Photopolymerisation by scanning a focused laser beam is a powerful method to build structures of arbitrary complexity with submicrometer resolution. We introduce parallel photopolymerisation to enhance the efficiency. Instead of multidimensional scanning of a single focus, the structure is generated simultaneously with diffractive patterns. We used fixed diffractive optical elements (DOEs), kinoforms, and Spatial Light Modulators (SLMs). The possibilities of photopolymerisation using SLM were investigated: the added flexibility using the programmable device is demonstrated. By using these DOEs, straight and helical cross shaped columns were produced with a single scan at a rate about an order of magnitude faster than by simple scanning. The produced helical structures could be rotated by optical tweezers.
© 2007 Optical Society of America
Building of micrometer sized structures to perform different tasks in microfluidics is an emerging field that requires the use of very diverse techniques. The tasks include e.g. heating , sorting [2,3], directing , and stirring of liquid. Such elements can be built by silicon-based microlithographic methods [5–7]. In this procedure 3D objects are built in several steps layer-by-layer repeating production processes like coating with photoresist, illumination, etching, doping, etc. Laser ablation methods are also available for material microprocessing .
Photopolymerisation is an alternative method to produce micrometer sized 3D structures. This procedure is a direct production using focused laser beam and photopolymer. In this case a laser beam is focused into the photopolymer, and with carefully chosen parameter set polymerisation occurs in the close vicinity of the focal spot. By scanning the focal point in the photopolymer along a pre-defined trajectory entire objects can be built in one or a few number of steps. One can scan either the sample relative to the laser focal spot or vice versa. The spatial resolution of the method can be improved by the use of femtosecond lasers, in this case two photon absorption (TPA) can be achieved in the photopolymer, reducing the size of the polymerised region [9,10]. Recent papers reported successful two photon polymerisation of organic-inorganic materials with sub-micron resolution . The method is capable of generating different microscopic mechanical devices like springs [12,10], gears [13,14] or micromanipulator arms . The actuation of photopolymerised objects mainly involves laser tweezers which either simply holds or relocates  or, in case of special-shaped objects also rotates them [6,16]. An optical motor including surface-attached micro gear and integrated light-guide was also created . This fully integrated system represents a next step in the development of optically driven micromachines: optical tweezers are not needed for the function.
In this paper we demonstrate that manufacturing of micrometer sized structures with increased efficiency can be achieved using laser beam shaping methods. We introduce and characterize basic procedures to improved several properties of the photopolymerisation technique. The purpose of this approach is to accelerate the fabrication process either by polymerising an entire structure using complex light patterns and/or achieving simultaneous polymerisation of the same object in several copies at the same time using several focal spots. With a single polymerising focal spot the production of a complex structure requires (multiple) scans along all three axes. As an alternative, using appropriate beam shaping that produces focal spots of special 2D arrangement or continuous 2D light patterns the polymerisation either will not require the complicated or repeated 3D scanning of the sample at all, but only a static illumination or it requires just a simple 1D scan. From the reduced number of scans an increased production rate of the polymerised microstructures follows.
We studied the possibilities of the parallel polymerisation with several laser beams generated by splitting the original polymerising beam using diffractive optical elements (DOE). The split and then focused beams form several individual polymerising focal spots in a predefined arrangement in the photopolymer. If the diameter of the polymerised area, defined by the two-photon polymerisation threshold , is larger than the distance of the centers of the adjacent focal spots, the polymerised resin in the neighboring focal spots merge. Thus a single extended object is formed instead of several individual voxels. This way a predetermined structure is polymerized in a single shot. In addition, these basic light patterns can be scanned to produce structures of higher complexity.
We used two different types of DOE to shape the polymerising beam. One is a passive, transmission-type DOE (a kinoform), and the other is an active, computer-controlled reflective Spatial Light Modulator (SLM).
The structures described below were produced by two-photon polymerisation of the negative photoresist SU-8 2007 (Michrochem, Newton, MA, USA) applying the beam of a Ti:Sapphire laser. The polymerising laser (FemtoRose 100 TUN, R&D Ultrafast Lasers Ltd, Budapest, Hungary) provided 150 fs pulses with 80MHz repetition rate and up to 6.3 nJ pulse energy (500 mW average power) in mode locked operation at 790 nm wavelength. In our system an average power of 3-8 mW was used in each beam for the polymerisation. A 100x magnification, 1.25NA oil immersion objective (Zeiss Achroplan, Carl Zeiss, Germany) was used to focus the beam into the photopolymer.
The treatment of the photopolymer layers was the following: glass microscope cover slips were cleaned with KOH overnight, rinsed with deionized water then treated by oxygen plasma (PDC-002, Harrick Plasma, Ithaca, NY, USA) for 10 minutes. 50 μL of the photoresist was spin-coated (P6700, Specialty Coating Systems, INC., Indianapolis, IN, USA) onto the glass surfaces with a 3000 RPM, 30 s duration spin. The samples were then soft-baked on a homemade hot plate for 2 minutes at 95 Celsius, and transferred into the sample holder. After the completed illumination process the SU8 layer was post-baked for 2 minutes at 95 Celsius on the hot plate and then immersed it the developer solution (mr-Dev 600, Micro Resist Technology GmbH, Berlin, Germany) for 45 minutes with very gentle stirring to remove the non-polymerised resin. Finally the remaining structures were rinsed with ethanol for 2 minutes and dried under a gentle nitrogen stream.
The glass-resin sample was mounted on an X-Y piezo-translator (Physik Istrumente, Germany) of 100x100 μm travel length and nanometer precision displacement perpendicular to the optical axis. The microscope objective was also mounted on a piezo translator (PIFOC, Physik Istrumente, Germany) with 80 μm travel length in the Z-direction, along the optical axis. The piezo stages were mounted on a Zeiss Axiovert 40 inverted microscope equipped with a conventional video camera (VCAMBWPAL, World Precision Instruments, Sarasota, FL, USA). The polymerising laser beam was directed into the objective through the epifluorescent opening at the rear of the microscope. The laser beam access to the sample was controlled by an electrically driven shutter (VS14S2ZM1, Vincent Associates, Rochester, NY, USA) and attenuated by a continuously variable neutral density filter (NT54-535, Edmund Optics GmbH, Karlsruhe, Germany) before it entered the microscope. The reported laser power values were measured at the entrance of the objective. The shutter and the translator movement were controlled by a LabView program running on a PC.
2.2 Polymerised structures
The structures to be polymerised were designed to explore the capabilities of the multi-beam polymerisation method and to demonstrate that it is possible to create extended structures with a single scan that can be manipulated with laser tweezers. First, to test the resolution of the method we generated objects by focusing 5 laser beams into the resin with the arrangement depicted in Fig. 1(a). By scanning the sample parallel to the direction of the beam propagation, i.e. perpendicular to the sample surface, five rods were polymerised starting at the glass surface with length equal to the thickness of the resin layer. With this arrangement we could test the parameter requirements for the rods to remain separated or merge, and for the space between four rods in a square arrangement to be filled.
It has been demonstrated that non-spherical objects trapped by a single-beam laser tweezer assume a position in which their longest axis is parallel to the optical axis and that objects with helical structure rotate when trapped by laser tweezers [6,16,17]. In the second group we fabricated straight columns with cross-shaped base that position themselves in a laser tweezer with their longest axis along the optical axis. In this set the polymerising focal spots formed a cross shape with fixed lateral positions (see Fig. 1(b) and 1(c)) and during polymerisation the focal spots were scanned through the sample along its normal in the Z direction. The third set of structures was prepared with the intention to produce structures that rotate when trapped by a laser tweezer. For this columns with cross shaped base were polymerised but the columns had also a helical twist along their long axis. During the polymerisation of this set the focal spots were also forming a cross shape and translated along the optical path, but during the Z scan the cross formed by the focal spots was rotating as shown in Fig. 1(b) and 1(c).
The 3D structures described above consist of 5 and 9 rods polymerised simultaneously during the same scan along the Z axis. Had we used only a single polymerising focal spot, we should have had to perform Z scan 5 and 9 times to polymerise the entire object and also X and Y scans for the helical columns. The production of the structures is therefore accelerated at least as many times as the number of focal spots. The structures themselves were polymerised one at a time with one single Z scan. We tested the applicability of two types of DOE to perform the beam splitting and to generate the above described structures. The optical arrangement for the DOEs is seen on Fig. 2.
2.3 The transmitting DOE
The first DOE was transmitting-type kinoform. It is a plastic disc the phase shifting 2D structure of which is factory-molded to achieve a specially patterned surface. This piece of optics generated one pre-defined type of beam shaping: it split the one incoming laser beam into nine beams in a symmetric cross shaped pattern (see Fig. 1(c)). The angular deviation between the adjacent beams after the DOE was 0.4 degree for the 790 nm wavelength of the polymerising Ti:sapphire laser. The focusing of the nine beams with the objective resulted in simultaneous polymerisation in the nine focal spots. Because of the fixed beam deviation of the kinoform the desired focal spot distance in the resin was achieved by choosing a 1:10 relay optics (see Fig. 2). The d3 distance of the adjacent focal spots was 1.3±0.1 μm.
With the kinoform straight and helical columns with cross-shaped cross section were produced as shown in Fig. 1(e) and 1(f). The straight and the helical columns were polymerised with the scanning speed ranging from 0.25 μm/s to 25 μm/s along the Z axis and with the power of the original incoming single laser beam from 3.5 mW to 30 mW. There is only the fraction of the total power in each of the nine split beams. It was a characteristics of the kinoform at this wavelength that the total laser power was not split equally among the nine beams, leaving the central one with at least 1.5 times larger power than the others. For the helical columns the orbiting of the focal spots was achieved with the mechanical rotation of the kinoform using a simple stepper motor and a home-made holder. The synchronization of the rotation speed and the travel speed of the focal spots along the optical axis resulted in one complete turn in 20 μm.
2.4 The reflective DOE: Spatial Light Modulator (SLM)
The second DOE was a computer-controlled, active SLM (LC-R 2500, Holoeye Photonics AG, Berlin-Adlershof, Germany). The operation of this SLM is similar to those widely reported for holographic optical tweezers [18,19]. When connected to a PC the SLM acts as a second display of the computer. The computer-generated hologram on the surface of the SLM resulted in the split of the incoming beam into the given number of beams which were reflected off towards the sample. The beams were also separated from the 0th order beam. The hologram was calculated by the application software of the SLM. First a simple 1-bit bitmap image file was made representing the positions of the split beams. Then the application software converted it into an 8-bit grayscale pattern acting as the hologram. This static grayscale pattern appeared on the second display of the computer which was set to be the SLM. After the pattern was transferred, the SLM transformed it to a 0-2π phase shift. The variation of the angular deviation of the beams was achieved through changing the zooming level of the grayscale pattern. Larger deviation requires smaller pattern size. When a hologram pattern that changes in time was needed, an animated GIF file was created in which the subsequent frames generate the consecutive arrangements of the deviated beams. In our experiments the orbiting of the focal spots around the central one was achieved using a GIF file of 4 frames. This means that in each consecutive frame the cross-shape arrangement of the beams rotated by approximately 22.5 degrees around the central one. The dynamic reconfigurability of the holographic pattern on the SLM enabled us to change the angular deviation of the beams and consequently the distance of the focal spots as well as the orbiting of the marginal beams around the central one without any physically moving part. It is the dynamic reconfigurability of the SLM that gives it more potential to use in the parallel polymerisation technique as compared to the static kinoform. We note that the use of the static holograms can be justified in case of some special applications because those are much cheaper, experimentally more easily applicable and the quality of the beam split by them is somewhat superior to that of the SLM.
With the SLM the original beam was split into five beams which were directed into the focusing objective via a 1:1 relay optics (see Fig. 2(b)). With this method all three types of the described structures could be made: the test structures (see Fig. 1(d)), the straight, cross-shaped columns (see Fig. 1(e)) and the helical columns (see Fig. 1(f)). The parameter set for the test structures was the following: 1 μm, 1.3 μm and 1.7 μm focal spots separation (dimension d1), 1 mW, 2 mW and 3 mW laser power for each of the five beams and 1 μm/s, 3 μm/s and 5 μm/s translation speed in Z direction. The straight and the helical columns were prepared with 1 μm, 1.25 μm and 1.5 μm focal spot separation (dimensions d2 and d3), laser power from 2 mW to 8 mW for each beam and 1 μm/s, 2 μm/s and 3 μm/s translation speed. The translation speed and the speed of the orbiting of the focal spots were set such that a full rotation was performed during a translation of about 10–12 μm. Unlike for the kinoform, in the case of the SLM the power is distributed evenly among the laser beams, resulting in more uniform polymerisation.
2.5 Laser tweezer
The optical tweezers system consisted of a Zeiss Axiovert 135 microscope and a Cell Robotics Laser Tweezers 980-1000 unit. The tweezers are formed by an infrared diode laser (SDL 5762 A6 of 994 nm wavelength) focused by the objective lens (Zeiss Plan-Apochromat oil immersion 100X, NA=1.4). The maximal trapping intensity was about 20 mW. In order to be able to manipulate the chosen structures by the laser trap, first we put a drop of about 50 μL aqueous solution of 5% TWEEN 20 over the columns, and then we mechanically broke them off with the help of a micromanipulator using a tapered glass capillary tube. After this the structures, laying on the surface of the cover slip, could easily be trapped by the optical tweezers.
The test structures, presented in Fig. 3, were polymerised using the SLM for beam splitting. The reconfigurability of this device enabled us to set the parameters of the desired test pattern, namely the arrangement of the focal spots and their distances. Second, to study the quality of parallel polymerisation of identical structures equal power level in each split laser beams was necessary which could not be achieved with the kinoform. With the normal operation of the polymerisation using a single laser beam good quality rods with circular cross section can be created when the focal spot is scanned along the optical axis. One of our aims was to study whether it is possible to achieve this geometry with the parallel polymerisation method in a quality comparable to that of the single focus scanning. We also wanted to find the parameter regime where the individually polymerised rods merge or remain separated so that the cross shape that forms the cross section of the structure has good quality.
As clearly seen on Fig. 3(a) and 3(c), merging of the rods can naturally be controlled with the focal spot separation distance (d1), laser intensity and scanning velocity. With the smallest, 1 μm separation only the highest 5 μm/s scanning speed and the lowest 1 mW laser power resulted in separated rods. The other two separation distances (1.3 μm and 1.7 μm) the limit for merged structures is at lower scanning speeds and higher laser powers. The polymerisation using the highest, 3 mW laser power resulted in the merging of the rods at all translation speed values regardless the separation, while the 1 mW power resulted in separated rods in all cases except for the 1 μm separation. The merging at 2 mW power level is a strong dependence of the translation speed and focal spot separation.
The effect of the translation speed is apparent on Fig. 3(d), if one compares it to Fig. 3(b). The focal spot separation in both cases is 1.3 μm, the power level is 2 mW, and the only difference is a factor of 3 in the translation speed. At low translation speed additional fringes are visible among the polymerised rods. These fringes were possibly polymerised by the interference maxima generated by the distinct laser beams focused 1.3 μm from each other.
These maxima can induce polymerisation only at very low scanning speeds which is highly unwanted when regular structures are to be made. The SEM images on Fig. 3 and Fig. 4 show polymerised rods with smooth surfaces, and apparently circular cross sections. From this we conclude that the splitting does not introduce any appreciable aberration in the polymerising laser beams. If the beam had aberration, a case we do not exclude, it does not affect the quality of the polymerised structures beyond applicability.
We also produced straight and helical columns using the SLM (see Fig. 4). The straight columns do not add new information to the results obtained from the test structures. These were made with the same parameters as the helical columns, except the focal spots were not orbiting around the central one, and served as references. In the parameter range we applied in this case the polymerised rods were completely merged in all the columns except when the lowest, 2 mW laser power was used. The dimension of the polymerised columns agreed well with the design parameters: in case of 1.25 μm focal spot separation the centers of the outmost rods were about 2.5 μm from each other (see Fig. 4(a) and 4(b)). Below 3 μm/s scanning speed and above 6 mW laser power the polymerised structures always showed the extra fringes characteristic to the interference maxima of the adjacent focal spots (see Fig. 4(a)). When the helical columns were polymerised with these parameters the extra fringes resulted in unpredictable substructures. The thresholds for merging of the individual rods and for the appearance of the fringes define the parameter range where the method of parallel polymerisation can be successfully applied.
Helical columns with acceptable quality for manipulation in a laser tweezer were polymerised with 3 μm/s scanning speed from 2.9 mW to 5.5 mW laser power and 1.25 μm focal spot separation in about 2-3 seconds as defined by their length and the scanning speed used for their polymerisation. These structures did not display the fringes among the individual rods but clear helicity (see Fig. 4(c)). On the optical microscopic image (see Fig. 4(d) and 4(e)) the spirals seem to be even and clearly visible. The electron micrographs, on the other hand, show that the helical rods are not perfectly continuous but there is a periodicity in the thickness of the outmost rods. We attribute this to the non-continuous rotation of the extreme focal spots around the central one resulting in the polymerisation of a step-like helical rod instead of a continuously spiraling one. Had we chosen smaller than 22.5 degrees rotation angle it would have been possible to achieve smoother helical structures. In spite of this imperfection, the images show that we successfully applied the SLM-split multiple beam arrangement for the polymerisation of extended 3D objects. It can also be seen that the helical structures are not simply the twisted versions of the straight ones (see Fig. 4(b) and 4(c)) although the same parameters were used. The gaps between the rods forming the columns are “filled up” with SU-8 to some extent. In spite of this a successful trapping, alignment and rotation of a helical column by a laser tweezer is presented in Fig. 4(f). As soon as the trap was on, the column aligned itself along the optical axis, therefore only a slightly blurred axial view was visible in the optical microscope. When a trapping beam of about 20 mW was used, a rotation rate of 0.56±0.08 1/s (N=4) was measured.
The nine laser beams split by the kinoform displayed a non-uniform power distribution. The central beam carries about 1.5 times more power than any other of the remaining eight beams. In order to achieve the merging of all nine beams at moderate laser power levels, translation speed as low as 0.25 μm/s had to be used. The non-helical objects polymerised by the nine, kinoform generated beams are shown in Fig. 5(a) and 5(b) as viewed from the direction of the optical axis. From this direction the cross-shape is clearly visible as well as the good reproducibility of the objects. These figures reflect the effect of the laser power. It is clearly seen that the SU-8 rods polymerised along the path of the central five beams always merge. When the 30 mW split beam was used (see Fig. 5(a)), only the 0.75 μm/s and lower scanning speeds yield objects where all the nine beams are merged. Using the 18 mW beam, even the slowest 0.25 μm/s scan speed does not produce completely merged crosses, leaving only the central five to merge. At this low power level some of the outmost focal spots do not even polymerise the SU-8 (see Fig. 5(b)). In the 30 mW case, below 0.75 μm/s scanning speed the previously described fringe-like features appear among the paths of the focal spots. This is, just like in the case of the SLM, probably due to the interference pattern of the adjacent beams. This feature makes the helical columns made with 30 mW and higher unsuitable for the rotation by the laser tweezer.
The helical objects using the kinoform were made by 12 mW, 18 mW and 27 mW total laser power with the scanning speed ranging from 0.2 μm/s to 2 μm/s. The resulted objects chosen for manipulation in a laser tweezer are shown in Fig. 5(c) and 5(d). Viewing from the side the helicity of the structures is clearly visible. The 2-3 μm wide central region, however usually does not show helicity but the arms of the crosses are spiraling around this. Even with the spiraling arms no big variance in the thickness of the helices can be observed. It is due to the fact that at this power level only the central five polymerised rods merged. After visual inspection through the optical microscope the helical columns made with 12 mW laser power and 2 μm/s scan speed were chosen to be tested in the optical trap. Immediately after turning on the trap the columns aligned along the optical axis and started to rotate. Snapshots of the spinning object are shown in Fig. 5(e-i). With about 20 mW laser power the rotation rate was about 0.5 1/s. One of the reasons for this relatively slow rotation is likely that only the central five beams formed the helical column. If the outmost beams also merged the central ones, a helical object of larger variance in thickness would probably rotate with higher rate.
We successfully demonstrated the possibility of photopolymerising micrometer-sized objects with beam shaping methods using diffractive optical elements. By splitting the single original laser beam into several ones we performed simultaneous polymerisation with and without rotating the diffracted beams around the central one during the process. We demonstrated that individual rods of about 500 nm diameter can be produced and identified a parameter range where fringe-free but merging individual rods can be polymerised. We produced helical columns which can be easily trapped and rotated in a laser tweezer. With the extension of the method to more beams, using more sophisticated algorithms and more powerful laser sources we believe that the method of parallel polymerisation will be capable of producing rather complicated structures. The presented method can also be expanded to completely different beam arrangements in the future, achieving merged or separated simultaneously polymerised structures at will. Some of the most obvious limitations of the technique are the available laser power and the speed of the desired reconfiguration of the separated beams during polymerisation.
The kinoform used in the experiments was a generous gift from Dr. Kishan Dholakia, Optical Trapping Group, StAndrews University, UK. We are grateful to Dr. Erzsébet Mihalik, Department of Botany and Botanic Garden, University of Szeged, for the scanning electron micrographs. This work was supported by grants from the Hungarian Scientific Research Fund (T046747) for P. O., from the National Office for Research and Technology (NKFP1/0007/2005) for L. K. and also from EU6 ATOM 3D (contract 508952).
References and links
1. J. H. Park, A. M. Derfus, E. Segal, K. S. Vecchio, S. N. Bhatia, and M. J. Sailor, “Local heating of discrete droplets using magnetic porous silicon-based photonic crystals” J. Am. Chem. Soc. 128, 7938–7946 (2006) [CrossRef] [PubMed]
2. T. A. J. Duke and R. H. Austin, “Microfabricated sieve for the continuous sorting of macromolecules” Phys. Rev. Lett. 80, 1552–1555 (1998) [CrossRef]
3. A. Y. Fu, C. Spence, A. Scherer, F. H. Arnold, and S. R. Quake, “A microfabricated fluorescence-activated cell sorter” Nature Biotech. 17, 1109–1111 (1999) [CrossRef]
4. B. J. Kirby, T. J. Shepodd, and E. F. Hasselbrink Jr, “Voltage-addressable on/off microvalves for high-pressure microchip separations” J. Chromatography A , 979, 147–154 (2002) [CrossRef]
5. R. C. Gauthier, R. N. Tait, H. Mende, and C. Pawlowicz, “Optical selection, manipulation, trapping, and activation of a microgear structure for applications in micro-optical-electromechanical systems” Appl. Opt. 40, 930–937 (2001) [CrossRef]
6. E. Higurashi, H. Ukita, H. Tanaka, and O. Ohguchi, “Optically induced rotation of anisotropic micro-objects fabricated by surface micromachining” Appl. Phys. Lett. 64, 2209–2210 (1994) [CrossRef]
7. J. Voldman, M. L. Gray, and M. A. Schmidt, “Microfabrication in Biology and Medicine” Annu. Rev. Biomed. Eng. 1, 401–425 (1999) [CrossRef]
9. S. Maruo and S. Kawata, “Two-photon-absorbed near-infrared photopolymerisation for three-dimensional microfabrication” J. Microelectromech. Sys. 7, 411–415 (1998) [CrossRef]
11. M. Straub, L. H. Nguyen, A. Fazlic, and M. Gu, “Complex-shaped three-dimensional microstructures and photonic crystals generated in a polysiloxane polymer by two-photon stereolithography” Opt. Mater. 27, 359–364 (2004) [CrossRef]
14. H. B. Sun, T. Kawakami, Y. Xu, J. Y. Ye, S. Matuso, H. Misawa, M. Miwa, and R. Kaneko, “Real three-dimensional microstructures fabricated by photopolymerisation of resins through two-photon absorption” Opt. Lett. 25, 1110–1112 (2000) [CrossRef]
15. S. Maruo, K. Ikuta, and H. Korogi, “Force-Controllable, Optically Driven Micromachines Fabricated by Single-Step Two-Photon Microstereolithography” J. Microelectromech. Sys. 12, 533–539 (2003) [CrossRef]
16. P. Galajda and P. Ormos, “Complex micromachines produced and driven by light” Appl. Phys. Lett. 78, 249–151 (2001) [CrossRef]
17. P. Galajda and P. Ormos, “Rotation of microscopic propellers in laser tweezers”, J. Opt. B. 4, S78–S81 (2002) [CrossRef]
18. J. Leach, G. Sinclair, P. Jordan, J. Courtial, and M. J. Padgett, “3D manipulation of particles into crystal structures using holographic optical tweezers” Opt. Express , 12, 220–226 (2004) [CrossRef] [PubMed]
19. P. J. Rodrigo, L. Gammelgaard, P. Bøggild, I. R. Perch-Nielsen, and J. Glückstad, “Actuation of microfabricated tools using multiple GPC-based counterpropagating-beam traps” Opt. Express 13, 6899–6904 (2005) [CrossRef] [PubMed]