Cascade energy transfer versus charge separation in ladder-type oligo(p-phenylene)/ZnO hybrid structures for light-emitting applications

Hybrid inorganic/organic semiconductor structures (HIOS) are promising candidates for achieving improved or even new opto-electronic functionalities. Control of the electronic coupling at the heterointerface is a crucial aspect in this context. In HIOS comprised of a ZnO quantum well (QW) and an adjacent organic layer, efficient conversion of Wannier excitons into Frenkel excitons via Förster-type resonant energy transfer (FRET) has been demonstrated.^1,2 Similar observations were made for GaN, InGaN, or GaAS as donor material.^3–5 In this way, the difficulty of injecting high carrier densities directly into the organic material can be circumvented. Hybrid light emitters based on this concept will combine the exceedingly high electrical injection rates of inorganic semiconductors with the unsurpassed radiation output of molecular systems. For achieving truly superior operation in comparison to the single components, the inorganic/organic excitonic coupling must be very efficient and loss channels of the excitation energy at the hybrid interface rendered inactive. To meet the first challenge, we introduce ladder-type oligo(p-phenylenes) (LOPPs) as efficient FRET partners for ZnO.^6–8 Due to their rigid planar geometry, LOPPs exhibit weakly Stokes-shifted absorption and emission spectra featuring well-resolved vibronic progressions. Their absorption coefficients are much larger than those of the respective oligo(p-phenylenes) with non-rigidified backbone and the LOPPs' fluorescence yield reaches unity in solution.⁶ Moreover, the exciton transition of ladder-type quarterphenyl (L4P) and its spiro-derivatives is almost in resonance with that of ZnO. However, despite all these favorable features, an intrinsic loss channel of the excitation energy might be present due to an unfavorable energy level alignment between the frontier molecular levels and the band edges of the inorganic semiconductor at the hybrid interface.^9,10 The electron affinity of ZnO is about 4.2 eV, which is larger than that of most common conjugated organic molecules. Therefore, the formation of a type-II energy level alignment with ZnO accepting an electron from the excited organic component is expected^10–13,15 resulting in nonradiative deactivation.^9,10 Such configuration, detrimental when targeting light-emitting applications, holds for many other conventional inorganic semiconductors as well.^9,16,17 Indeed, it was shown in previous work that the photoluminescence (PL) of a spirobifluorenene derivative is substantially quenched when in contact with ZnO due to interfacial exciton dissociation.¹⁰ The same situation is encountered in polyfluorene/GaN HIOS¹⁴ hampering so far the exploitation of the full potential of this class of heterostructures in light-emitting applications. Here, we present an effective strategy based on a FRET cascade to overcome this loss channel and to achieve highly luminescent HIOS.

The HIOS studied in this work are composed of a ZnO/ZnMgO QW structure covered with a thin layer of LOPPs. The inorganic part is grown on a-plane sapphire substrates by radical source molecular beam epitaxy. Details on the ZnO epitaxy can be found elsewhere.¹⁸ The 3.5-nm thick ZnO QW is placed on top of a 500-nm thick lower Zn_0.9Mg_0.1O barrier layer. The upper Zn_0.9Mg_0.1O barrier separating QW and organic layer is kept sufficiently thin (2 nm) to allow for efficient FRET. The HIOS are completed with a top organic layer grown by molecular beam deposition. The growth rate (0.1 nm/min) and layer thickness are determined by a quartz crystal microbalance. In case of organic blends, the composition is adjusted via the deposition rates of the molecular components. The area of the organic layer is defined by a shadow mask leaving always an uncapped QW structure as reference. Growth of organic and inorganic component is performed in ultra-high vacuum while the samples are transferred from the inorganic to the organic growth chamber under high-vacuum conditions. FRET and charge separation are evaluated by a combination of PL, PL excitation (PLE), and time-resolved PL spectroscopy while the energy level alignment at the LOPP/ZnO interface is derived from ultraviolet photoelectron spectroscopy (UPS).

Before discussing the QW-LOPP FRET, we characterize the isolated components of our HIOS on reference samples. The chemical structures of two representatives of the LOPP family suited as FRET acceptor molecules for ZnO, ladder-type quarterphenyl (L4P) and a triply spiro-annulated derivative (L4P-Sp3), are depicted in the insets of Figs. 1(a) and 1(b), respectively. Absorption spectra of a thin L4P-Sp3 film and of L4P dispersed in a PMMA matrix are shown together with PL spectra of a ZnO QW recorded at low and room temperature in Fig. 1(c). Both molecules exhibit well-resolved vibronic features with the S_0,ν=0 → S_1,ν=0 transitions overlapping extensively with the room-temperature QW PL so that one essential condition for efficient excitonic FRET is ensured. The transition energies of the molecules can be fine-tuned through attachment of different substituents at the bridging carbon atoms. Spiro groups, as shown here, cause a red-shift of the absorption spectrum of some 10 meV. Furthermore, packing of the molecules in the solid state as well as the film-forming properties can be adjusted in this way. L4P readily crystallizes.⁷ However, for vacuum deposition on ZnO, Volmer-Weber growth prevails as evident from the atomic force microscopy (AFM) image in Fig. 1(a). The resultant films are composed of separated three-dimensional islands and are not suited for use in HIOS. On the contrary, spiro-annulated LOPPs grow in a two-dimensional mode, consistent with their reduced tendency to crystallize. In particular, L4P-Sp3 forms homogeneous films with a very smooth surface morphology [Fig. 1(b)]. In addition, the extended spiro groups prevent close packing of the molecules; and hence, the solid-state absorption spectrum is very similar to that of the isolated molecules in solution. In view of these advantageous features, L4P-Sp3 is selected in the present study.

We first evaluate FRET in a L4P-Sp3/ZnO/ZnMgO HIOS. The experiments described in the following are performed at cryogenic temperatures (T = 5 K) better suited to disentangle the relevant transfer processes. The PLE spectra reported in Fig. 1(d) are recorded by setting the detection energies to the L4P-Sp3 PL transition at E_det = 3.12 eV [Fig. 2(b)] or to the QW PL (E_det = 3.352 eV). The PLE spectrum of the QW displays the exciton ground- and excited-state absorption features as well as the ZnMgO band edge (E_ex > 3.55 eV), the latter indicating efficient capture of photo-excited carriers by the QW. All these features are also present in the molecular PLE demonstrating unambiguously exciton transfer from the ZnO QW to L4P-Sp3. PL transients are obtained by time-correlated single-photon counting using a mode-locked frequency-doubled fs-Ti:sapphire laser (5 W/cm², 76 MHz) as excitation source. The PL decay transient of the QW in absence of L4P-Sp3 is in good approximation single-exponentially with a time constant of τ_QW = 200 ps. In the hybrid part covered with L4P-Sp3, the QW decay time shortens to $τQWh=56 ps$ verifying opening of an additional decay pathway for the QW excitons (Fig. 2(a)). Assuming that no parasitic losses on the QW side are introduced, we find a characteristic inorganic-organic FRET time of $τETio=(1/τQWh−1/τQW)−1=77 ps$ and an efficiency of $ηETio=τQWh/τETio=0.72$⁠, i.e., about three out of four excitons photogenerated in the QW are transferred to the L4P-Sp3 layer. However, inspection of the PL spectrum (Fig. 2(b)) shows that there is barely light produced by the organic layer. The quantum yield of L4P-Sp3 is sufficiently high in solid state and cannot serve as explanation. Consequently, there has to be an efficient pathway for the loss of excitation energy at the ZnMgO/L4P-Sp3 interface.

Valuable insights into the above issue are provided by UPS measurements, from which the energy level scheme depicted in Fig. 2(c) is derived. The experiment is performed at PM4 beamline at Bessy II (Germany) consisting of an interconnected sample preparation analysis UHV apparatus (base pressures 10⁻⁸ mbar and 10⁻¹⁰ mbar). Photoelectrons excited with 35 eV UV light are detected by a Scienta SES 100 spectrometer with ca. 200 meV energy resolution. Films of L4P-Sp3 are evaporated onto bulk ZnO single crystals (Tokyo Denpa) cleaned be repeated sputter/annealing cycles (1 kV, 20 mA, 40 min/400 °C). The energy offset between the valence band maximum (VBM) of ZnO and the highest occupied molecular orbital (HOMO) level of L4P-Sp3 deduced from the binding energy onsets amounts to 1.2 eV. The positions of the conduction band minimum (CBM) of ZnO and the lowest unoccupied molecular orbital (LUMO) level of L4P-Sp3 are estimated by adding the exciton binding energies [60 meV for ZnO and 450 meV for L4P-Sp3 (Ref. 19)] to the optical band gaps (3.3 eV for ZnO and 3.25 eV for L4P-Sp3) determined from the respective absorption spectra. The L4P-Sp3 LUMO is hence situated ≈1.5 eV above the ZnO CBM. As speculated above, a type-II interface is formed where the excitons undergo charge separation, with the electron in ZnO and the hole on the molecular side. In the present HIOS, the organic layer is in contact with the upper Zn_0.9Mg_0.1O barrier with a band gap widened by only 0.3 eV relative to ZnO.¹⁸ Therefore, the excitons initially collected in the QW and passed by FRET into L4P-Sp3 dissociate here as the energy gained by the electron transfer to the Zn_0.9Mg_0.1O barrier still substantially exceeds their binding energy. The time-resolved molecular PL fully confirms the charge separation process [Fig. 2(d)]. The decay time of L4P-Sp3 in a reference layer is about τ_M = 500 ps, whereas a shortening to $τMh≈95 ps$ is found for the HIOS. The exciton quenching proceeds in two steps, namely, exciton diffusion towards the hybrid interface and subsequent dissociation of electron and hole. The 95-ps overall decay can thus be taken as an upper limit for the characteristic time of the charge separation process. As a consequence, a large fraction of the excitation energy is not converted into light. To be of use in light-emitting applications, exciton dissociation at the hybrid interface must therefore be prevented.

Tuning the energy levels into a type-I alignment, either by lowering the work function of ZnO via surface functionalization or by increasing the electron affinity of LOPPs by chemical modification, is difficult as the energy offsets are very large.^20,21 Note that also the use of molecules with substantially smaller HOMO-LUMO gaps will not circumvent the problem of competing charge separation. We performed experiments with the red-absorbing polymer poly((9,9-dioctylfluorene)-2,7-diyl-alt-[4,7-bis(3-hexylthien-5-yl)-2,1,3-benzothiadiazole]-2′,2″-diyl), which is even used as an acceptor in organic photovoltaics²² and observed a substantial decrease of the PL lifetime from 1.1 ns to 315 ps when deposited on ZnO. We propose therefore a different strategy based on a fast and highly efficient FRET cascade: Following the primary FRET step starting at the ZnO QW, the excitation energy is funneled away from the hybrid interface by a secondary energy transfer step within the organic HIOS part before interfacial exciton dissociation can take place. As secondary acceptor and FRET partner for L4P-Sp3, a longer chain LOPP, namely, ladder-type sexiphenyl (L6P) is selected [inset of Fig. 3(a)].⁶ The more extended π-electron system of L6P shifts its S₀ → S₁ absorption in resonance with the L4P-Sp3 emission, as demonstrated in Fig. 3(a). Furthermore, there is also spectral overlap between the S_0,ν=0 → S_1,ν=2 transition and ZnO so that direct energy transfer from the QW to L6P is in principle also possible. The FRET process within the organic layer is elucidated in blends obtained by co-evaporation of L4P-Sp3 and L6P on an inert substrate (Al₂O₃). A L6P fraction of 10% is low enough for avoiding phase separation, but sufficient for adequate FRET rates. The PL spectrum of the blend is indeed dominated by the L6P emission, while the signal from L4P-Sp3 is almost entirely quenched due to the presence of the acceptor molecules [inset Fig. 3(b)]. PLE data corroborate highly efficient FRET within the organic blend: The spectrum recorded at the vibronic S_1,ν=0 → S_0,ν=1 PL feature of L6P reproduces exactly the absorption spectrum of L4P-Sp3 (Fig. 3(b)). The PL decay of L4P-Sp3 in the blend can hardly be resolved by our photon counting setup providing an upper estimate of the time constant of the organic-organic FRET $τEToo<30 ps$ at the present L4P-Sp3/L6P mixing ratio and a lower estimate of the efficiency $ηEToo>0.9$⁠. The decay time observed for L6P (curve not shown) is τ_L6P ≈ 440 ps.

Competition between the FRET cascade and interfacial exciton dissociation is studied on a HIOS composed of a ZnO/ZnMgO QW covered by a layer of a L4P-Sp3/L6P blend. To avoid interaction between ZnO and L6P, a 0.5-nm thick interlayer of pure L4P-Sp3 is introduced. The thickness of the blended layer is 3.5 nm so that the total width of the organic part is the same as in the “non-cascade” HIOS of Figs. 1(d) and 2. Indeed, in the “cascade” HIOS, prominent molecular emission of L6P arises [Fig. 4(a)]. The energy of the pump photons is chosen such that direct excitation of L6P is negligible. Details on the cascade process are attained from time-resolved PL measurements. First, the QW PL transients in the hybrid and reference part of the “cascade” HIOS depicted in Fig. 4(b) provide practically the same transfer time $τETio=65 ps$ and efficiency $ηETio=0.77$ as for the non-cascaded version in the sole L4P-Sp3 HIOS. That is, the first step of the cascade is not modified. Second, the donor role of L4P-Sp3 in the next cascade step is verified by a further shortening of its decay time, now clearly below the 30-ps time-resolution. Third, in marked contrast to L4P-Sp3 in the single-stepped FRET, the PL decay time of L6P in the HIOS $(τMh≈400 ps)$ is nearly identical to that of the blend on an inert substrate. All these observations imply that the excitation energy is effectively funneled away from the interface. The recovery of the molecular emission (R_EM) is measured by the ratio of the (spectrally integrated) yield of L6P in the “cascade” HIOS versus that of L4P-Sp3 in the “non-cascade” sample. As the inorganic-organic FRET is almost identical, the experimental ratio R_EM ≈ 10 can be inferred by scaling the yield in both samples relative to that of the QW emission. Theoretically, denoting by σ_L6P and σ_L4P-Sp3, the radiative efficiency of the molecules in the respective HIOS, it holds $REM=ηETooσL6P/σL4P–Sp3$⁠. The radiative efficiency is $σM=τMh/τMrad$⁠, where the total molecular lifetime $τMh$ is directly measured [95 ps (L4P-Sp3), 400 ps (L6P)], while the ratio of the radiative lifetimes can be quantified by comparing the molecular extinction coefficients providing $τL4P–Sp3rad/τL6Prad≈2$ and, finally, $REM≈8 ηEToo$⁠. This confirms the above estimate of $ηEToo≈1$ and demonstrates that FRET from L4P-Sp3 to L6P outpaces almost entirely the quenching of L4P-Sp3 excitons at the hybrid interface.

Cascade FRET persists up to room temperature as testified by the same set of measurements used above. The primary transfer from the QW to the L4P-Sp3 stays reasonably efficient $(ηETio=0.32)$ and the L6P emission is bright (Fig. 4(c)). These findings are remarkable as, generally, the quantum yield of inorganic QW structures declines at higher temperatures. Nonradiative recombination processes are thermally activated, while the radiative lifetime of the excitons becomes intrinsically longer.^23,24 The PL yield of the present ZnO/ZnMgO QWs drops roughly by a factor of 10 from T = 5 to 300 K. The FRET contribution from the QW excitons to the total L6P emission is measured under the present excitation conditions by the ratio $ηETioAQW/[ηETioAQW+AL4P–Sp3]=0.3$⁠, where A_L4P-Sp3 = 0.07 and A_QW = 0.1 are the fractions of photons absorbed in the L4P-Sp3 layer and the QW, respectively. Therefore, a substantial part of QW excitons otherwise lost by nonradiative recombination is recovered by FRET for light emission.

In conclusion, LOPPs are well suited FRET partners for ZnO. Charge separation caused by an unfavorable interfacial energy level alignment can be surpassed by cascade FRET where the second step takes place within a blend of suitably adjusted organic components. The resulting cascade does not noticeably reduce the total FRET efficiency. At room temperature, about one third of the excitons primarily excited in the QW are passed to the LOPP and emit light, whereas maximally one tenth would radiatively recombine in the sole QW. Hybrid FRET can thus effectively bypass non-radiative recombination. FRET-based hybrid light emitters are thus promising in terms of power consumption. Cascade FRET to suited acceptor molecules can furthermore be used for efficient color conversion or white light generation.²⁵ The HIOS presented in this work are not yet optimized and further improvements can be achieved by minimizing the FRET distance or optimizing the molecular packing.

Financial support of the Deutsche Forschungsgemeinschaft (SFB 951) is acknowledged.