Article pubs.acs.org/JPCC
Exciton Dynamics at CuPc/C60 Interfaces: Energy Dependence of Exciton Dissociation G. J. Dutton and S. W. Robey* National Institute of Standards and Technology, 100 Bureau Dr., Gaithersburg, Maryland 20899, United States S Supporting Information *
ABSTRACT: Donor−acceptor interfaces are critical for the operation of organic photovoltaic devices. Exciton dynamics at these interfaces play a significant role in determining efficiency and controlling open circuit voltage (VOC), short circuit current (JSC), or fill factor (FF). These fundamental interfacial dynamical processes are dependent on the interfacial electronic and molecular structure. In this report we use time-resolved two-photon photoemission (TR-2PPE) to investigate exciton dissociation, recombination, and relaxation processes occurring at well-characterized prototypical donor−acceptor interfaces of copper phthalocyanine (CuPc) layers on C60. S1 excitons are created by excitation in the CuPc Q-band. The excited S1 population is probed as a function of time via photoemission with a UV probe pulse. TR-2PPE measurements provide a picture of subpicosecond charge separation and recombination processes as a function of distance from the CuPc/C60 interface, starting with a CuPc single layer. Analysis via rate equation modeling reveals that the bulk intersystem crossing and intraband relaxation occur on picosecond to subpicosecond time scales, resulting in rapid relaxation of the exciton population. At the interface, these processes compete with electron transfer to C60. The rate constant governing exciton dissociation is energy-dependent, decreasing by orders of magnitude for excitons below the energy of interfacial chargetransfer states. Connections to semiclassical models of charge transfer and implications for device performance are discussed.
■
INTRODUCTION Next-generation photovoltaic concepts, particularly organic photovoltaics (OPVs), require interfaces between donor and acceptor components to dissociate the excitons created by optical excitation.1,2 Charge separation and recombination processes at the interface thus play key roles in determining power conversion efficiencies. The relevant time scales for the critical interfacial dynamical processes range from subpicosecond to nanosecond, requiring experimental probes with similar time resolution. Ultrafast optical and terahertz methods have been used to probe exciton dynamics in OPV systems,3−18 but the separation of interfacial contributions from dynamics in the bulk is not always transparent, leading to some controversy over the relative importance of interfacial versus bulk processes.17−22 Theoretical work showed that interfacial molecular orientation has a significant effect on interfacial dynamics,23−25 highlighting the utility of investigating wellcharacterized interfacial systems. Clearly identifying and monitoring interfacial dynamics and discriminating from bulk processes requires ultrafast time resolution coupled to interface sensitivity. Most all-optical pump−probe methods lack the ability to easily distinguish the interfacial response. However, laser-based time-resolved photoemission techniques, such as time-resolved two-photon photoemission (TR-2PPE), can satisfy both requirements. Thus, we report here on results of TR-2PPE studies of exciton dynamics at a well-charaterized organic donor−acceptor interface composed of copper phthalocyanine (CuPc) and C60. Previous work amply demonstrated the capability of TR-2PPE to This article not subject to U.S. Copyright. Published 2012 by the American Chemical Society
investigate surface and interface dynamics, including interfacial charge-transfer processes in organic and molecular systems.26−33 The sensitivity of TR-2PPE to the near-surface region and the ability to build the heterojunction layer-by-layer using organic molecular beam techniques provides a means of clearly separating interfacial charge-separation and relaxation processes from bulk behavior. The focus of this investigation is the C60/CuPc heterojunction, a prototypical donor−acceptor (D-A) pair studied extensively for small molecule solar cells.34,35 In the TR-2PPE measurements described here, CuPc S1 exciton populations are excited using subpicosecond pulses in the near-IR. A subsequent UV probe pulse stimulates photoemission from this population that is monitored to extract information on the dynamics of the excited state. The measurements are performed as the interface is built up from a single monolayer of CuPc on C60 to a bulk CuPc film, allowing us to compare the exciton dynamics at the D−A interface with those in the CuPc bulk. From these results, we determine rate constants for charge transfer, recombination, and relaxation at the interface as a function of energy. We begin by describing the experimental details involved in the formation of the D−A interfaces using organic molecular beam expitaxy (MBE), characterization of these films, and a description of the TR-2PPE measurement methods. This is Received: June 8, 2012 Revised: August 23, 2012 Published: August 23, 2012 19173
dx.doi.org/10.1021/jp305637r | J. Phys. Chem. C 2012, 116, 19173−19181
The Journal of Physical Chemistry C
Article
followed by a discussion of results for CuPc/C60 interfaces as a function of film thickness and analyses of the time-dependent data employing rate-equation modeling. The picture of the dominant charge-transfer, relaxation, and recombination processes at interface developed from the modeling is used to provide connections to semiclassical (Marcus) treatments of interfacial charge transfer (ICT).36−38 This report completes a previous brief description of this research project26 with additional experimental results and a much more extensive data analysis.
■
EXPERIMENTAL METHODS CuPc/C60 interfaces were formed by thermal deposition of CuPc onto 20 monolayer (ML) thick C60 films on Ag(111). A monolayer unit is defined here as the molecular density for complete surface coverage one molecule thick (C60 or CuPc). With this definition, 1 ML C60 corresponds to ∼6 × 1013 cm−2 for a well-ordered hexagonal layer on Ag(111). For CuPc deposited on this C60 surface, scanning tunneling microscopy (STM) has shown that the CuPc molecules are oriented nearly “upright” to the surface forming a bulk-like phase with a surface density of ∼4 × 1013 cm−2.39,40 The nominal thicknesses for each layer are 1 nm for C60 and 1.2 nm for the upright CuPc molecules. The initial C60 monolayer was deposited at 570 K to promote order, and additional layers were deposited at 425 K. CuPc was deposited in thicknesses ranging from 1 to 5 ML at a substrate temperature of 315 K. Typical deposition rates for both materials were in the range of 0.05 ML min−1. The work function of the initial C60 surface, based on the midpoint of the low-energy threshold in UPS or 2PPE, was 4.60 ± 0.05 eV and decreased to 4.35 ± 0.05 eV with the addition of CuPc. The coverages for C60 or CuPc were calibrated by following the work-function change and shifts in image state energies. The interface orientation is shown schematically in the inset of Figure 1. STM measurements39,40 indicated well-ordered 2D growth with no indication of significant second-layer formation before completion of the first layer, suggesting good layer-by-layer growth. The point of monolayer completion determined by work function change agreed very well with calibrations based on a quartz crystal thickness monitor, also suggesting first-layer formation without significant second-layer growth. STM measurements could not be performed for thicker CuPc films on C60. We expect growth of subsequent layers of the bulk-like CuPc to continue with some variation from the nominal layer thickness with increasing coverage, as observed for bulk-like pentacene film growth on weakly interacting substrates.41 Polarized ultraviolet photoelectron spectra were identical for all CuPc thicknesses beyond 1 ML, indicating no significant change in the CuPc orientation. Both one-photon He(I) ultraviolet photoelectron spectroscopy (UPS) and 2PPE data were collected with a hemispherical analyzer with typical resolutions of 100 and 60 meV, respectively. A He resonance lamp was used as the excitation source for UPS. For 2PPE measurements, we employed the fundamental of a tunable Ti:sapphire oscillator (Coherent Chameleon XR, 150 fs, ≤ 20nJ)42 and its third harmonic (INRAD 5-050). These measurements, along with inverse photoemission (IPES) data43 from the literature, yield the occupied and unoccupied interfacial electronic structure presented in Figure 1. Figure 1 also shows the excitation scheme employed for TR2PPE. The lowest energy optical absorption in the
Figure 1. Interfacial electronic structure determined by UV photoemission and 2PPE from this work and previous inverse photoemission42 and optical absorption data. The TR-2PPE excitation scheme is shown, along with important processes involving CuPc exciton dynamics at the interface. The work function, not shown in the Figure, was 4.35 eV ± 0.05 eV for CuPc layers on C60. This was determined based on the midpoint of the low-energy photoemission threshold. The inset shows the relative molecular orientations of CuPc and C60 at the interface determined by STM.39,40
phthalocyanine, the Q-band, arises from allowed π→π* HOMO−LUMO transitions, producing a band centered at ∼650 nm and extending to 800 nm.44 The Ti:sapphire fundamental at 750 nm (1.65 eV) lies within this absorption band and was used to pump CuPc S1 exciton levels. The third harmonic (4.96 eV) was used as probe. This wavelength combination was chosen as a compromise to match the pump to the CuPc Q-band absorption while suppressing the onephoton background signal from the UV probe to a manageable level. The pump beam was focused to ∼100 μm diameter, producing a peak power density of ∼0.3 GW cm−2. The probe power density was significantly lower. The pump and probe beams were noncollinearly overlapped at the sample with perpendicular polarizations (s-pump, p-probe) to eliminate contributions from coherent two-color photoemission and isolate dynamics associated with real intermediate states. Initial 2PPE investigations were performed to characterize excitations in thick C60 and CuPc films. These measurements revealed the absence of measurable response from pure C60 films using the pump−probe scheme described above, consistent with the lack of significant absorption in C60 at 750 nm.45 In contrast, pump-induced transitions to CuPc exciton levels were clearly observed in thick CuPc films. These studies also revealed that the photoemission intensity is dominated by the outermost CuPc layer because of electron attenuation by ∼1.2 nm “upright” layer of CuPc molecules. Analogous complete attenuation of C60 intensity by the first CuPc ML was also observed for UPS measurements. 19174
dx.doi.org/10.1021/jp305637r | J. Phys. Chem. C 2012, 116, 19173−19181
The Journal of Physical Chemistry C
Article
zero delay are similar, consisting of a broad peak centered close to 0.5 eV above EF. The location of this peak is consistent with transitions originating from the CuPc S0 HOMO level, shifted by the pump energy. These pump-induced excitations are identified as CuPc π −π* S1 excitons. Examination of the spectra as a function of pump−probe delay also reveals that for both CuPc thicknesses the spectral intensity is centered in one of two energy regions, depending on the delay. We illustrate this point by plotting spectra at small probe delays (