Open Access
Add to BibSonomyAbstract
The generation of diffractive optical elements often requires time and cost consuming production techniques such as photolithography. Especially in research and development, small series of diffractive microstructures are needed and flexible and cost effective fabrication techniques are desirable to enable the fabrication of versatile optical elements on a short time scale. In this work, we introduce a novel process chain for fabrication of diffractive optical elements in various polymers. It is based on a maskless lithography process step, where a computer generated image of the optical element is projected via a digital mirror device and a microscope setup onto a silicon wafer coated with photosensitive resist. In addition, a stitching process allows us to microstructure a large area on the wafer. After development, a soft stamp of the microstructure is made from Polydimethylsiloxane, which is used as a mold for the subsequent hot embossing process, where the final diffractive optical element is replicated into thermoplastic polymer. Experimental results are presented, which demonstrate the applicability of the process.
© 2015 Optical Society of America
1. Introduction
Diffractive optical elements (DOE) are a versatile tool in nearly any optical application including beam shaping, imaging, and holography [1]. While glass as a material for DOE is inevitably connected to high production cost, several polymers such as Polymethylmethacrylate (PMMA) or Cyclic-Olefin-Copolymers (COC) have gained increasing interest in industrial as well as research applications as an alternative material [2]. Polymers offer the possibility to employ a wide range of low-cost and high throughput reproduction processes for microstructures, which make DOE accessible for the mass market [1–3]. Among the most common replication techniques injection molding is widely used [2]. During the process, a polymer is heated above its melting temperature and is injected into a master mold under high pressure, cooled down, and released from the mold. The great advantage of injection molding is a high throughput compared to, e.g., hot embossing. There, a master stamp and a polymer substrate are heated above the glass transition temperature of the polymer. Subsequently, the master stamp is imprinted into the substrate, cooled below the polymer’s glass transition temperature, which is followed by a demolding process step [3]. However, in academic and industrial research and development, high throughput is not required but high flexibility in master stamp production and replication in polymer substrates are key issues, which makes hot embossing the most favorable replication technique [3]. The master stamp for hot embossing is often made from silicon or metal [2, 3]. Silicon offers a wide range of techniques for microstructuring such as various etching techniques [4]. Alternatively, stamps are fabricated in metal by ultra precision milling or turning [5]. Both, silicon and ultra precision turning and milling technologies are very time and cost consuming and limit their applicability in research and development.
A more cost effective way to produce microstructures in polymers is to create an inverse of the master geometry in photoresist by lithography, which is subsequently replicated in Polydimethylsiloxane (PDMS) [3]. The PDMS stamp is used as master stamp for hot embossing. The drawback of this technique is that a lithography mask is still required, which limits the flexibility of the process. A very versatile method is micromaskless lithography, which was developed for industrial and research applications. In general, these techniques can be distinguished into two groups. The first group relies on single point processing, where a focused laser beam is moved over a substrate material, which is coated with photoresist [6]. Such a technique was used, for example, by Gale et al. to write continuous reliefs in photoresist, which were transferred into nickel shims by electroplating after development of the resist, followed by a hot embossing process to replicate the optical elements [7]. Direct laser writing of optical elements in glass is also reported [8]. Lasers were also used to support etching processes in glass, which enables direct writing by selectively activating an etching region on the glass substrate [9]. In polymers, recent work focuses on direct writing of three-dimensional structures such as photonic crystals [10]. To increase the lateral resolution of single point laser writing, two photon polymerization (2PP) was extensively studied to produce optical elements including photonic crystals [11]. While single point processes greatly enhance the flexibility compared to mask supported lithography, process times are often extraordinary high and can reach up to several hours. Therefore, latest work focuses on a second group of maskless lithography processes utilizing spatial light modulators (SLM), which generate a binary or a grayscale intensity pattern. The pattern is directly projected onto a substrate, which is coated with photosensitive resist [12]. A binary pattern is utilized to obtain microstructures with only two different height levels. Alternatively, a pattern with different intensity values can be generated to create microstructures with more than two height levels, which is usually denoted as grayscale lithography [1]. If the light source utilized emits coherent light, the SLM can also be used to generate holographic intensity distributions to obtain three dimensional structures [14]. The latter technique was also extended for writing three-dimensional structures in polymer using 2PP [15]. The applications of maskless structuring range from fabrication of microelectromechanical systems (MEMS) [16] to optical elements [12]. As SLM, either liquid crystal displays (LCD) or digital micro mirror devices (DMD) are used [13–17]. The advantage of the latter is a lower energy absorption and transmission in the ultraviolet wavelength region but it requires more sophisticated beam shaping when using incoherent or partially coherent light sources as theoretically investigated [17].
In this work, we combine DMD based maskless lithography and soft stamp hot embossing for large area production of diffractive optics. With this novel process chain, we are able to fabricate a hot embossing mold with arbitrary geometry. The mold is embossed into polymer to create the diffractive microstructure, subsequently. The emphasis of our work lies on large area, low-cost, and fast production by soft stamp lithography rather than on complicated three-dimensional structures, which are not suitable for hot embossing due to undercuts. Our maskless lithography setup is based on a standard DMD and a simple microscope setup, which demagnifies the DMD and projects its image onto a silicon wafer coated with photosensitive resist. Since the area which is exposed at a time is limited by the physical extent of the DMD and the field of view of the microscope, we utilize a piezoelectric actuated translation stage to expose a sequence of images next to each other onto the wafer. From a scientific point of view, the presented process chain is not only capable of fabricating diffractive optical elements and structures but also opens the possibility for investigation of a large variety of novel micro optical structures in polymers equipped with extensive optical functionality including fiber based optical sensors. For example, the structures produced could lead to novel fiber or waveguide based polymeric networks for distributed optical sensing, e.g. measurement of quantities such as temperature and strain, using photonic components [18]. In addition, our process is capable of creating basis structures for, e. g., Fiber-Bragg-Gratings, ring resonators and even more complex systems such as Arrayed-Waveguide-Gratings [19–21]. Hence, the presented methodology may also be relevant for research aiming at applications of such structures as integrated polymer sensors.
2. Production process
2.1. Master stamp production by maskless lithography
To manufacture the master stamp for hot embossing, we produced an inverse of the stamp in photoresist (Shipley S1813) utilizing a self-made maskless lithography system. The lithography setup is shown in Fig. 1. A high power LED with a center wavelength of 435 nm and 350mW optical power is collimated by a DOE and a lens and illuminates a DMD device (Vialux ALP 4.1) with a pixel pitch of 13.8 μm at a resolution of 1024×786 pixels. The exposure time of the setup is controlled via an Arduino micro controller board and was set to 10s. Before exposure, a binary pattern of the desired microstructure is generated on a personal computer and uploaded onto the DMD device. After the image is displayed by the DMD, the exposure is initialized by a trigger signal, which is send to the Arduino board by the personal computer. During the exposure time, the binary pattern is projected onto a silicon wafer by a standard microscope setup consisting of a tube lens (Carl Zeiss) and a highly corrected plan-achromatic objective (Carl Zeiss Epiplan) with a numerical aperture (NA) of 0.3 and a magnification of 10. The field of view of the microscope is approximately 1.4 mm × 1.1 mm at a resolution of 1.35 μm. Higher resolutions can be achieved by using higher NA lenses such as immersion lenses with a NA up to 1.4, which yields a theoretical resolution limit of the setup of 270 nm. The silicon wafer was placed on a piezo-electrical translation stage (Physik Instrumente GmbH), which is able to move the wafer in x- and y- direction perpendicular to the optical axes of the microscope. The stage has a maximum travel range of 50 mm × 50 mm and enables stitching of images to increase the area on the wafer, which can be used for lithography to the maximum travel range of the translation stage. To obtain a sharp image of the pattern to be projected onto the wafer, the pattern is imaged onto a CCD camera (Thorlabs) via a beam splitter and a second tube lens. By adjusting the distance between the wafer and the microscope along the optical axis of the objective (z-axis) manually, we obtained a sharp projection of the pattern on the surface of the silicon wafer prior to the actual lithography process. As substrate material, we used a polished standard silicon wafer for the inverse master stamp. A thin film of photoresist (Shipley S1813) was spin coated onto the substrate to obtain a homogeneous layer of resist with a thickness of approximately 500 nm. After spin coating, we applied a prebake at 115°C for 60 seconds.
As an example for a diffractive microstructure, we chose a logo (of our collaborative research center PlanOS), which contains a rectangular diffraction grating with a period of 5.2 μm. The logo was fabricated using 5×5 single patterns, which were transferred into the resist by applying the stitching process yielding a microstructured area of 7 mm × 5 mm. After exposure, the resist was developed for 60 s in a MF-324 developer bath.
2.2. Stamp production and replication by soft stamp hot embossing
Since the photosensitive resist on the silicon substrate is highly sensitive to thermal loads, it is not suitable as stamp material for direct hot embossing. As an alternative, we used a soft stamp made from PDMS (Sylgard 184), which is a negative copy of the microstructures in photoresist. The process steps for transferring the microstructures made by maskless lithography into polymers are shown in Fig. 2. First, we coated a layer of PDMS onto the microstructured surface of the silicon substrate. Subsequently, it was cured at room temperature for 48 hours. The PDMS was removed after curing and placed in a hot embossing machine and used as soft stamp. The hot embossing process was carried out in a commercial hot embossing machine HEX03 from Jenoptik. As polymer material, we used PMMA sheets with a thickness of 500 μm. During the hot embossing process, the PDMS mold and the PMMA substrate were heated up to 135°C, which is well above the glass transition temperature of 105°C of the PMMA [3]. Subsequently, an embossing pressure of 6.2 kPa was applied for 5 min. After cooling to a release temperature of 40°C, the PDMS mold was removed from the PMMA manually.
3. Experimental results and discussion
Figure 3(a) shows a photograph of a hot embossed logo with an incorporated diffraction grating with a period of 5.2 μm, which was fabricated using our maskless lithography setup and soft stamp hot embossing. The logo was embossed into a PMMA foil with a thickness of 500 μm. On the photograph, the optical grating inside the logo splits the incident white light into its spectral components. The logo consists of 5 × 5 single images, which were combined to a single large area diffractive structure by applying the stitching process described in the previous sections. Please note, that in Fig. 3(a) no overlap between single images is observable, which leads to the desired uniform appearance. The enlarged image in Fig. 3(b) shows a microscope image of the structured resist on the silicon substrate, which was used for PDMS stamp production. Due to the maskless setup, we are able to produce an arbitrary geometry of the diffractive structure in resist, which yields the macro-geometry of the logo. To quantify the geometry error introduced by the stitching process, we recorded a microscope image of the stitching region, which is shown in Fig. 3(c). The stitching error in y-direction accounts for approximately 300nm, which corresponds to the positioning accuracy of the translation stage and is well below the pixel size of 1.35 μm. The stitching offsets in x- and y-direction were calibrated using the diffraction pattern used for the logo. The grating vector of the diffraction grating points in the y-direction, which leads to a well calibrated stitching offset in y-direction but a rather large stitching offset of approximately 2 μm in x-direction. Using diffraction patterns with different orientations of the gratings will also improve the stitching offset in x-direction in the future. The measured offset of 300nm is sufficiently small when using a pixel size of 1.35 μm but needs to be improved in further work if feature sizes will decrease to the sub-micron range.
Fig. 3 Photograph of a logo with an incorporated diffraction grating replicated in a thin PMMA foil (a), magnified image of the diffractive structure on the logo (b), and microscope image of a diffraction grating within the stitching region (c).
To evaluate the accuracy of each molding step, we recorded microscope images of characteristic features of the microstructures. These features ensure that each image was taken at the same sections on stamp, PDMS mold, and PMMA replica. An example is shown in Fig. 4, where Fig. 4(a) shows the inverse master mold produced by maskless lithography, Fig. 4(b) the negative PDMS copy, and Fig. 4(c) the replicated PMMA foil. The PDMS stamp in Fig. 4(b) is inverted because it is a negative copy of the microstructure shown in Fig. 4(a). In general, the microstructures are well replicated with no observable defect due to the replication process. The red arrows in Fig. 4(c) indicate the smallest realizable feature size of 1.35 μm, which we achieved utilizing an objective with a NA of 0.3 at a magnification of approximately 10. Therefore, the smallest feature size corresponds to the pixel pitch of the DMD of 13.8 μm, which is demagnified down to 1.35 μm. The theoretical diffraction limited resolution of the microscope setup at a wavelength of 435nm is 884nm when using an objective with a NA of 0.3. Hence, the optical setup resolves 8.9 μm in the image plane on the DMD chip. As a consequence, a single mirror on the DMD is resolved by the setup. Since the mirrors exhibit a rectangular shape and are inclined by approximately 12°, the image on the wafer and, therefore, the developed microstructure show defects due to the mirror edges. This leads, for example, to periodic structures at straight edges of gratings, which are observable in the enlarged section of Fig. 4(c). To avoid these artifacts, one could either use a high definition DMD with a pixel size smaller than the diffraction limited resolution of the microscope or decrease the NA of the lens and, hence, the resolution of the setup. The first method is more favorable since it also decreases the total resolution of the setup below 1.35 μm. However, further work will also include using objectives with higher NA to create smaller feature sizes. For example, using an objective with a NA of 0.7 and a magnification of 50 yields a diffraction limited resolution of 380nm on the wafer, which corresponds to a resolution of 19 μm in the DMD image plane. Hence, using a setup with an objective having a NA equal or larger than 0.7 results in unresolved pixels and, thus, in straight edges when creating microstructures.
Fig. 4 Microscope images of the diffractive grating structure in resist (a), its PDMS copy (b), and the hot embossed structure in PMMA (c). The inset shows the enlarged region in (c) indicated by the red square.
To evaluate the accuracy of the whole process and possible defects of the microstructures during replication, we carried out confocal surface topography measurements (Keyence VK X100) at the same sections of the inverse stamp, the PDMS master mold, and its PMMA replica. These sections correspond to the ones shown in Fig. 4. Cross-sections of the confocal topography measurements are shown in Fig. 5, which were measured along the grating vector of the diffraction grating. One would expect a rectangular grating profile due to the lithography setup, while the measured profile shows inclined flanks. This inclination is caused by a removal of edge artifacts in the confocal measurement data, such as overshooting at edges, followed by interpolation of missing data points. In general, the entire topography of a microstructure cannot be determined by standard tactile or optical topography measurement devices. However, the measurements yield reliable data of period and profile heights. The result of the measurements are given in table 1. The initial period of the grating in photoresist of 5.22 μm increases slightly during the process leading to a period of 5.47 μm of the PMMA structure. This increase is caused by a deformation of the PDMS mold during hot embossing. Furthermore, the profile heights of resist, PDMS mold and PMMA replica differ by a few ten nanometers, which is a result of the polymerization process during curing of the PDMS and thermal expansion of the mold during hot embossing [22].
Fig. 5 Surface profiles measured by confocal microscopy of the initial microstructure on silicon substrate (top), PDMS soft stamp (middle), and its PMMA replica made by soft stamp hot embossing (bottom).
Table 1. Measured period and profile height determined from the measured profiles in Fig. 5
However, the presented results demonstrate that the process chain introduced in this paper is capable of creating reproducible microstructures with feature sizes of 1.35 μm, an arbitrary macro geometry, and height levels in the few hundred nanometer range. Further work will also include immersion lithography with higher NA lenses to decrease the feasible feature size well blow 1 μm as well as grayscale lithography to produce multilevel microoptics in the sub-micron range. Especially, the production of blazed diffractive gratings is an interesting application of the presented process chain. Here, the minimum achievable period of such gratings is an important feature and depends on the number of height levels. If we consider eight height levels, the minimum period with the current setup is limited to 10.8 μm, which is eight times the resolution of the setup when using an objective with a NA of 0.3. However, if objectives with a NA of up to 1.4 are used, the minimum period is expected to be reduced to 1.5 μm. Further applications of the presented work include holographic structures for product copy-protection, micro-optical devices and microfluidic platforms for lab-on-chip sensors.
4. Conclusion
We presented a process chain for the production of polymer based diffractive optical elements, which is based on replication of a master stamp into polymer by hot embossing. The master stamp is a PDMS copy of a microstructure in photosensitive resist, which is fabricated by a lablevel maskless lithography setup. The optical setup consists of a simple microscope, a DMD device, and a high power light emitting diode. The micro- and macrostructure of the diffractive element is generated on a personal computer, transferred to the DMD, and projected onto the photoresist. Larger areas for microstructuring are achieved by using a piezoelectrical translation stage, which is used for stitching a sequence of single images of approximately 1.4 × 1.1mm2 size to a maximal size of several square centimeters. The smallest feasible feature size of the setup is 1.35 μm utilizing a microscope objective with NA = 0.3. The lithography setup is made from standard optical components and easy to implement in a standard optical lab environment. Our process efficiently enables low cost trial series for applications in research and development. In combination with soft stamp hot embossing, which is also easy to implement with low-cost components, our process chain offers fast and versatile production of large area diffractive optical elements and holographic structures at low production costs. In future, the process chain presented also offers the potential to investigate a broad range of polymer based micro optical elements and devices such as micro ring resonators or Fiber-Bragg-Gratings, which could lead to novel and highly-functional integrated optical sensing architectures for applications in fields as diverse as medicine, life sciences or aeronautics. Further work will also focus on the fabrication of multilevel structures such as blazed gratings using non-binary intensity patterns as well as on an improvement of the stitching offset using positioning stages with a bidirectional position accuracy well below 50nm. We estimate that when calibrating the setup carefully, a stitching offset below 30nm is achievable.
Acknowledgments
This work was funded by the German Research Foundation ( DFG) in the framework of the collaborative research center “PlanOS” and the German Federation of Industrial Research Associations (AiF) under grant IGF 16123/N. We acknowledge support by Open Access Publishing Fund of Leibniz Universität Hannover.
References and links
1. H.-P. Herzig, (ed.), Micro-optics - Elements, Systems and Applications (Taylor & Francis, 1997).
2. S. Bäumer, (ed.), Handbook of Plastic Optics (Wiley-VCH, 2005). [CrossRef]
3. M. Worgull, Hot Embossing: Theory and Technology of Microreplication (William Andrew, 2009).
4. S. Sinzinger and J. Jahns, Microoptics (Wiley-VCH, 2003). [CrossRef]
5. X. Luo, K. Cheng, D. Webb, and F. Wardle, “Design of ultraprecision machine tools with applications to manufacture of miniature and micro components,” J. Mater. Process. Technol. 167, 515–528 (2005). [CrossRef]
6. Y.-C. Chan, Y.-L. Lam, Y. Zhou, F.-L. Xu, C.-Y. Liaw, W. Jiang, and J. Ahn, “Development and applications of a laser writing lithography system for maskless patterning,” Opt. Eng. 37(9), 2521–2530 (1998). [CrossRef]
7. M. T. Gale, M. Rossi, J. Pedersen, and H. Schuetz, “Fabrication of continuous-relief micro-optical elements by direct laser writing in photoresists,” Opt. Eng. 33(11), 3556–3566 (1994). [CrossRef]
8. A. Smuk and N. Lawandy, “Direct laser writing of diffractive optics in glass,” Opt. Lett. 22(13), 1030–1032 (1997). [CrossRef] [PubMed]
9. F. Chen, H. Liu, Q. Yang, X. Wang, C. Hou, H. Bian, W. Liang, J. Si, and X. Hou, “Maskless fabrication of concave microlens arrays on silica glasses by a femtosecond-laser-enhanced local wet etching method,” Opt. Express 18(19), 20334–20343 (2010). [CrossRef] [PubMed]
10. M. Deubel, M. Wegener, S. Linden, G. von Freymann, and S. John, “3D-2D-3D photonic crystal heterostructures fabricated by direct laser writing,” Opt. Lett. 31(6), 805–807 (2006). [CrossRef] [PubMed]
11. J. Serbin, A. Egbert, A. Ostendorf, B. Chichkov, R. Houbertz, G. Domann, J. Schulz, C. Cronauer, L. Fröhlich, and M. Popall, “Femtosecond laser-induced two-photon polymerization of inorganic organic hybrid materials for applications in photonics,” Opt. Lett. 28(5), 301–303 (2003). [CrossRef] [PubMed]
12. L. H. Erdmann, A. Deparnay, F. Wirth, and R. Brunner, “MEMS-based lithography for the fabrication of micro-optical components,” Proc. SPIE 5347, 79–84 (2004). [CrossRef]
13. B.-A. Behrens, R. Krimm, J. Jocker, E. Reithmeier, B. Roth, and M. Rahlves, “Method to emboss holograms into the surface of sheet metals,” Key Eng. Mater. 549, 125–132 (2013). [CrossRef]
14. G. Vizsnyiczai, L. Kelemen, and P. Ormos, “Holographic multi-focus 3D two-photon polymerization with real-time calculated holograms,” Opt. Express 22(20), 24217–24223 (2014). [CrossRef] [PubMed]
15. N. J. Jenness, R. T. Hill, A. Hucknall, A. Chilkoti, and R. L. Clark, “A versatile diffractive maskless lithography for single-shot and serial microfabrication,” Opt. Express 18(11), 11754–11762 (2010). [CrossRef] [PubMed]
16. J.H. Lake, S.D. Cambron, K.M. Walsh, and S. McNamara, “Maskless grayscale lithography using a positive-tone photodefinable polyimide for MEMS applications,” J. Microelectromech. S. 20, (6)1483–1488 (2011). [CrossRef]
17. C. Liu, X. Guo, F. Gao, B. Luo, X. Duan, J. Du, and C. Qiu, “Imaging simulation of maskless lithography using a DMD,” Proc. SPIE 5645, 307–314 (2005). [CrossRef]
18. C. Kelb, E. Reithmeier, and B. Roth, “Foil-integrated 2D optical strain sensors,” Procedia Technology 15, 711–716 (2014). [CrossRef]
19. M. Rosenberger, G. Koller, S. Belle, B. Schmauss, and R. Hellmann, “Polymer planar Bragg grating sensor for static strain measurements,” Opt. Lett. 38(5), 772–774 (2013). [CrossRef] [PubMed]
20. R. Kirchner, A. Finn, R. Landgraf, L. Nueske, L. Teng, M. Vogler, and W.-J. Fischer, “Direct UV-imprint of hybrid-polymer photonic microring resonators and their characterization,” J. Lightwave Technol. 32(9), 1674–1680 (2014). [CrossRef]
21. Q. Zhengkun, Y. Yue, S. Jia, Z. Huiping, W. Guofeng, S. Yongxin, and Wang Yuhai, “Analysis of fabrication results for 17 × 17 polymer arrayed waveguide grating multiplexers with flat spectral responses,” J. Semicond . 34, 094011 (2013). [CrossRef]
22. S. W. Lee and S. S. Lee, “Shrinkage ratio of PDMS and its alignment method for the wafer level process,” Microsyst. Technol. 14(2), 205–208 (2008). [CrossRef]
