first time the presence of a dense distribution of long-range CT states that is band-like in nature, yielding a binding energy

low enough to parallel the process of direct charge formation in inorganic materials and lead to lower energetic losses at the heterojunction.

2. Results and Discussion

Rubrene thin films deposited via thermal evaporation are amor-phous. We have reported previously[16] a simple heat treatment to induce crystallization of these amorphous films. Incorpo-ration of a thin (5 nm) layer of tris[4-(5-phenylthiophen-2-yl)phenyl]amine (TPTPA) prior to rubrene deposition further improves the crystallization process[17] and allows access to pinhole-free films with crystalline domains exceeding 1 mm in size, as seen in the polarized optical microscope image in Figure 1a. The uniform coloring of the grains indicates a single molecular orientation within the domain. The atomic force microscope image in Figure 1b shows that growing C60 atop this crystalline rubrene template layer results in plate-like


Figure 1. a) Polarized optical microscope image of the crystalline rubrene template layer showing single-colored crystal domains occasionally exceeding 1 mm in size. Scale bar is 500 µm. b) Atomic force microscope image of 5 nm of C60 grown on the crystalline rubrene template, showing columnar grains on the order of 200 nm in size. Scale bar is 1 µm. c) GIXD data of C60 grown on the crystalline rubrene template showing that the rubrene molecules (reflections identified in white) adopt the orthorhombic crystal structure with the (h00) planes parallel to the substrate, and C60 (reflections identified in red) takes on a highly oriented, face-centered cubic phase with the (111) plane parallel to the substrate. d) Top view of the calculated lowest-energy alignment of the crystalline rubrene and crystalline C60 lattices at the interface.



crystalline C60 domains with lateral dimensions on the order of 200 nm that adopt the crystal structure of bulk C60 (ref. [18]). The grazing-incidence X-ray diffraction (GIXD) pattern in Figure 1c of C60 grown atop the crystalline rubrene template shows that rubrene (reflections identified in white) adopts its orthorhombic crystal structure with its (h00) planes preferentially organized parallel to the substrate while C60 (reflections identified in red) adopts a highly oriented face-centered cubic crystal structure with its (111) plane parallel to the substrate.[19,20] These data provide ensemble-averaged structural information for each of the heterojunction materials; to ascertain how rubrene and C60 orient at the molecular scale at the D/A interface, we carried out molecular mechanics simulations whose calculated lowest-energy configuration of this interface is shown in Figure 1d (additional simulation results can be found in Figures S1–S4 in the Supporting Information).

External quantum efficiency (EQE) measurements allow us to visualize photocurrent generation from the CT state at energies less than that of the Frenkel molecular absorption. Since C60 defines the low-energy absorption edge in this heterojunction

pair, the measured EQE of a single-layer C60 device (Figure 2a) shows that the photocurrent produced below 1.6 eV is a result of direct excitation of the CT state (EQE and absorption spectra for single-layer amorphous and crystalline rubrene control devices are shown in Figure S5 in the Supporting Information). Amorphous rubrene planar heterojunction (PHJ) and bulk het-erojunction (BHJ) devices with C60 show a single CT state with a typical Gaussian lineshape centered at ≈1.6 eV. Data fitting yields a CT state energy (ECT) of 1.48 eV and reorganization energy (λ) of 0.11 eV for the amorphous PHJ device, in agree-ment with previous work on amorphous rubrene/C60 solar cells.[21] However, the CT state photocurrent from a crystalline rubrene/C60 PHJ (“crystal PHJ”) device is much broader and does not resemble the Gaussian lineshape observed in many previous studies[21,22] that characterizes a local CT excitation between neighboring D and A molecules.

To study the impact of local structural order on CT state behavior, we investigate a method by which to modulate the degree of crystallinity of the heterojunction between these extreme cases of order (crystalline C60 atop crystalline rubrene)


Figure 2. a) Comparison of the CT state region for the highly ordered (“crystal PHJ” where X = 0 nm) and disordered systems. The EQE of a single-layer C60 device is included to show that photocurrent below ≈1.6 eV results from direct excitation of CT states. b) Device stack showing a certain thickness (X) of a rubrene:C60 blend grown atop the crystal template. c) Modulation of the disordered (≈1.55 eV) CT state prominence as a function of the thickness of the disordered rubrene:C60 blend. The thicker the disordered blend, the more prominent the disordered CT state becomes. Arrows identify the high-energy (dashed arrow) and low-energy (solid arrow) CT features. d) EL shows a single, narrow peak for the crystal PHJ device and a broader, blue-shifted peak for the amorphous BHJ device. The dashed curves are EQE data for the respective devices. The amorphous BHJ device includes a Marcus fit to the EQE data (dotted curve).



and disorder (C60 atop amorphous rubrene). Despite the fact that pristine rubrene adlayers will propagate the crystallinity of the crystalline rubrene template,[23] Bragg–Brentano X-ray diffraction reveals that a 1:1 blend of rubrene:C60 grown on top of the crystalline rubrene template is largely disordered (Figure S6a, Supporting Information). Therefore, the interfacial disorder will increase with increasing thickness of this disor-dered blend until any remaining exposed areas of the ordered crystal PHJ interface are covered.

A schematic of the device stack is shown in Figure 2b. The thickness of this disordered blend, X, is swept from 0 to 100 nm and the resulting EQE measurements of the CT state behavior are shown in Figure 2c (a finer thickness sweep is shown in Figure S6b in the Supporting Information). As X increases, we see a distinctly parabolic shape centered at 1.6 eV rise in prominence. Remarkably, for the thickest values of X, the peak EQE for this CT state exceeds 1.5%, which, to our knowledge, is the highest-ever reported EQE for a CT state. Even when X = 2 nm, the high-energy feature (≈1.6 eV, Figure 2c dashed arrow) has greatly enhanced prominence, producing nearly two orders of magnitude higher photocurrent, but the low-energy tail (≈1.1 eV, Figure 2c, solid arrow) remains largely unchanged, indicating that these features possess a different origin. Since the CT states in the amorphous rubrene devices are energeti-cally aligned with the higher-energy (Figure 2c, dashed arrow) CT feature that appears for X > 0 nm, it suggests that the higher-energy feature is a result of disorder at the heterojunc-tion. The interfacial disorder (and hence prominence of the higher-energy feature) saturates when X > 20 nm (Figure S6b, Supporting Information). As witnessed by the dramatic rise in prominence of the disordered (Figure 2c, dashed arrow) CT fea-ture after depositing only 2 nm of the disordered blend on top of the crystal template, this feature is extremely sensitive to the amount of interfacial disorder.

Electroluminescence (EL) measurements of the crystal PHJ and amorphous BHJ devices (Figure 2d) show a single, narrow CT state emission peak for the crystal PHJ device and a broader, blue-shifted peak for the amorphous BHJ device. The energy offset between the two peaks is consistent with a decreased open-circuit voltage (VOC) when comparing the crys-talline and amorphous devices[16] due to a 0.5 eV highest-occu-pied molecular orbital level shift,[24] and it has been shown that VOC follows the CT state energy[21,22] and can be influenced by morphology.[25] Shown overlaid in Figure 2d as dashed curves are the corresponding CT state EQE measurements for the two devices. The center of the Gaussian fit shown for the amor-phous BHJ device occurs at 1.6 eV and is blue shifted from the CT state emission peak by about 0.2 eV, roughly twice the reor-ganization energy of 0.12 eV, as expected from Marcus theory. Relaxation within the disorder-induced CT density of states may also contribute to the energy shift. In contrast, the CT state emission peak in the crystal PHJ case is nearly coincident with the CT absorption onset inferred from the EQE, a situation that is difficult to rationalize on the basis of Marcus theory (since it would imply a vanishing reorganization energy), instead appearing closer in form to the band-edge transitions character-istic of direct-gap inorganic semiconductors.

To explore this possibility, we sought to access CT state absorption in the crystal PHJ device directly through anisotropy

in the transverse electric- (TE) and transverse magnetic- (TM) polarized EQE. At an abrupt crystalline PHJ, the CT state tran-sition dipole moment is expected to be predominantly aligned orthogonal to the interface, leading to preferential absorption of TM-polarized light that provides a means of isolating the CT state from the isotropic excitonic background absorption. Figure 3a shows that this is indeed observed for the crystal PHJ device in the vicinity of the low-energy CT feature at ≈1.1 eV. In contrast, the same measurements carried out on an amor-phous PHJ device in Figure 3b display no such anisotropy since molecular-scale intermixing at the interface largely randomizes the CT state orientation.

The polar plot in Figure 3c summarizes the EQE anisotropy at a single photon energy in the CT region for devices with (X = 100 nm and X = 4 nm) and without (X = 0 nm) the disor-dered blend layer. For the X = 100 nm and X = 4 nm devices, the absorption at ≈1.55 eV (corresponding to the disordered CT state in the blend) is isotropic, with slight deviations due to optical interference effects, thereby confirming the disordered origin of this CT feature. Crucially, however, the low-energy CT state at ≈1.1 eV in the crystal PHJ and the X = 4 nm devices must be of different origin because it is strongly anisotropic, with a nearly cos2 θ polarization pattern indicative of well-aligned transition dipoles.

The anisotropic optical response of the low-energy CT state provides a way to explicitly isolate its photocurrent contribu-tion by eliminating common-mode isotropic absorption via the difference, EQETM–EQETE (ref. [26]). The result plotted in Figure 3d reveals a step function-like CT state absorption spec-trum in the range 1–1.6 eV that is clearly different from the Gaussian lineshape observed for other localized CT excitations using this same method.[26] Instead, the shape of the absorp-tion spectrum in Figure 3d is reminiscent of that observed for 2D electronic systems such as inorganic quantum wells and 2D transition metal dichalcogenides[27–30] that have step func-tion-like density of states (DOS) functions, which is the same dimensionality expected for an extended CT state that is Cou-lombically confined to the interface. In this context, we applied a 2D DOS direct bandgap model[31] to the EL and polarized EQE subtraction data of the crystal PHJ device (see the Supporting Information for model details). Using this reciprocity approach, we simultaneously fit these two data sets to yield a shared set of parameters (see inset of Figure 3d) and found that such a model provides a consistent description of the data. Compared with a CuZnSnSSe solar cell,[31] the crystal PHJ device has a ≈20% smaller energy broadening parameter, γ, underscoring the narrowness of its emission peak as a result of the highly ordered system.

To further investigate the CT state delocalization and band-like characteristics, we applied quantum chemical calculations to the ordered and disordered systems under investigation. Due to random molecular orientations in the disordered system, we model the disordered CT state by considering single rubrene and C60 molecules in solution. We take the minimum energy geometry to be the most likely dimer configuration (Figure 4a) and compute a calculated CT state energy of 1.61 ± 0.2 eV in good agreement with the Marcus fit value of ECT = 1.48 eV. Although this simple single-solvated dimer model does not account for the configurational broadening of the CT state




peak, and does not include structural orientation effects of the surrounding disordered interface beyond a mean-field level, it still provides a useful comparison between theory and experi-ment; we see that localization of the CT state on a single repre-sentative rubrene/C60 dimer is sufficient to explain the energy of the center of the disordered CT state peak. Upon incorpo-rating additional rubrene molecules with the interfacial align-ment shown in Figure 1d based on the experimental molecular orientation of the crystalline rubrene and crystalline C60 films, we calculate that the lowest-energy crystalline CT state is char-acterized by a hole wavefunction that is delocalized over sev-eral rubrene molecules along the interface (Figure 4b). We note that provisional calculations did include an additional C60 molecule, but the results of the calculations were unchanged. The correspondence of this particular CT state to that identi-fied in experiment is supported by the computed DOS distribu-tion, which is in reasonable agreement with the experimental CT EL spectrum. Delocalization is driven along the interface by the large π-orbital overlap along the parallel rubrene c-axis

and is limited mainly by the binding electron in C60. While the hole-only system is fully delocalized, the presence of the elec-tron on C60 breaks translational symmetry, localizing the hole wavefunction. Further, the C60 molecule reduces the rubrene–rubrene hole-hopping coupling along the c-axis from 0.14 to 0.09 eV. However, we note that it has been shown in the litera-ture that crystallization can lead to delocalization on either the donor[13] or acceptor[12,15] side, with varying effects on CT state behavior.

The spatial character of the crystalline CT state is also evi-dent from electroabsorption (EA) measurements in the CT energy range with an electric field applied normal to the het-erojunction in reverse biased devices (EA measurements on single-layer control devices can be found in Figure S7 in the Supporting Information). The data in Figure 5a show that the X = 100 nm device exhibits a single CT feature at ≈1.55 eV, while the crystal PHJ (X = 0 nm) device exhibits a single CT feature at ≈1.1 eV, confirming that the high-energy (disordered) CT state is not present in this device. The X = 4 nm device displays both


Figure 3. a) Polarized EQE spectra for the crystal PHJ device showing a large degree of CT state anisotropy at ≈1.11 eV indicating that the dipole orientation preferentially absorbs TM-polarized light. b) Polarized EQE spectra for an amorphous rubrene/C60 PHJ device showing no polarization dependence for the CT feature at ≈1.55 eV, thus indicating a random orientation of CT state dipoles leading to isotropy. c) Polar plot summarizing the CT state anisotropy for devices with varying thicknesses of the disordered blend, X, showing excitation of the high-energy (disordered, isotropic) CT state, and low-energy (ordered, anisotropic) CT state band. d) The difference (blue circles) between the TM- and TE-polarized EQE data for the crystal PHJ device in (a) yields a band-like CT state spectrum whereby the CT state absorption is flat above the band edge, indicative of a 2D DOS. A band model with shared fit parameters based on a 2D DOS (shown in red solid lines) fits both the polarized EQE subtraction and EL data (black dots).



high- and low-energy CT features in accord with our interpre-tation of the EQE in Figure 2c. Similar to the EQE, the low-energy CT EA response is strongly TM-polarized, whereas the higher-energy disordered CT EA feature is largely independent of polarization. Figure 5b shows that, in contrast to the usual quadratic Stark effect exhibited by the disordered CT state, the crystal PHJ EA arises from the linear Stark effect due to its highly oriented net CT dipole moment.[32] In this case, a uniform electric field, F, applied in reverse bias couples to the orientation-averaged net CT dipole moment and leads to a decrease in its energy given by ΔECT = −⟨F · µCT⟩, which leads to a red-shift of the CT absorption spectrum[32] and thus to the first derivative differential absorption lineshape shown in Figure 5a. This red-shift is directly evident in the crystal PHJ EQE spectra recorded for different reverse bias fields shown in Figure S8 in the Supporting Information. Normalizing the EQE spectra to account for improved charge collection in reverse bias reveals a ≈10 meV red-shift of the crystal PHJ CT absorption edge in Figure 5c that is absent from the disordered CT absorption in Figure 5d. This red-shift enables the crystal PHJ CT dipole moment normal to the interface to be determined[32] as 31.8 ± 12.6 D, which is consistent with the calculated value of 56.0 D that is determined from the calculated CT state charge density distribution in Figure 4b. Meanwhile, the dipole moment for

the X = 100 nm device, as determined via EA measurements,[32] yields a smaller value of 8.4 ± 2.4 D, thereby indicating that the average electron–hole separation in the crystal PHJ CT state is significantly larger than that of the disordered CT state. EA car-ried out with the electric field applied in-plane should alterna-tively probe the substantial CT delocalization predicted along the D/A interface; however, it has so far proven challenging to isolate signal from noise in this measurement configuration, preventing a direct comparison with the simulated results.

Our computational models provide important implications for material selection and reducing energy losses in OPV devices. To begin, we compute a CT exciton binding energy of ≈0.1 eV in the crystal PHJ device, substantially lower than the ≈0.6 eV CT exciton binding energy computed for the amor-phous devices. The CT state binding energy is computed as the difference between the lowest energy (bound) state and the lowest energy unbound state of the lattice model. This analysis suggests a substantially reduced energy loss mechanism for the crystal PHJ device compared with the amorphous devices, which is confirmed via temperature-dependent VOC meas-urements (Figure 6). In the amorphous PHJ device, we find that VOC increases linearly with decreasing temperature until ≈225 K where it saturates at 1.08 V. In the crystal PHJ device, however, there is no VOC saturation down to 125 K at which


Figure 4. a) The calculated lowest-energy ground state molecular geometry for single rubrene and C60 molecules in solution and the charge distribution across the molecules for a charge transfer state in this orientation. The calculated CT state energy for this configuration is 1.61 ± 0.2 eV in line with the Marcus fit value of ECT = 1.48 eV for the amorphous BHJ system. b) Diagram showing delocalized positive charge across seven rubrene molecules in the lowest-energy crystalline CT state. c) The computed CT state band DOS (solid line) is in good agreement with the measured EL data for the crystal PHJ device.



point, the VOC is 0.81 V. The saturation of VOC at low tempera-ture has previously been interpreted as a result of decreased CT state dissociation efficiency and therefore suggests that charge separation proceeds more easily at the crystal PHJ than at the amorphous PHJ[33] due to a smaller CT state binding energy in the former. This smaller binding energy manifests directly as a reduced loss in potential relative to the CT-defined optical absorption energy gap. To make this comparison, we con-sider the difference between the highest measured VOC for each device (at low temperature) relative to its CT state energy (1.48 eV, amorphous PHJ) or CT band gap (1.07 eV, crystal PHJ) measured at room temperature. These calculations yield differ-ences of 0.40 eV for the amorphous PHJ and 0.26 eV for the crystal PHJ, proving that the crystal PHJ has a reduced energy loss of at least 140 meV compared with the amorphous device. We emphasize that this number represents a lower bound to the amount of energy loss reduction since the crystal PHJ VOC still has not saturated even at the lowest temperatures we can probe.

It has been suggested[4,6,22] that energy losses for a D/A inter-face can be quantified by the difference between the room tem-perature qVOC and ECT at 0 K, a metric that is normally between

0.5 and 0.7 eV, and, further, that linear extrapolations of qVOC(T) and ECT(T) converge at 0 K. Indeed, using this method, the D/A systems, both polymer and small molecule, listed in Table 1 have an average energy loss of 0.57 eV, with a standard deviation of 38 meV. Remarkably, the crystal PHJ device pre-sents a difference qVOC(0 K)–qVOC(295 K) = 0.35 eV, thereby exhibiting a reduced energy loss of 220 meV. The crystal PHJ energy loss is significantly smaller than it is for the amorphous control device (0.54 eV) and represents an energy loss nearly six standard deviations below the mean of the material systems in Table 1. While the VOC of the crystalline rubrene/C60 device is lower than that of the amorphous device, the reduced energy loss indicates that an alternative acceptor with a shallower lowest unoccupied molecular orbital level than C60 could be used to reduce the gap between VOC and the Frenkel absorption edge even further to realize higher power conversion efficiency. These exceptional results confirm that structural order is an important design principle to reduce energy loss and allow for more efficient OPVs.

Finally, we consider alternative crystalline interfaces between rubrene and C60 (Figure S4, Supporting Information) in order to understand the role of molecular interfacial orientation in these


Figure 5. a) EA measurements of the X = 100, 4, and 0 nm devices displaying either a high-energy CT feature (X = 100 nm), low-energy CT feature (X = 0 nm), or both features (X = 4 nm). The red-dashed curve is a fit to the high-energy EA feature, taken as the second derivative of the Gaussian fit to the high-energy CT feature in EQE. b) EA measurements as a function of DC bias show a linear dependence for the high-energy CT feature (X = 100 nm device), whereas the low-energy feature (X = 0 nm device) is independent of DC bias. c) Normalizing the biased EQE spectra (measured with TM-polarized light) for the X = 0 nm device displays a red-shift of the CT absorption onset with increasing reverse bias. d) Normalizing the biased EQE spectra for the X = 100 nm device shows that the applied field does not significantly alter the CT state.


© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim1701494 (8 of 10)Adv. Energy Mater. 2018, 8, 1701494

devices. We see that both exciton binding energy and transition dipoles are governed by the distance between rubrene mol-ecules along the direction of the interface; more closely packed rubrene molecules lead to a lower exciton binding energy and higher oscillator strength, while simultaneously creating higher hole mobility. Thus, the experimental interface realized in this work is optimal because it aligns the close-packed, highly conductive c-axis of rubrene along the interface. As a result, among the various interfaces modeled, band-like charge pho-togeneration was unique to the experimental system. Since rubrene single crystals are known to exhibit band-like charge transport,[9] the question arises as to whether or not band-like

charge transport is a prerequisite to band-like charge pho-togeneration. Though we are not yet able to fully answer this question, we have found that pentacene, which also possesses band-like charge transport,[34] does not exhibit band-like charge photogeneration of the CT state when paired with C60 (ref. [26]).

3. Conclusion

In summary, we have found a domain where OPVs not only work in a delocalized fashion but also, through the existence of a band-like CT state, can dissociate interfacial CT excitons into free charges in a process that parallels inorganic semi-conductors where formation of charge is a direct process. The results presented here distinguish the CT state of the crystal PHJ device from previous reports in that the delocalization is so great that the CT state exhibits band-like properties, such as narrow band-edge emission, a consistent description of emis-sion and absorption via a band model, and reduced energy loss in solar cell devices. Using the design guidelines presented here, future high-efficiency OPVs should utilize molecules with high degrees of structural order and interfacial orientations that maximize charge delocalization. This opens the door for organic solar cells to operate in a new regime in which energy losses are reduced by avoiding the need to overcome large CT state binding energies in order to produce free charges. Fur-ther, this report underscores the importance of the local envi-ronment at the D/A interface on charge photogeneration in organic solar cells.

4. Experimental SectionGlass substrates prepatterned with indium tin oxide (ITO) were cleaned via ultrasonication in deionized water, acetone, and isopropanol before undergoing an O2 plasma treatment. All organic materials were sourced from commercial vendors and purified via thermal gradient sublimation prior to use: TPTPA (Lumtec), rubrene (Nichem), C60 (Nano-C), and bathocuproine (BCP, Nichem). All layers were deposited via thermal evaporation with a base pressure of ≈3 × 10−7 Torr. The following evaporation rates were used: TPTPA (0.5 Å s−1), rubrene (1 Å s−1), rubrene:C60 (0.3:0.3 Å s−1), C60 (1 Å s−1), BCP (1 Å s−1), and Al (1.5 Å s−1). To fabricate the crystalline samples, after the initial 22 nm of rubrene deposition, the samples were removed from vacuum and heated at 141 °C for 7 min in a clean, dry N2 environment to facilitate rubrene crystallization. The samples were then reintroduced into the vacuum chamber where the subsequent layers were deposited. The structure of the amorphous PHJ device was: ITO/rubrene(20 nm)/C60(20 nm)/BCP(10 nm)/Al(100 nm). The structure of the amorphous BHJ device was: ITO/rubrene:C60[1:1](40 nm)/BCP(10 nm)/Al(100 nm). For polarized EQE and EL measurements, samples were encapsulated with a N2 atmosphere using a solvent-free epoxy and a glass cap.

Polarized optical microscope images were taken with an Olympus BX60F5, and atomic force microscopy images were taken with a Veeco Innova, model 840-012-711.

GIXD was collected at the G1 station at the Cornell High Energy Synchrotron Source. The X-ray beam (9.95 ± 0.05 keV) was selected to be 0.05 mm tall and 1 mm wide. The beam was aligned above the critical angle of the film but below that of the substrate, at a 0.18° incident angle. GIXD images were collected on a 2D charge coupled device (CCD) detector positioned 111.6 mm from the sample. Due to the short sample-to-detector distance, the width of the samples was kept at 5 mm to reduce geometric smearing. 2D GIXD images

Table 1. Energy losses from the CT state to VOC.

Material system ECT–qVOC [eV]

Crystalline rubrene/C60 0.35

Amorphous rubrene/C60 0.54

6T/DIP (ref. [4]) 0.62

Pentacene/C60 (ref. [4]) 0.60

DIP/C60 (ref. [4]) 0.51

CuPc/C60 (ref. [4]) 0.57

6T/C60 (ref. [4]) 0.52

APFO3:PCBM 1:4 (ref. [6]) 0.59

APFO3:PCBM 1:1 (ref. [6]) 0.59

P3HT:PCBM 1:1 (ref. [6]) 0.53

RRa P3HT:PCBM 1:4 (ref. [6]) 0.61

MDMO-PPV:PCBM 1:4 (ref. [6]) 0.58

DIP = diindenoperylene, CuPc = copper phthalocyanine, APFO3 = poly[2,7- (9-di-octyl-fluorene)-alt-5,5-(4′,7′-di-2-thienyl-2′,1′,3′ benzothiadiazole], PCBM = [6,6]-phenyl-C61-butyric acid methyl ester, P3HT = poly[3-hexylthiophene], MDMO-PPV = poly[2-methoxy-5-(30,70-dimethyloctyloxy)-1,4-phenylene vinylene].

Figure 6. Temperature-dependent open-circuit voltage for the crystal and amorphous PHJ devices. Open-circuit voltage linearly increases (solid lines are guides to the eye) as temperature decreases, but the amorphous PHJ saturates below 225 K whereas the crystal PHJ does not, even down to 125 K, indicating that charge separation occurs more easily in this device due to increased delocalization and reduced binding energy.



have been background subtracted, and polarization and absorption corrections were applied.

EQE measurements were taken using a Newport TLS-300X tunable light source system. Measurements were performed under short-circuit conditions using a Stanford Research Systems SR830 lock-in amplifier and SR570 current preamplifier to detect the photocurrent produced by the device from the incident chopped (110 Hz) light. Calibrated Si and Ge photodetectors (Newport) served as reference cells.

Steady-state EL measurements were collected under constant current bias of 250 mA cm−2 for the X = 100 nm device and 1 A cm−2 for the crystal PHJ device using a Horiba iHR fiber-coupled spectrometer with an air-cooled Si CCD array and a N2-cooled InGaAs array. All figures present EL data with units of normalized photons/area × time × bandwidth.

Polarized EQE measurements were performed using an Energetiq laser-driven Xe light source passed through a Princeton Instruments SP2150 monochromator and chopped at a frequency of 400 Hz. Synchronous photocurrent was detected using an Amtek 7230 lock-in amplifier and SR570 current preamplifier. Monochromatic light was passed through a broadband wire grid polarizer (Edmund Optics) fixed to a precision rotation mount to define the polarization of the beam which was then incident on the device at an angle of 30°. Calibrated Newport Si and Ge photodetectors were used as reference cells. Biased EQE measurements were similarly obtained using TM-polarized light and the current preamplifier to hold the devices in reverse bias. EA measurements were performed using the same system. TM-polarized light was incident on the device and the reflection off the Al cathode (i.e., incident light passed through the active layer twice) was collected with calibrated Si and Ge photodetectors. Spectra were obtained by depleting all devices with a −1.5 V reverse bias and a 3 V peak-to-peak, f = 400 Hz sinusoidal voltage to measure the change in reflected light. EA as a function of reverse DC bias was collected using a constant AC voltage of 3 V peak-to-peak.

Bragg–Brentano X-ray diffraction measurements were performed on a Bruker D8 Discover diffractometer with CuKα source, parallel mirror optics, 0.6 mm divergence slits, and a point scintillation detector.

For low-temperature VOC measurements, devices were mounted in a Janis ST-100 liquid–nitrogen cryostat. Cold finger and device temperatures were controlled and monitored with a LakeShore 335 Cryogenic Temperature Controller. A Si diode mounted directly on to the device substrate was used to carefully monitor device temperature throughout all measurements. VOC was measured using a Keithley 2400 SourceMeter at cold finger temperatures ranging from 77 K to room temperature. A 405 nm laser diode was used as the light source, and incident intensity was tuned using a continuous neutral density filter to yield a VOC similar to one-sun conditions at room temperature for each device. Intensity was monitored during measurements with a calibrated Si Photodiode and ThorLabs digital power meter.

Density functional theory calculations were performed using the Q-Chem 4.3 package[35] with the B3LYP functional[36] in a 6–31 g* basis set.[37] Charge transfer state energies, couplings, and transition dipole moments were computed using constrained density functional theory configuration interaction (CDFT-CI).[38] CT state energies computed with CDFT-CI were expected to be accurate to within 0.2 eV. Dielectric effects were simulated using the conducting polarizable continuum model.[39]

Molecular dynamics calculations were performed using the GROMACS package.[40] The Lennard-Jones parameters for carbon were σ = 0.3431 nm and ε = 0.4396 kJ mol−1. The Lennard-Jones parameters for hydrogen were σ = 0.2571 nm and ε = 0.1842 kJ mol−1. The Lennard-Jones parameters were taken from the OPLS-aa force field for aromatic carbon and for hydrogen bound to an aromatic carbon.[41] Further computational details can be found in the Supporting Information.

Supporting InformationSupporting Information is available from the Wiley Online Library or from the author.

AcknowledgementsThe authors thank R. A. Kerner for assistance with Bragg–Brentano X-ray diffraction measurements. The authors acknowledge support for this work from the U.S. Department of Energy, Office of Basic Energy Sciences, and Division of Materials Sciences and Engineering under award numbers DE-SC0012458 and DE-SC0012365. The theory work on this project was supported as part of the Center for Excitonics, an Energy Frontier Research Center funded by the US Department of Energy, Office of Science, and Office of Basic Energy Sciences (award no. DE-SC0001088, MIT). G.E.P. was supported by the Department of Defense (DoD) through the National Defense Science and Engineering Graduate Fellowship (NDSEG) Program. CHESS is supported by the NSF & NIH/NIGMS via NSF award DMR-1332208. Y.-L.L. acknowledges support by the National Science Foundation (NSF) Materials Research Science and Engineering Center program through the Princeton Center for Complex Materials (DMR-1420541). This work was performed, in part, at the Center for Nanoscale Materials, a U.S. Department of Energy Office of Science User Facility, and supported by the U.S. Department of Energy, Office of Science, under Contract No. DE-AC02-06CH11357.

Conflict of InterestThe authors declare no conflict of interest.

Keywordscharge transfer state, crystal, delocalized, rubrene, solar cells

Received: May 31, 2017Revised: August 7, 2017

Published online: December 8, 2017

[1] S. R. Forrest, Nature 2004, 428, 911.[2] M. A. Green, Y. Hishikawa, W. Warta, E. D. Dunlop, D. H. Levi,

J. Hohl-Ebinger, A. W. Y. Ho-Baillie, Prog. Photovoltaics 2017, 25, 668.

[3] T. M. Clarke, J. R. Durrant, Chem. Rev. 2010, 110, 6736.[4] U. Hörmann, J. Kraus, M. Gruber, C. Schuhmair, T. Linderl, S. Grob,

S. Kapfinger, K. Klein, M. Stutzman, H. J. Krenner, W. Brütting, Phys. Rev. B 2013, 88, 235307.

[5] M. Gruber, J. Wagner, K. Klein, U. Hörmann, A. Opitz, M. Stutzmann, W. Brütting, Adv. Energy Mater. 2012, 2, 1100.

[6] T. M. Burke, S. Sweetnam, K. Vandewal, M. D. McGehee, Adv. Energy Mater. 2015, 5, 1500123.

[7] N. C. Giebink, G. P. Wiederrecht, M. R. Wasielewski, S. R. Forrest, Phys. Rev. B 2011, 83, 195326.

[8] R. A. J. Janssen, J. Nelson, Adv. Mater. 2012, 25, 1847.[9] V. Podzorov, E. Menard, A. Borissov, V. Kiryukhin, J. A. Rogers,

M. E. Gershenson, Phys. Rev. Lett. 2004, 93, 086602.[10] N. A. Minder, S. Ono, Z. Chen, A. Facchetti, A. F. Morpurgo, Adv.

Mater. 2012, 24, 503.[11] H. Najafov, B. Lee, Q. Zhou, L. C. Feldman, V. Podzorov, Nat.

Mater. 2010, 9, 938.[12] B. Bernardo, D. Cheyns, B. Verreet, R. D. Schaller, B. P. Rand,

N. C. Giebink, Nat. Commun. 2014, 5, 3245.[13] A. A. Bakulin, A. Rao, V. G. Pavelyev, P. H. M. van Loosdrecht,

M. S. Pshenichnikov, D. Niedzialek, J. Cornil, D. Beljonne, R. H. Friend, Science 2012, 335, 1340.

[14] K. Vandewal, S. Albrecht, E. T. Hoke, K. R. Graham, J. Widmer, J. D. Douglas, M. Schubert, W. R. Mateker, J. T. Bloking,



G. F. Burkhard, A. Sellinger, J. M. J. Fréchet, A. Amassian, M. K. Riede, M. D. McGehee, D. Neher, A. Salleo, Nat. Mater. 2013, 13, 63.

[15] S. Gélinas, A. Rao, A. Kumar, S. L. Smith, A. W. Chin, J. Clark, T. S. van der Poll, G. C. Bazan, R. H. Friend, Science 2014, 343, 512.

[16] B. Verreet, P. Heremans, A. Stesmans, B. P. Rand, Adv. Mater. 2013, 25, 5504.

[17] M. A. Fusella, S. Yang, K. Abbasi, H. H. Choi, Z. Yao, V. Podzorov, A. Avishai, B. P. Rand, Chem. Mater. 2017, 29, 6666.

[18] A. L. Briseno, S. C. B. Mannsfeld, M. M. Ling, S. Liu, R. J. Tseng, C. Reese, M. E. Roberts, Y. Yang, F. Wudl, Z. Bao, Nature 2006, 444, 913.

[19] D. L. Dorset, M. P. McCourt, Acta Crystallogr. 1994, A50, 344.

[20] O. D. Jurchescu, A. Meetsma, T. T. M. Palstra, Acta Crystallogr. 2006, B62, 330.

[21] K. R. Graham, P. Erwin, D. Nordlund, K. Vandewal, R. Li, G. O. Ngongang Ndjawa, E. T. Hoke, A. Salleo, M. E. Thompson, M. D. McGehee, A. Amassian, Adv. Mater. 2013, 25, 6076.

[22] K. Vandewal, K. Tvingstedt, A. Gadisa, O. Inganäs, J. V. Manca, Phys. Rev. B 2010, 81, 125204.

[23] M. A. Fusella, F. Schreiber, K. Abbasi, J. J. Kim, A. L. Briseno, B. P. Rand, Nano Lett. 2017, 17, 3040.

[24] Y. L. Lin, M. A. Fusella, O. V. Kozlov, X. Lin, A. Kahn, M. S. Pshenichnikov, B. P. Rand, Adv. Funct. Mater. 2016, 26, 6489.

[25] U. Hörmann, C. Lorch, A. Hinderhofer, A. Gerlach, M. Gruber, J. Kraus, B. Sykora, S. Grob, T. Linderl, A. Wilke, A. Opitz, R. Hansson, A.-S. Anselmo, Y. Ozawa, Y. Nakayama, H. Ishii, N. Koch, E. Moons, F. Schreiber, W. Brütting, J. Phys. Chem. C 2014, 118, 26462.

[26] A. N. Brigeman, M. A. Fusella, Y. Yan, G. E. Purdum, Y.-L. Loo, B. P. Rand, N. C. Giebink, Adv. Energy Mater. 2016, 6, 1601001.

[27] W. Stolz, J. C. Maan, M. Altarelli, L. Tapfer, K. Ploog, Phys. Rev. B 1987, 36, 4310.

[28] K. F. Mak, C. Lee, J. Hone, J. Shan, T. F. Heinz, Phys. Rev. Lett. 2010, 105, 136805.

[29] J. S. Weiner, D. S. Chemla, D. A. B. Miller, T. H. Wood, D. Sivco, A. Y. Cho, Appl. Phys. Lett. 1985, 46, 619.

[30] G. Livescu, D. A. B. Miller, D. S. Chemla, M. Ramaswamy, T. Y. Chang, N. Sauer, A. C. Gossard, J. H. English, IEEE J. Quantum Electron. 1988, 24, 1677.

[31] J. K. Katahara, H. W. Hillhouse, J. Appl. Phys. 2014, 116, 173504.

[32] A. Horvath, G. Weiser, G. L. Baker, S. Etemad, Phys. Rev. B 1995, 51, 2751.

[33] F. Gao, W. Tress, J. Wang, O. Inganäs, Phys. Rev. Lett. 2015, 114, 128701.

[34] O. Ostroverkhova, D. G. Cooke, S. Shcherbyna, R. F. Egerton, F. A. Hegmann, R. R. Tykwinski, J. E. Anthony, Phys. Rev. B 2005, 71, 035204.

[35] Y. Shao, M. Wormit, J. Kussmann, A. W. Lange, A. Behn, X. Feng, D. Ghosh, M. Goldey, P. R. Horn, L. D. Jacobson, I. Kaliman, R. Z. Khaliullin, T. Kus, J. Liu, E. I. Proynov, Y. M. Rhee, R. M. Richard, A. Rohrdanz, R. P. Steele, E. J. Sundstrom, H. L. W. Iii, M. Zimmerman, D. Zuev, B. Albrecht, E. Alguire, B. Austin, G. J. O. Beran, Y. A. Bernard, E. Berquist, K. Brandhorst, B. Bravaya, S. T. Brown, D. Casanova, C.-M. Chang, S. H. Chien, K. D. Closser, D. L. Crittenden, R. A. Distasio, H. Do, A. D. Dutoi, G. Edgar, S. Fatehi, L. Fusti-Molnar, A. Ghysels, A. Golubeva, J. Gomes, M. W. D. Hanson-heine, H. P. Philipp, A. W. Hauser, E. G. Hohenstein, Z. C. Holden, T.-C. Jagau, H. Ji, B. Kaduk, K. Khistyaev, J. Kim, J. Kim, R. A. King, P. Klunzinger, D. Kosenkov, T. Kowalczyk, C. M. Krauter, K. U. Lao, A. E. L. D. Laurent, K. V. Lawler, V. Levchenko, C. Y. Lin, F. Liu, E. Livshits, R. C. Lochan, A. Luenser, P. Manohar, S. F. Manzer, S.-P. Mao, A. V. Marenich, S. A. Maurer, N. J. Mayhall, E. Neuscamman, C. M. Oana, R. Olivares-amaya, P. O. Neill, J. A. Parkhill, T. M. Perrine, R. Peverati, D. R. Rehn, E. Rosta, N. J. Russ, S. M. Sharada, S. Sharma, D. W. Small, A. Sodt, T. Stein, D. Stück, Y.-C. Su, A. J. W. Thom, T. Tsuchimochi, V. Vanovschi, O. Vydrov, T. Wang, M. A. Watson, J. Wenzel, A. White, C. F. Williams, J. Yang, S. Yeganeh, S. R. Yost, I. Y. Zhang, X. Zhang, Y. Zhao, B. R. Brooks, K. L. Chan, D. M. Chipman, C. J. Cramer, W. A. Goddard, M. S. Gordon, W. J. Hehre, A. Klamt, H. F. S. Iii, M. W. Schmidt, C. D. Sherrill, D. G. Truhlar, A. Warshel, X. Xu, A. A. N. Aspuru-guzik, R. Baer, A. T. Bell, N. A. Besley, D. Chai, A. Dreuw, B. D. Dunietz, T. R. Furlani, R. Steven, C.-P. Hsu, Y. Jung, J. Kong, D. S. Lambrecht, W. Liang, C. Ochsenfeld, V. A. Rassolov, V. Lyudmila, J. E. Subotnik, T. V. Voorhis, J. M. Herbert, I. Anna, P. M. W. Gill, M. Head-gordon, Z. Gan, E. Epifanovsky, A. T. B. Gilbert, Mol. Phys. 2015, 113, 184.

[36] A. D. Becke, J. Chem. Phys. 1993, 98, 1372.[37] R. Krishnan, R. Seeger, J. A. Pople, J. S. Binkley, J. Chem. Phys. 1980,

72, 650.[38] Q. Wu, C.-L. Cheng, T. V. Voorhis, J. Chem. Phys. 2007, 127,

164119.[39] D. M. Chipman, J. Chem. Phys. 2000, 112, 5588.[40] M. J. Abraham, T. Murtola, R. Schulz, J. C. Smith, B. Hess,

E. Lindahl, SoftwareX 2015, 1–2, 19.[41] W. L. Jorgensen, D. S. Maxwell, J. Tirado-Rives, J. Am. Chem. Soc.

1996, 118, 11225.

Documents

Band‐like Charge Photogeneration at a Crystalline ... · as the potential for low-cost fabrication on flexible substrates, but they lag behind their inorganic counterparts in performance