1. Introduction

Opto-electronics based on synthetic compounds such as organic light emitting diodes (OLEDs) or organic photo-sensors are of high technological interest due to their excellent luminescence behavior and pronounced photo-sensitivity in combination with low-cost and easy preparation routines. However, those stacked devices composed of several planar organic layers and metal electrodes have some disadvantages, which reduce the in- or out-coupling of light and thereby, their overall efficiency. For instance, in OLEDs the emitted photons can effectively couple to surface plasmon polaritons (SPPs) at the organic-metal interfaces or to wave-guide modes which can add up to total losses of about 50% [1]. A suitable approach to increase the out-coupling efficiency in OLEDs is rendered possible by patterning of the active device area [2]. In this contribution, we establish nanostructuring by implementation of a periodically ordered silver nanoprism array beneath the organic emitter, which has the additional benefit of enhancing the emission efficiency by coupling to its localized surface plasmon resonances (LSPRs) [3].

Aiming for a preparation route of technological relevance, we fabricated these plasmonic nanoarrays by means of Shadow Nanosphere Lithography (SNSL), utilizing the advantages of deterministic positioning, scalability and reproducible manufacturing [4]. Moreover, SNSL provides possibilities to create long-range ordered arrays of geometrically varying nanoparticles, ranging from triangles to half-spheres [5–7], to stack different materials, which is a primary requirement for the preparation of metal-organic NCs employed in this study, or to generate even more complex structures of nanometer sizes. Therefore, this general approach allows for a combination of effects, namely the nanopatterning of the active device region and the LSPR of metallic nanoparticles, positively affecting the light out-coupling efficiency. Moreover, regarding the proposed sample architecture, the metal particles located on top and at the bottom of the emissive organic volume are suitable to operate as electrodes in future opto-electronic devices.

Here, we investigate the change in photoluminescence intensity caused by plasmonic coupling of two-dimensional silver nanoarrays to the excited states of the archetypical organic dye zinc phthalocyanine (ZnPc), the latter showing an exceptionally high photo-stability and broad chemical variety along with specific emission characteristics, including structure dependent optical transitions in the visible [8]. Moreover, we take advantage of controlled co-evaporating this organic emitter at low concentrations into a matrix of the well-established OLED compound tris(8-hydroxyquinolinato)aluminum (Alq₃) [9] and of preparing nanocavities (NCs) with defined linear dimensions for better out-coupling characteristics compared to closed thin films.

By these preparative approaches and selected metal-organic architectures we are able to demonstrate a significant enhancement in photoluminescence emission reaching gain factors of up to 700 with respect to the corresponding thin film counterparts, when normalized to structure size.

2. Experimental

Alq₃ and ZnPc were purchased from Sigma Aldrich and purified twice by gradient sublimation. For the preparation of the monolayer shadow mask we used polystyrene nanospheres of 500 nm diameter purchased from Polysciences Europe GmbH. The diluted spheres were centrifuged to replace the water by ethanol and thereby, to reduce the surface tension for subsequent spin coating. Afterwards, the solution is spin-coated onto a cleaned thin glass substrate for 60 s at 4000 rpm. Onto the resulting nanosphere monolayer 30 nm thick silver was thermally evaporated in a vacuum chamber at a base pressure of 10⁻⁶ mbar. For thin film samples the PS mask is removed prior to evaporating the following layers. In case of the nanocavities (NCs) organic layers were sublimed by molecular beam deposition directly on top of the nanosphere shadow mask in a vacuum system at a base pressure of 10⁻⁷ mbar. Alq₃ was deposited at a rate of about 8 nm/min and ZnPc was evaporated at 2 nm/min for neat films and at about 0.3 nm/min for co-evaporated films. It is important to ensure the same angle of incident, ϴ, (see Fig. 1(a)) for the subsequent evaporation of the 60 nm thick silver capping as otherwise, partial overlapping of the various NC layers will lead to significant reduction of the NCs’ optical quality. Finally, samples were encapsulated in a glove box under N₂ atmosphere by glass slides and a two-component glue to prevent photo-degradation of the active organic compound in the subsequent optical studies.

 

Fig. 1 (a) Scheme of sample preparation and thickness dependent shadowing during deposition leading to the formation of tilted sidewalls. (b) Investigated sample geometries: (I) thin film (TF) on glass and (II) on silver nanoprism array (TF on Ag prisms), (III) Ag/organic/Ag nanocavities (NC) and (IV) neat organic pillars (NP). (c) Molecular structure of ZnPc (left) and Alq₃ (right). AFM images of (d) silver nanoprism array, (e) ZnPc layer on silver nanoprism array (II), (f) ZnPc nanocavities (III) and (g) neat ZnPc pillars (IV).

Download Full Size | PDF

Photoluminescence measurements were performed with a home-built confocal microscopy setup at 638 nm linearly polarized cw laser excitation. Samples are moved in three dimensions with nanometer resolution by a piezo stage (nPoint). A high aperture (NA = 1.49) oil immersion objective (100XOTIRF, Olympus) fulfills two functions, focusing of the laser spot onto the sample surface at normal incidence and collecting the emitted light. The latter is guided through a pinhole of 35 µm diameter and a 650 nm long pass filter. Photoluminescence images are recorded by an avalanche photodiode (Count-100C-FC, Laser-components GmbH) and the corresponding spectra by a PIXIS 400B CCD camera in combination with an Acton SP2300 spectrograph (both Princeton Instruments). Time-correlated-single-photon-counting (TCSPC) experiments were performed with the confocal setup described above employing a pulsed SuperK extreme supercontinuum laser (NKT Photonics) operated at 635 nm wavelength in combination with a PicoHarp 300 TCSPC module (PicoQuant).

Finite-difference-time-domain (FDTD) simulations were carried out with a commercial software package (FDTD Solutions, Lumerical). We simulated scattering and absorption cross sections as well as the electrical field distribution of the hexagonal arrays of 500nm lattice constant and composed of 30nm thick, round-cornered silver nanoprisms supported by glass substrates (n = 1.45). According to the experimental conditions described below, the metal nanoprisms exhibit tilted side facets of 160 nm bottom length and 90 nm top length. We would like to stress that consideration of this tilted sidewall geometry is essential as it leads to spectral modulations of the plasmonic resonances due to additional couplings [10], as observed also later in our PL studies. For the calculations, we assumed the nanoprism arrays to be covered by either a 30 nm thick organic layer or by an organic-metal combination comprising a 30 nm thick organic prism below a 60 nm thick silver nanoprisms. The corresponding refractive index of the organic semiconductor matches either ZnPc (n = 1.9) or Alq₃ (n = 1.7). To calculate the scattering and absorption cross section we used linear polarized plane waves.

3. Confocal microscopy PL spectra analysis

In order to determine the specific influences of co-evaporation and nanocavity-engineering four different sets of sample geometries were analyzed. At first, we will present a comparative discussion of the optical properties of neat ZnPc reference thin films (TF) and of ZnPc layers on top of nanosized Ag arrays (Figs. 1(b,I) and (b,II), respectively), followed by a spectral characterization of SNSL structured Ag/ZnPc/Ag NCs (Fig. 1(b,III)). To judge the impact of exciton-exciton annihilation and quenching at metal-organic interfaces, co-evaporated thin films of Alq₃:ZnPc on Ag arrays (Fig. 1(b,II)) and of Ag/Alq₃:ZnPc/Ag NCs (Fig. 1(b,III)) are studied in regard to the corresponding ZnPc sample configurations. Finally, to distinguish between plasmonic and geometrical effects on the emission spectra and the PL enhancement we compare the previous results with optical characteristics of neat ZnPc and Alq₃:ZnPc nanopillars (NPs) (Fig. 1(b,IV)).

Before presenting the results and their interpretation, it is important to note the morphological impact of the specific arrangement in SNSL as illustrated in Fig. 1(a). During deposition, the layer thickness atop the polystyrene (PS) beads, used as shadow mask in this lithographic process, increases as well. As a consequence, at fixed evaporation angle, ϴ > 0 deg., the projected gap between the PS spheres is reduced continuously leading to tilted sidewalls of the nanostructures [10]. This specific geometrical feature of the prisms is considered in our FDTD simulations described below and, by the additional couplings, results in the observed spectral modulations of the plasmonic resonances.

3.1 ZnPc thin film (TF)

Starting with the sample series based on ZnPc as active organic medium, we will first highlight the morphological and geometrical aspects related to the SNSL method applied. For this purpose, Figs. 1(d)-1(g) show representative Atomic Force Microscopy (AFM) images recorded after each step of the metal-organic NC fabrication. As clearly seen by Fig. 1(d), already the initial prisms of the array differ slightly in size and tip radius, caused by the non-perfect nanosphere monolayer and the related variation in shadow mask apertures. This will lead to slight variations in the energetic position of the plasmonic resonances of the individual silver prisms [11] and thus, to spatially different enhancement characteristics across the array of a given sample as confirmed by our data below. The silver nanoprisms exhibit edge lengths of 150 ± 20 nm and a thickness of about 30 nm. The lattice constant equals the diameter of the PS spheres of 500 nm. Figure 1(e) illustrates the array of sub Fig. 1(d) after vacuum deposition of a 30 nm thick ZnPc thin film on top. Obviously, the ZnPc layer mimics the underlying silver nanopattern, which is considered favorable for reduced SPP losses in an actual device, but in addition, forms small crystallites between the prisms as one would expect for ZnPc thin film growth on an unstructured support [8].

Figure 2(a) shows a photoluminescence intensity map taken by confocal microscopy of a bare ZnPc layer deposited on an Ag prism array (638 nm excitation wavelength, NA = 1.49). The same hexagonal symmetry and lattice constant of both the patterned silver nanostructures and the lateral PL intensity modulation are evident. Already for this Ag/ZnPc bilayer geometry it becomes apparent, that the plasmonic excitations of the metallic nanostructures lead to an enhancement of the ZnPc photoluminescence on top and in between the prisms. According to the aforementioned variation in size and tip profile, the brightness measured on the single nanoprisms differs somewhat. Complementary FDTD simulations of the electrical field distribution in ZnPc regions with and without silver nanostructure underneath result in the E-field enhancement displayed in Fig. 2(b), for an excitation by x-polarized light at 785 nm. These theoretical calculations resemble the intensity increase experimentally observed on top and at the tips of the silver nanoprisms.

 

Fig. 2 (a) Confocal photoluminescence image of a ZnPc layer grown atop a silver nanoprism array as in (1e). (b) FDTD simulations of the in-plane electrical field enhancement inside a nanocavity comprising a material of refractive index n = 1.9, according to ZnPc, and excited by light linearly polarized along the x-axis.

Download Full Size | PDF

To elucidate the origin of this gain in emission in more detail we performed spectrally resolved PL measurements on this set of samples. As illustrated in the upper graph of Fig. 3(a), the ZnPc emission centered at around 1.3 eV (930 nm) and attributed to an excited dimer (excimer) state [12] is strongly enhanced for the nanostructured sample (solid line) versus that of the neat molecular layer (dashed line). The corresponding intensity ratio yields the wavelength dependent enhancement factor shown in the upper graph of Fig. 3(b) (blue line). Additionally, the FDTD simulated scattering cross section C_scatt is shown and depicts the energetic position of the contributing localized surface plasmon (black curve). For better comparability, we have scaled this quantity to the basal area of the nanoprisms leading to an area normalized scattering cross section per prism, area norm. C_scatt. The scattering cross section is proportional to the electrical polarizability α which, according to the Clausius-Mossotti relation, is highly dependent on the dielectric environment. Due to the geometrical complexity of our nanostructures, C_scatt has to be determined numerically, assuming ε₂ to be constant.

 

Fig. 3 (a), (c) Photoluminescence spectra and (b), (d) resulting enhancement factors (colored lines) as well as simulated localized surface plasmon resonances (black lines) of ZnPc and Alq₃:ZnPc based samples. The respective upper graphs show the PL of neat TFs (dashed lines) and TFs on Ag prisms (solid lines); lower graphs show the PL of NCs (solid lines) and TFs (dashed lines). (b), (d) Overlap of both curves (PL enhancement factor and simulated LSPR) indicates major plasmonic contributions to the PL enhancement.

Download Full Size | PDF

Compared to free silver nanoprisms, the LSPR of our nanostructures shows a pronounced red-shift of about 300 nm due to the refractive index of the glass substrate (n = 1.45) and the local environment, which is assumed to be n = 1.9 [13] in the spectral emission region of the ZnPc cover layer (800-1000 nm). The modulation of the calculated scattering cross section at around 1.4 eV can be attributed to the tilted sidewalls of the nanoprisms accounted for in our modellings and caused by SNSL fabrication as illustrated in Fig. 1(a) [10]. Under these circumstances, the spectral overlap between LSPR and ZnPc emission leads to a threefold amplification of the PL intensity in presence of the nanostructured silver support.

A second peak of the spectral enhancement factor in the upper graph of Fig. 3(b) is located at about 1.8 eV. Based on spectroscopic studies on ZnPc in solution and thin films [12] as well as on the optical properties of ZnPc-blended Alq₃ layers discussed below, the second peak can be assigned to the emission of an excited monomeric ZnPc state. We suggest this monomeric state to originate from the amorphous fraction of the ZnPc volume on top of the silver nanoprisms, whereas the excimer emission arises from the complementary (short-range) ordered phase of the volume.

3.2 ZnPc nanocavity (NC)

The confocal microscopy image in Fig. 2(a) clearly reveals that the largest PL intensity enhancement occurs on top and at the tips of the nanoprisms. To take advantage of this phenomenon and to improve the PL out-coupling even further we prepared nanocavities consisting of a bottom and a top silver prism separated by a 30 nm thick ZnPc spacer as schematically depicted in Fig. 1(b,III).

The spectra of such a NC and a neat ZnPc thin film are shown in the lower graph of Fig. 3(a). It becomes obvious that the shape of the spectral emission is quite different in both cases. Whereas the PL of the planar ZnPc layer is dominated by the excimer luminescence, the spectrum of the Ag/ZnPc/Ag NC is governed by emission peaks between 1.5 eV and 1.85 eV. Concordantly, the AFM images in Figs. 1(e) and 1(f) show the formation of small crystallites embedded in the ZnPc thin film but not in the ZnPc NCs. We assume the distortion of the film morphology and thus the suppression of an (long-range) ordered phase to be induced by the much stronger substrate interaction atop the silver nanostructures compared to that at the ZnPc/glass interface, eventually leading to monomeric emission behavior. For instance, X-ray analyses revealed that on oxide surfaces, the weak substrate interaction supports an almost upright-standing, co-facial alignment of metal-phthalocyanines [8,14] whereas on metallic surfaces the strong interaction, often being of chemisorptive nature, forces a flat-lying configuration of the molecules [15]. To release the substrate-induced constraint and to promote the transition into the thermodynamically favored standing-up orientation at larger thicknesses, the organic layer forms an amorphous phase up to a certain critical thickness. The occurrence of monomer emission in larger aggregates was also confirmed for e.g. ZnPc nanoparticles in various solvents [16], perylene nanoparticles [17] and PTCDI nanostructures [18]. We substantiated this hypothesis by excitation power dependent PL measurements on ZnPc nanopillars (not shown). At sufficiently high laser powers the luminescence of the monomeric fraction starts to decrease, whereas the excimer contribution is still increasing. As higher intensities lead to higher thermal loads, we expect a thermally induced recrystallization within the nanosized ZnPc volume. It has to be stressed that all PL spectra used to determine the enhancement factors shown in Figs. 3 and 4 were recorded at sufficiently low excitation powers to avoid such irreversible morphological changes.

 

Fig. 4 (a) PL spectra of ZnPc-based (upper graph) and Alq₃:ZnPc-based (lower graph) NCs (solid lines), thin films (dashed lines) and NPs (dashed-dotted lines) . (b) Enhancement factors of ZnPc-based (upper graph) and Alq₃:ZnPc-based (lower graph) sample structures. Intensity gain of nanocavities relative to the respective thin film, representing the overall enhancement (black lines), and relative to pillar structures, representing the plasmonic contribution to the enhancement (red lines, in case of ZnPc multiplied by a factor of 10 for better visualization). The respective simulated LSPR are presented by the green dashed lines. Enhancement factors of the nanopillars relative to thin film samples, representing the structural contribution to the PL gain (blue lines).

Download Full Size | PDF

The strongest enhancement appears between 1.65 and 1.9 eV due the emerging monomeric peaks for the NC structures as shown in the lower graph of Fig. 3(b) and the absence of intensity for the planar ZnPc layer in this energy range. Also the geometrical design of the nanocavities allows for a more directional out-coupling of the emitted photons in comparison to thin film samples, which contributes to a higher enhancement factor. The mismatch between the positions of our simulated LSPR and the main PL increase by more than 100meV suggests the existence of such a mechanism. The enhancement profile indicates two peaks at about 1.7 eV and 1.85 eV, which can be unraveled by single peak fitting into three transitions located at 1.85 eV, 1.77 eV and 1.69 eV and correlated with the monomer emission as described above. The three narrow peaks between 1.76 eV and 1.79 eV are caused by Raman scattering [19], which is known to be enhanced in proximity to metal nanostructures [20].

In contrast, the PL in the energy regime below 1.7 eV is less increased. As indicated by our simulations, the LSPR of the Ag/ZnPc/Ag NCs is shifted towards higher energies in respect to that of an array fully covered by ZnPc due to the reduced effective refractive index in the vicinity of the silver nanoprisms. Additionally, due to coupling between the upper and lower nanoprism the shape of the resonance changed from thin film to NC structures. Therefore, the spectral mismatch between localized plasmonic resonance of the silver nanoprisms and ZnPc fluorescence inhibits a further enhancement and, in addition, the photoluminescence is effectively diminished by exciton-exciton-annihilation, reflection and absorption losses as well as by non-radiative quenching processes at the metal-organic interfaces. Finally, the absolute number of molecules within the illuminated NC volume is much smaller than for the corresponding thin film, resulting in a reduced portion of molecules adopting an excimer configuration as well.

3.3 Alq₃:ZnPc thin film (TF)

Within a bare ZnPc layer excitons are able to interact with each other leading to exciton-exciton annihilation [21]. As a second loss channel, excitons can diffuse towards metal/organic interfaces, where they can undergo an efficient quenching process [22]. Self-absorption by Förster-resonance energy transfer might constitute a third non-radiative loss process and is anticipated for the ZnPc compound under study. To avoid these non-radiative dissipation processes, dilution of fluorophores within an optically inactive host provides a suited strategy. In the present case, we doped ZnPc molecules into an Alq₃ matrix at a concentration of about 4% thus preventing interaction between excitonic species as well as their diffusive transport towards the silver boundaries. Beneficially, both HOMO and LUMO level of ZnPc are located inbetween those of Alq₃ [23,24] excluding quenching of optically excited ZnPc states by the matrix. Tuning our cw laser excitation wavelength to 638 nm the Alq₃ matrix becomes transparent and absorption occurs mainly by the phthalocyanine guest molecules. With respect to morphology, Alq₃ bears the further advantage of growing amorphous [25], thus promoting the formation of sharp interfaces with the subsequently deposited metal top-layer.

The upper graph in Fig. 3(c) shows the PL spectra of Alq₃:ZnPc thin films with (solid green line) and without (dashed green line) silver array underneath. Correspondingly, the spectrally resolved photoluminescence enhancement is illustrated in the upper graph in Fig. 3(d). The increase of PL intensity at 1.85 eV is assigned to an emission, which only occurs in nanostructured neat and co-evaporated samples in addition to the peak at 1.8 eV. We assume this blue-shift to be indicative for surface excitons, the latter experiencing a slightly lower relative permittivity by the reduced polarization of their local environment leading to an uplift in energy. However, an amplification of PL intensity by more than one order of magnitude appears in the range between 1.25 and 1.5 eV. The specific scattering efficiency simulated for a silver nanoprism array covered by a material of refractive index n = 1.7, such as for amorphous Alq₃ [26], is displayed, too. The comparison of both curves clearly demonstrates the coincidence of the localized surface plasmon resonance with the highest enhancement factor and therefore, the plasmonic origin of the latter. Furthermore, the slightly blue-shifted LSPR position with respect to ZnPc-covered Ag nanoarrays in Fig. 3(b) is a consequence of the lower refractive index of Alq₃ with respect to ZnPc (n_ZnPc = 1.9) [27].

3.4 Alq₃:ZnPc nanocavity (NC)

In a last configuration we exploited both the advantage of co-evaporated layers and of nanostructured design by means of Ag/Alq₃:ZnPc/Ag NCs. The lower graph in Fig. 3(d) shows the related spectral enhancement factor as well as the simulated LSPR. With respect to Ag/ZnPc/Ag NCs the main enhancement occurs at lower energies (1.6 to 1.7 eV) and the LSPR shows a very small blue-shift of 10 meV, which again is caused by the slightly smaller refractive index of Alq₃ compared to ZnPc [13,26]. But here this effect is less pronounced compared to the thin film references as the silver prisms are not completely covered by the organic layer and hence experience an overall lower effective refractive index. As before, the peak at 1.85 eV occurs as a specific feature of the nanostructured sample design, but in addition, overlaps with the enhanced reflectivity of the laser at the silver/glass interface, the latter convoluted by the applied long-pass filter with a cut-off wavelength of 650 nm. Although the total effect is very small it appears to be significant in the drawing of the comparative PL enhancement. Moreover, between 1.6 and 1.7 eV a more than 40-fold gain in PL intensity occurs, which in fact is even higher taking into account the smaller volume fraction illuminated in case of NCs versus closed Alq₃:ZnPc thin films. Adjusting this geometrical contribution and dividing the laser spot size by the area of a single NC, a correction factor of about 16 has to be taken into account, which finally yields a high geometrically adjusted enhancement of about 700 demonstrating the power of this combinatorial approach.

3.5 Distinction between plasmonic and geometrical impacts

The previous sample structures are not sufficient to unambiguously distinguish between the plasmonic and structural contribution to the observed PL increase in the metal-organic hybrid structures. To tackle this issue, we prepared nanopillars (NPs) of the neat organic compounds by Shadow Nanosphere Lithography. These structures allow us to distinguish the LSPR contribution by the metal front ends from confinement effects induced by total internal reflection at the sidewalls of the NPs, the latter promoting a directional out-coupling of emission intensity along the surface normal. The spectral emission of neat ZnPc NPs occurs at higher energies in respect to the corresponding thin film as illustrated by the upper graph in Fig. 4(a), supporting the previous considerations on monomer emission in nanostructured samples. Moreover, it can be seen that the respective peaks of NP and NC emission appear at the same energetic positions, however unequally distributed in intensity. Figure 4(b) illustrates the enhancement factors of NCs in relation to the corresponding NP structures and thin film. By the upper graph of this diagram it becomes obvious that the enhancement factor is mainly governed by the different structural properties of the three geometries as already the bare ZnPc nanopillars exhibit huge gains in PL intensity compared to the corresponding thin film. The comparison of ZnPc-based NPs and NCs reveals an enhancement profile, which spectrally coincides with the simulated LSPR yielding a plasmonic induced enhancement of up to one order of magnitude. Since the overall PL increase relative to that of the neat molecular layers is more than two orders of magnitude larger, this corroborates the strong impact of the applied nanopattering on the ZnPc NC emission gain. The absence of monomer emission in thin film samples prevent a reliable estimation of the respective enhancement factors, which we have carried-out for the co-evaporated sample series in the following.

Unlike neat ZnPc nanostructures, co-evaporated Alq₃:ZnPc samples exhibit a somewhat different behavior. Co-evaporated Alq₃:ZnPc samples with a phthalocyanine content of about 10% are presented in the lower graphs of Figs. 4(a) and 4(b). The higher ZnPc concentration was chosen to highlight the additional effect of molecular aggregation which is indicated by the reduced PL enhancement by a factor of four compared to the Alq₃ samples with 4% ZnPc content in Figs. 3(c) and 3(d). Additionally, this effect is confirmed by the spectrally resolved PL of the corresponding thin film references, where the excimer emission is increased with increasing ZnPc ratio compared to the monomeric emission. In contrast to thin films, the PL of NPs as well as of NCs reveals an additional peak at 1.85 eV like for patterned samples of neat ZnPc, confirming its emergence by nanostructuring. The luminescence enhancement of Ag/Alq₃:ZnPc/Ag NCs with respect to both, thin films as well as neat NPs is in good agreement with our simulated spectral distribution of the localized plasmon mode proving its dominant plasmonic origin. Therefore, alongside the structural influence, the main part of the observed overall enhancement in co-evaporated Alq₃:ZnPc NCs evolves from the coupling of the localized surface plasmon at the metal ends to the excited states of the ZnPc dyes.

4. Time-correlated-single-photon-counting (TCSPC) measurements

Due to the cw laser excitation, the previous analyses solely rely on the steady-state behavior of the metal-organic hybrid structures under study. But as the coupling to the localized surface plasmons directly affects the dynamics of the dye’s photo-excited states and, in particular, their radiative lifetime, we evaluated this property by means of time correlated single photon counting (TCSPC) measurements. The upper graph in Fig. 5 shows the time dependent PL intensity decay in neat and co-evaporated thin films. Multi-exponential fitting yields to time constants of 0.49 ns and 19 ns for ZnPc as well as of 0.36 ns and 2.52 ns for Alq₃:ZnPc. The lifetime component of 19 ns for ZnPc represents the decay at extended timescales, whereas the shortest lifetime of 0.36 ns is limited by the resolution of our setup as confirmed by deconvoluting the instrument response function (IRF) via a gaussian and an exponential function as shown in the upper graph of Fig. 5 (black curve). Therefore, we can only qualitatively identify the existence of an additional relaxation channel on shorter timescales, which becomes more prominent in the co-evaporated samples. Since in the neat ZnPc thin film predominantly excimers are present rather than monomers, we conclude the 0.49 ns decay component to correspond to the excimer lifetime in our system, whereas the 2.52 ns time constant of the co-evaporated Alq₃:ZnPc film is assigned to the monomeric decay. The existence of the short life time component (0.36 ns) exclusively in co-evaporated films is presumably due to the intermolecular ZnPc arrangement. Settels et al. [28] showed that different orientations between adjacent molecules alter the potential energy landscape of the excited states in such a way that efficient relaxation of bright into dark states via conical intersection might occur within hundreds of femtoseconds. Vice versa, the polycrystallinity of the neat ZnPc samples might hinder the ultrafast relaxation into dark states in a similar manner.

 

Fig. 5 TCSPC measurements. Upper graph: TCSPC signal of 30 nm thick films of ZnPc (blue) and Alq₃:ZnPc (green) fitted by a multi-exponential decay (red) together with the instrument response function (IRF) (black). Lower graph: TCSPC signal of 30nm thick Alq₃:ZnPc nanocavities (green) and nanopillars (black) modelled by multi-exponential fits (red).

Download Full Size | PDF

The lower graph in Fig. 5 displays the exciton decay in co-evaporated Alq₃:ZnPc NCs and NPs. In NPs and NCs the decay constants are determined to 0.99 ns and 2.49 ns as well as to 0.86 ns and 2.43 ns, respectively. The NCs show an additional fast decay component of 0.36 ns, which indicates the plasmonic coupling of the nanoprisms to the molecular emitters reducing their excited state lifetime by at least one order of magnitude. Comparing the decay time of 0.36 ns for the coupled nanoprism arrangement (NC) with that of 2.49 ns for the uncoupled geometry (NP) allows us to determine a lower limit for the Purcell factor of F_p » 10. Other groups report on Purcell factors between 200 and 900 for similar systems [29,30]. Though representing also lower estimates due to the limited time resolution, the higher F_p values cited in these studies can be attributed to the materials used, showing much longer decay times and thus higher effects on the emission rate, as well as to enhanced absorption losses in our organic-metal hybrid structures caused by metal penetration upon thermally depositing the silver nanoprisms on top [31].

The decrease in exciton lifetime in our samples matches the LSPR caused augmentation of the NCs’ enhancement factor compared to that of NPs very well and thus allows us to quantify the neat plasmonic contribution to the PL increase in our hybrid metal-organic nanocavity arrays. Due to the spatially inhomogeneous electrical field enhancement inside the cavity as indicated in Fig. 2(b), a number of molecules is still unaffected by the plasmonic coupling as shown by the similar monomeric lifetime constant of about 2.45 ± 0.5 ns as in the NP structures.

5. Conclusion

The applied shadow nanosphere lithography technique offers the possibility to scale-up the generation of two-dimensional arrays of organic nanopillars (NPs) and hybrid metal-organic nanocavities (NCs). By systematic variation of the surroundings on nanometer length scales we were able to provide a detailed analysis of the mechanisms leading to PL intensity enhancement in neat ZnPc as well as ZnPc blended Alq₃ matrices prepared either as thin films, NPs or NCs. While the emission increase in Ag/ZnPc/Ag NCs refers to the impact of morphological and geometrical changes, the enhancement in co-evaporated Ag/Alq₃:ZnPc/Ag NCs results mainly from an efficient coupling to the localized surface plasmon resonances of the silver nanoprisms. As demonstrated in comparison to neat films, for the Alq₃:ZnPc silver nanocavities at sufficiently low ZnPc concentration, i.e. in the case of spatially isolated dye molecules, the PL intensity gain can reach up to values of 700 after geometrical correction (factor of about 13). As confirmed by complementary TCSPC measurements, about one order of magnitude of this increase is caused by the decrease in exciton lifetime due to efficient coupling to the plasmonic resonances and is supported by a more directional out-coupling due to the columnar morphology (factor of about 5). Hence, our studies successfully demonstrate a scalable route towards generation of defined metal-organic nanocavity arrays with efficient light coupling properties, having high technological potential for highly integrated OLED structures or photo-detectors with sub-micron pixel sizes.