The application of large area OLEDs for lighting and signage purposes potentially requires essential changes of the common Lambert-like emission pattern. We demonstrate an array based micro optical approach for pattern shaping of area light sources based on distorted Fourier imaging of an aperture array with a micro lens array. Narrow angular emission patterns of ± 35° and ± 18° FWHM obtained experimentally demonstrate the pattern shaping with low stray light levels. The internal recycling of initially rejected photons yields intensity enhancements exceeding a factor two in forward direction that is still well below the theoretical limits due to limited reflectivity.
© 2012 OSA
Organic light-emitting diodes (OLEDs) [1, 2] promise to become efficient sources for general lighting applications and delivering monochromatic or white emission covering the whole visible spectrum from a very thin large area element. Different systems featuring a variety of properties have been demonstrated, the most efficient of which are bottom emitting OLEDs that emit light through the substrate glass into the air . The emission pattern usually exhibits a Lambert-like shape corresponding to a constant brightness for all viewing directions. Any OLED application demanding a certain light distribution inevitably requires optical shaping. Such approaches should retain the slim and light-weight structure of the original source.
The total internal reflection (TIR) at the substrate-air interface causes a significant amount of light emitted internally to be guided within the substrate [4, 5], thus reducing the device efficiency drastically. Record efficiencies have been achieved using hemi-spherical lenses  potentially combined with high refractive index substrates  in order to access the full substrate pattern. Other approaches to disturb the TIR by means of refractive, diffractive, or scattering elements are continuously applied for efficiency enhancement. Regarding refractive substrate outcoupling, elements like lenses [4, 8, 9] or prisms  are commonly used and can be applied to pixelated emitters accordingly .
Besides the pattern emitted inside the substrate glass, the detailed geometric shape of the refractive substrate outcoupling elements influences the far-field distribution generated. Hence, both parameters can be generally optimized . When applying purely refractive elements only, the degrees of freedom to obtain a certain angular far-field distribution or to avoid stray light within a specified angular interval are restricted, especially with regard to the large numerical aperture (NA~1.5) of light distribution inside the substrate glass. Additionally, the OLED light source is about to become a standard “lamp” and external shaping means have to be developed to meet the needs of future applications. These might cover any arbitrary pattern shapes, as demonstrated in Ref . In contrast, structures inside the emitting stack address efficiency goals primarily but provide methods for pattern shaping as well . Because such methods involve elaborate alterations of the active stack (that are generally not accessible) we restrict to external shaping.
In this paper a distorted Fourier imaging approach for an external beam shaping element is described. Commercial large area OLED sources have been equipped with custom made micro lens arrays for substrate coupling. Micro optical shaping arrays were applied to these sources to demonstrate the validity of the system by creating circular far field patterns with ± 35° and ± 18° FWHM. Therefore, the overall approach is introduced in section 2, followed by basic design considerations in section 3. Based upon lithographic technologies all micro optics arrays have been fabricated as explained in section 4. The results achieved will be given with a discussion in section 5.
2. Pattern shaping approach
Pattern shaping approaches of thin, large area sources – for backlights or lighting – generally feature equal extensions of the light source and the shaping optics. According to classical optics, etendue conservation causes an efficiency decrease when the angular interval of light emission is decreased. However, photons rejected by the shaping optics might be reflected sequentially by the shaper and the light source and leave the system in any later attempt. Closed light paths between the source and the optics can be avoided if the angle of propagation is additionally changed during the reflection. Such a “photon recycling” effect breaks the stringent rule of etendue conservation and gives rise to local radiant intensity enhancements in the radiation pattern.
Our general pattern shaping concept applies to any kind of reflective, large area light source. Two major concepts were followed for the present application to OLED: First, rejected photons are recycled by introducing a partially reflective shaping optics. Second, an efficient bottom emitting OLED is utilized that is readily equipped with substrate outcoupling means. As sketched in Fig. 1 a micro lens array (MLA) has been utilized for substrate coupling throughout the paper, but different approaches apply as well for the presented shaping concept. In the following the OLED + MLA will be considered as the light source. The shaping optics is placed on top of this kind of source without any lateral adjustment. The shaper consists of a glass substrate that is covered with an aperture array and a micro lens array, both being aligned with respect to each other. Preparing the aperture array of any well reflecting material ensures a potential recycling of photons, which do not transmit the aperture, as these are directed back to the (reflective) light source.
It is important to note that the detailed properties of the OLED stack as well as of the substrate coupling affect the shaping. An optimized system would take advantage of an adapted active stack  such that substrate emission pattern and thin film stack reflectivity are adjusted to the properties of the micro optic system and vice versa. But, as OLED sources have been made commercially available, shaping approaches should apply to sources without any expensive modification of the light source itself. Therefore, commercially available OLED sources (ORBEOS, OSRAM Opto Semiconductors, Germany) have been used for the experiments described in this work.
3. Optics design
3.1 Basic considerations
Some overall aspects of the shaping element according to Fig. 1 need to be discussed first. When considering single lens Fourier imaging, objects should be located close to the focal plane. The corresponding focal length f of a spherical lens with radius ρ inside the medium with refractive index n is . Although pattern shaping applications are often out of the paraxial limit and significant aberrations arise, apertures close to the paraxial focal plane are expected to yield best distorted Fourier imaging results. Applying the notation from Fig. 1 yields
When illuminating such an array with NA = 1 the marginal rays of the light cone () passing the aperture will reach neighboring lenslets, thus causing ghost images and stray light in the far field. In order to completely avoid such stray light, the lens – aperture distance should be restricted. Considering the marginal ray of the light cone from each aperture and the rim of the corresponding lenslet yieldsEq. (1). This defocus causes blurring of the image, which might be advantageous for smooth brightness distributions (depending on the application). Summarizing that, any design needs to make a compromise between both limiting situations, and small lenslet radii seem to be preferable as this yields the smallest focal length.
Besides the strict geometrical condition in Eq. (2) it is worth pointing out that light might be trapped inside the shaper due to Fresnel or total internal reflection. Additionally, light leaving one lenslet at oblique angles could be recaptured by the neighboring channel. Both effects initially reduce ghosts but cause multiple reflections in-between micro lens and aperture array, finally yielding stray light. Introducing an absorbing upper side of the metal aperture might effectively avoid this emission.
Now, we focus on the most simple aperture type with circular cross section and a relative area α compared to the unit cell. The divergence of the output light cone is governed by the detailed design with all defocus and aberration. In order to provide a reasonable estimation, we assume an ideal imaging of the aperture, resulting in a clipped Lambertian output distribution. Then, etendue conservation leads to the far field divergence angle θff of
3.2 Photon recycling
This estimation of angular divergence is strictly valid only if the optical path length is a unique function of the phase space variable, i.e., of the spatial coordinates and the ray angles . In the case of the intended recycling scheme multiple reflections may break the phase space conservation. Thus, potentially much more light can be concentrated at the aperture area with distinct ray divergence. In order to estimate recycling effects a Lambertian scattering concept [15, 16] is applied and all reflection R, transmission T and propagation p are modeled by average scalar values. Then, the emitted power P of the system can be approximated byEq. (4) describes the power directly going through the shaper with transmission αTs(α), while the denominator accounts for multiple internal reflections due to the metal pinholes (reflectivity (1-α)Rs) and the OLED (reflectivity ROLED). Although not strictly valid due to TIR and Fresnel losses inside the lenslets, the main linear dependency of TS and RS on the relative aperture size α has been readily regarded in Eq. (4). The factor p describes losses during propagation between the OLED and the aperture plane. It should be negligible away from edge losses that can become remarkably. For rather large OLED (lateral size L) and a micro optics height H the condition H/L<<1 holds and p can be estimated by .
Note, that in Eq. (4) the reflectivity ROLED represents the combination of OLED thin film stack and substrate coupling MLA. This value is significantly reduced with respect to that of a planar OLED. As the MLA couples light from air into substrate NA~1.5 reflected photons undergo recycling inside the substrate as well. Therefore, the reflection of light at the OLED with MLA usually involves multiple reflections at the thin film stack, corresponding to increased losses due to limited active stack reflectivity.
Figure 2 visualizes Eq. (4) and its denominator for different values of α. A shaping efficiency equal to unity is obtained with the shaper and the OLED being ideal reflectors and when neglecting edge losses. The dominant losses are expected to originate from a limited reflectivity of either the metal aperture or the OLED. Using Aluminum or Silver as reflecting aperture material enables for approximately 85% and 95% reflectivity, respectively, within the NA = 1 range inside the shaper substrate. In contrast, OLED thin film stack design preliminary ensures stability, lifetime, and efficiency, without paying prior attention to substrate internal reflectivity. With respect to this, Fig. 2 illustrates the enormous potential of power enhancement that can be obtained due to light recycling. The comparison of the quadratic dependence of the transmitted power in the case without recycling (R = 0) and the almost binary characteristics for high values of the product RSROLED necessarily drives any system optimization towards high reflectivity, especially for an intended narrow angular width of the far field emission. Taking into account the far field divergence of a clipped Lambertian output according to Eq. (3) allows estimating the system brightness B, i.e. the power emitted per solid angle, to beFig. 2(b) very large intensity enhancements can be reached for small far field divergence angles. In this case, high reflectivity products are required in order to redistribute the energy into the desired solid angle interval. By contrast, obtaining very small NA requires rather large aperture – lens distances that contradict condition (2). This illustrates a limit towards ideal collimation, although the corresponding range is contained in Fig. 2.
3.3 System design
Any specific system design for a pattern shaping element will start from the basic considerations above. Technological considerations, material data, and basic specifications of the OLED used are required in order to reach an exact layout of the optical system. Both, the OLED’s reflectivity and substrate pattern, can be accessed experimentally and thus be regarded in detailed optical system designs for one given OLED. Based on that, the system can be modeled and optimized with appropriate non-sequential ray-tracing tools. Such tedious numerical discussion is avoided here, giving a qualitative illustration instead.
Two different systems generating a far field angular half width of ± 35° and ± 18° are considered. The first mentioned pattern corresponds to standard spot like lighting, while the second one should generate a very narrow pattern with high brightness enhancement. Figure 3 plots the single transmission α∙TS of the shaper for different configurations of the system, obtained by sequential ray tracing in a quadratic unit cell. Such analysis illustrates qualitative features but neglects recycling effects discussed above. As apparent in Fig. 3 a narrow pattern is generated best by a rather large spacer thickness d, i.e., an aperture – lens separation close to the paraxial focal length. But, an optimum transmission into the desired angular range requires working well above the cross talk limit (Fig. 3(c)), as this would provide near ideal Fourier imaging of the aperture and would generate very steep bright – dark boundaries in the far field’s intensity distribution. In contrast, relaxing the requirements for far field angular divergence shifts the optimum transmission towards lower spacer thickness and thus towards the cross talk limit (Fig. 3(e,f)). Additionally, total transmission increases mainly due to the increased aperture size.
Comparing the diagrams Fig. 3(b) and Fig. 3(e) reveals that the optimum lens radius is near the half ball lens shape, because the short focal length enables working near the cross talk limit. But, half ball lens imaging suffers from very steep lens surfaces that might cause technological difficulties in replication procedures as well as total internal reflection at the lens – air interface. Therefore, systems with optimized lens radii and aperture sizes according to Eq. (3) have been set-up. The detailed layouts have been determined by a non-sequential ray tracing procedure, taking into account multiple internal reflections, absorption, side lobes etc. accurately. For the ± 35° shaper a reasonable compromise of a standard spot lamp angular distribution, an optimum transmission, and a low stray light level at large angles could be achieved using shaping lenslets of 55° contact angle and a relative aperture size of 0.6. The relative aperture has to be decreased to 0.4 for the ± 18° shaper, choosing a lens shape close to the half ball limit.
4. Micro optics fabrication
Micro lenses for substrate coupling as well as for pattern shaping have been replicated by UV molding. The master structures for such replications were fabricated by initially creating binary photo resist patterns and a subsequent reflow process to create spherical micro lenses [17, 18]. An UV transparent mold made of a soft elastormer is replicated from this resist master. Half ball lenses of 30 µm radius with an array fill factor above 90% have been prepared on top of 4” glass wafers for substrate outcoupling. This element is attached to the OLED substrate by means of an immersion oil that might be replaced by any index matched glue in the future.
The generation of the two different apertures was performed by initial magnetron sputtering of optically opaque aluminum layers with a thickness in the 200 nm-range on top of 4” thin glass substrates (D263, Schott). Apertures are generated using standard UV lithography combined with chemical wet etching and resist stripping. After the preparation of the corresponding masters, shaping lens arrays (pitch 200 µm) have been molded on top of the pinhole arrays. Lateral alignment as well as spacer thickness adjustment between pinholes and shaping lenslets is achieved in a mask aligner (Suess Microtec). Images of the shaper array prepared on 4” diameter substrates are shown in Fig. 4 .
An absorptive layer on top of the pinhole array could be prepared using black-matrix polymer resist material (PSK 2000, Brewer Science) patterned by UV-lithography. As the black matrix resist structure can be used as mask for the Aluminum wet etching, additional lithography and alignment steps can be avoided. This technology has been successfully used in a series of comparable non-imaging as well as imaging micro optical configurations [19, 20].
With regard to the fabrication of shaping elements, the pinhole – lens alignment, the lens curvature, and the spacer thickness d need to be ensured in order to achieve proper optical functionality. While lens curvature and spacer thickness tolerances yield deviations in the far field angle, any pinhole – lens misalignment introduces a decentering of the maximum emission angle. Such effects might be critical depending on detailed tasks. We ensured a lateral position accuracy of 2 µm for the present case of 200 µm pitch, but the system dimensions could be scaled in order to obtain reasonable tolerances. In contrast, the substrate coupling lens array has no pattern shaping functionality. Its lens shape and glue thickness need to ensure optimum coupling of energy only and even bubbles inside the glue do not compromise the shaping functionality.
5. Experimental results and discussion
All experiments were performed with a commercially available OLED (ORBEOS, OSRAM Opto Semiconductors, Germany) without initially mounted outcoupling elements. Angular resolved radiation patterns have been measured by mounting the system on a rotational stage. A multimode fiber (NA 0.22) with 200 µm core diameter was positioned in front of source and connected to a fiber spectrometer (SD2000, Ocean Optics) as detector. The distance between the source and the detector fiber ensured an angular resolution of the measurement better than 1.5°. A constant current source (GS610, Yokogawa) ensured reliable driving conditions. Radiation patterns have been calculated from the emission spectra by integrating the spectral intensity distribution for each angle of observation.
Figure 5 depicts the patterns obtained with the shaping elements as well as the patterns for the OLED with and without substrate coupling MLA. The well-known emission enhancement due to the micro lens array for substrate coupling results in an increase of the perpendicular emission in the 60..65% range. Adding the shaping element yields the desired suppression of emission into oblique angles while enhancing the perpendicular emission. This enhancement reaches a value of 195% compared to the planar OLED in case of the ± 35° pattern. As stated in the theoretical section, the maximum enhancement due to recycling effects can be obtained when further decreasing the angular width of far field emission. It exceeds 200% for the ± 18° pattern in Fig. 5. Note that both examples yield low stray light levels at oblique angles >60° where emission is in the range of 1% relative to maximum only. In order to illustrate the appearance of such a light source, Fig. 6 shows two photographs of the system taken at oblique and near perpendicular observation.
It should be noted, that introducing substrate coupling elements to a white emitting system generally alters the emission spectrum. According to the properties of the active thin film system the substrate spectrum differs from the one observed in air. Such an effect arises in our experiments when adding the substrate coupling MLA. But this spectrum remains virtually constant when further adding the shaping element. The reason is that the angular far field emission pattern is in first order approximation a Fourier image of the lateral intensity distribution in the aperture plane. Photon recycling and angular mixing by the outcoupling MLA provide a homogeneous lateral distribution there, resulting in an angular independent color of the far field pattern. However, any in depth analysis of angular dependent color variations is out of the scope of this paper as it requires considering the spectral properties of the thin film stack and of the mirror in combination with the total wavelength and angular dependent emission inside the substrate.
Although recycling obviously enhances the brightness of the pattern shaped systems, the overall efficiency drops to 65% (~40%) and 30% (~19%) for the ± 35° and ± 18° shaper, respectively, in comparison to the OLED without (with) substrate coupling MLA. This indicates a limited recycling of photons for the systems under study. As readily stated in section 3 this is due to two major reasons: First, for an OLED stack such as given in , comprising a Silver or Aluminum cathode combined with slightly absorbing organic active layers, reflectivity inside the substrate around 0.7 is feasible. This value drops to ~0.5 due to the presence of the substrate coupling MLA. Second, assuming 78 mm active area size and ~2 mm height of OLED substrate with micro optics, edge losses (p2) are roughly estimated to be in the 5..10% range still.
Returning to Fig. 2 with the relative efficiencies obtained and keeping the corresponding values of NA~0.57 and NA~0.30 in mind, allows for estimating the (averaged) reflectivity product in the range of RSp2ROLED~0.4. This result is well within the expected range when considering the losses discussed above.
The shaping results have been obtained without introducing an absorbing layer on top of the aperture. Within experimental errors analog systems utilizing such absorber yield the same emission curves (patterns not shown). This behavior can be attributed to the fact, that even the black polymer layer yields non-vanishing Fresnel reflections, giving rise to comparable stray light levels for both systems.
The emission pattern of an OLED as a white, large area, originally Lambertian emitting source has been strongly altered. Due to the Fourier imaging-like approach nearly complete stray light suppression at large observation angles becomes possible. This might become a feature whenever a large area illumination with narrow angular spectrum is desired. Utilizing photon recycling effects enables for perpendicular intensity enhancements in the range of a factor two. This value might be further improved by introducing increased reflectivity for either the OLED stack or the pinhole array. Recycling effects are especially advantageous in the case of very narrow angular patterns ( ± 18°). Such sources could become illumination candidates for ultra-thin projection systems  in the near future.
Existing lithographic technology has been utilized for preparing the shaping systems in 4” wafer scale. As OLED lighting is about to transit to volume lighting market, alternative technological routes for large area fabrication, either on rigid glass or flexible polymer substrates, will be required. The coating of glass microspheres on absorbing resins  or derived schemes might pave the way towards reasonable fabrication procedures.
According to Fig. 1 all optical channels of the shaper provide the same angular far field distribution. A more generalized approach could utilize a multitude of different optical channel designs, the superposition of which generating the desired far-field pattern. Applying such schemes to laterally patterned light sources might enable for pattern switching, and finally approach to imaging-like [19, 20] or projection-like  optical micro systems.
The authors thank the German Federal Ministry of Education and Research (BMBF) for funding under contract FKZ 13N10474 (TOPAS 2012).
References and links
1. C. W. Tang and S. A. Van Slyke, “Organic electroluminescent diodes,” Appl. Phys. Lett. 51(12), 913–915 (1987). [CrossRef]
2. J. H. Burroughes, D. D. C. Bradley, A. R. Brown, R. N. Marks, K. Mackay, R. H. Friend, P. L. Burns, and A. B. Holmes, “Light-emitting diodes based on conjugated polymers,” Nature 347(6293), 539–541 (1990). [CrossRef]
4. M.-H. Lu and J. C. Sturm, “Optimization of external coupling and light emission in organic light-emitting devices: modeling and experiment,” J. Appl. Phys. 91(2), 595–604 (2002). [CrossRef]
5. N. Danz, D. Michaelis, and C. Wächter, “OLED design: combined micro- and nanophotonics modeling, and routes to a complex optimization algorithm,” Proc. SPIE 6796, 67963F, 67963F-15 (2007). [CrossRef]
6. Z. B. Wang, M. G. Helander, J. Qiu, D. P. Puzzo, M. T. Greiner, Z. M. Hudson, S. Wang, Z. W. Liu, and Z. H. Lu, “Unlocking the full potential of organic light-emitting diodes on flexible plastic,” Nat. Photonics 5(12), 753–757 (2011). [CrossRef]
7. S. Reineke, F. Lindner, G. Schwartz, N. Seidler, K. Walzer, B. Lüssem, and K. Leo, “White organic light-emitting diodes with fluorescent tube efficiency,” Nature 459(7244), 234–238 (2009). [CrossRef] [PubMed]
8. S. Möller and S. R. Forrest, “Improved light out-coupling in organic light emitting diodes employing ordered microlens arrays,” J. Appl. Phys. 91(5), 3324–3327 (2002). [CrossRef]
9. J. P. Yang, Q. Y. Bao, Z. Q. Xu, Y. Q. Li, J. X. Tang, and S. Shen, “Light out-coupling enhancement of organic light-emitting devices with microlens array,” Appl. Phys. Lett. 97(22), 223303 (2010). [CrossRef]
10. B. C. Krummacher, M. K. Mathai, V. Choong, S. A. Choulis, F. So, and A. Winnacker, “General method to evaluate substrate surface modification techniques for light extraction enhancement of organic light emitting diodes,” J. Appl. Phys. 100(5), 054702 (2006). [CrossRef]
11. P. Melpignano, V. Biondo, S. Sinesi, M. T. Gale, S. Westenhöfer, M. Murgia, S. Caria, and R. Zamboni, “Efficient light extraction and beam shaping from flexible, optically integrated organic light-emitting diodes,” Appl. Phys. Lett. 88(15), 153514 (2006). [CrossRef]
12. M. Flämmich, D. Michaelis, N. Danz, C. A. Wächter, P. Dannberg, and A. H. Bräuer, “Micro-optical beam-shaper for tailoring light emission from OLEDs,” Proc. SPIE 7716, 771616, 771616-10 (2010). [CrossRef]
13. K. Saxena, V. K. Jain, and D. S. Mehta, “A review on the light extraction techniques in organic electroluminescent devices,” Opt. Mater. 32(1), 221–233 (2009). [CrossRef]
14. R. Winston, J. C. Minano, and P. Benitez, Nonimaging Optics (Elsevier Academic Press, Amsterdam, 2005).
15. A. Goetzberger, “Optical confinement in thin Si solar cells by diffuse back reflectors,” in Proceedings of the 15th Photovoltaic Specialists Conference (Institute of Electrical and Electronics Engineers, New York, 1981), pp. 867–870.
16. E. Yablonovitch, “Statistical ray optics,” J. Opt. Soc. Am. 72(7), 899–907 (1982). [CrossRef]
18. D. Daly, R. F. Stevens, M. C. Hutley, and N. Davies, “The manufacture of microlenses by melting photoresist,” Meas. Sci. Technol. 1(8), 759–766 (1990). [CrossRef]
20. A. Brückner, J. Duparré, R. Leitel, P. Dannberg, A. Bräuer, and A. Tünnermann, “Thin wafer-level camera lenses inspired by insect compound eyes,” Opt. Express 18(24), 24379–24394 (2010). [CrossRef] [PubMed]
21. M. Flämmich, J. Frischeisen, D. S. Setz, D. Michaelis, B. C. Krummacher, T. D. Schmidt, W. Brütting, and N. Danz, “Oriented phosphorescent emitters boost OLED efficiency,” Org. Electron. 12(10), 1663–1668 (2011). [CrossRef]
22. M. Sieler, P. Schreiber, P. Dannberg, and A. Bräuer, “Array projection optics: Multi-channel design for ultra slim projectors,” Proc. SPIE 7716, 7716A 1–10 (2010).
23. G. P. Morris, J. C. Rolfs, L. A. Meyer, R. S. Moshrefzadeh, H.-H. Chou, B. J. Tompkins, and T. N. Davis, “Glass microspheres for use in films and projection screen displays and methods,” Patent US 6.204.971 (2001).