Available online at www.sciencedirect.com ScienceDirect Procedia CIRP 34 (2015 ) 212 216 9th International Conference on Axiomatic Design ICADD 2015 Momenta ary Local Laser Heating in Thermal Roll-to-Plate Nanoimprinting; Node Decoupling with Axiomat tic Design Masayuki Nakao a*, Ken Takahashi a, Keisuke NagatoN a, Kenji Iino b a Department of Engineering Synsis, School of Engineering, The University of Tokyo, Hongo7-3-1, Bunkyo-ku, Tokyo, Japan b SYDROSE LP, 475N. 1st St., San Jose, CA95112, USA * Corresponding author. Tel.: +81-3-5841-6360; fax: +81-3-5800-6997. E-mail address: nakao@hnl.t.u-tokyo.ac.jp Abstract The independencee axiom of axiomatic design to decouple interference among functional requirements might be effective in designing new processes. This paper reports an application of this concept to our design of roll-to-platee rmal nanoimprinting using laser heating. The authors, instead of heating an entire roller, casted a laser beamm through a glass roller to a thin metal film on mold to produce p a momentary local heat source. The mold immediately cooled down after laser heating patterned resin did not reflow back to flat shapes; we succeeded in decoupling node of heating d cooling. 2015 The Authors. Published by Elsevier B.V. This is an open access article under CC BY-NC-ND license Pee (http://creativecommons.org/licenses/by-nc-nd/4.0/). er-review under responsibility of organizing committee of 9th International Conferencee on Axiomaticc Design. Peer-review under responsibility of organizing committee of 9th International Conference on Axiomatic Design Keywords: Decouple; Laser; Nanoimprinting 1. Introduction 1.11 Axiomatic design application in product processs design After Suh introduced axiomatic design in 1990 [1], it has been applied in a number of design analyses of products processes. The method is to apply independence axiom to select Design Parameter (DP) vector to remove interferences among Functional Requirements (FR), d n to apply information axiom to optimize DP vector to maximize sum of probability off success forr each functional requirement. At same time, DP vector has to satisfy Constraints (C) as a boundary condition off DPs, e. g., cost, life, lead time, rigidity, or system efficiency [2]. Figure 1 shows relation between FR DP vectors with Design Matrix (DM). The DM of a coupled design is non-diagonal with non-zero non-diagonal components; however, decoupling, i.e., removing interferences with some new DPs will make all non-diagonal components of matrix zero, thus matrix turns diagonal. The Cs are expressed by linear relations of DP vector. Fig. 1. Decouplingg in Axiomatic Design. 2212-8271 2015 The Authors. Published by Elsevier B.V. This is an open access article under CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of organizing committee of 9th International Conference on Axiomatic Design doi:10.1016/j.procir.2015.07.001
Masayuki Nakao et al. / Procedia CIRP 34 ( 2015 ) 212 216 213 Fig. 2. Categories of nanoimprinting. Note that products processes have a large number of functional requirements [3]. When number of FRs is 20, for example, we have to check interferencesi for 20x20=400 cases, such a strict analysis burdens designer. From our experience, keeping number of FRs to 5 or less, is quite effective to keep designer concentrated on important FRs. For this paper, we excluded FRs with success rates of 1 or 0, meaning those are always or never accomplished, set n=3. 1.2 Designing nanoimprinting process Chou reported nanoimprinting in 1995 [4]. The process produces geometric patterns of about 30nm width on silicon replicates inverse pattern on resin by pressing. The resin is n used for masks. A number off researches have been since n advancing process. Figure 2 shows 4 groups of se follow-up studies. First, y are divided into 2 groups based on wher resin is hardened rmally by cooling rmoplastic resin after heating, softening pressing it against a patterned mold, or optically by putting photo-curing resin, in form of liquid or powder, under light after is has been pressed against a metal mold. The second categorization distinguishes between stamping rolling metal mold. Altoger 2x2 categorizations define 4 types. The optically hardening process proceeds under room temperature, thus it eliminates need for rmal design; however, photo-curing resin r costs about 10 times higher price of rmoplastic resin. Rolling is a widely used process in metal rolling or paper printing. The process control is deming with short time localized area when roller sheet come in contact; however, rolling has advantage in its about 10 times faster process time than stamping. The process wee developed report in this paper is in lower left-h corner of Figure 2, i.e., rolling rmal nanoimprinting. Among four, this processs is most likely to have lowest production cost is suited for production of out-coupling film for organic electroluminescence displays or anti-reflection filmm for liquid crystal television screens. 2. Axiomatic design of a laser assisted roll-to-plate to accomplish decoupling Figure 3 (a) shows a conventional roll-to-plate rmal nanoimprinting. For this conventional system of replicating nano-structures on roller mold to resin sheet, processes for heating, pressing, cooling, mold releasing have to be rmal nanoimprinting 2.1 Records of ourr thinking process implemented near contact area of roller sheet. Past methods of using a metal rollerr with a heater for raising resin sheet temperature succeeded in softening resin replicating mold m patterns; however, t metal roller remained hot. Thee resin sheet n is released from mold before cooling down to a temperature of e glass transition point (Tg), surface tension in still s soft patterns would reflow resin r back to flat shapes. Axiomatic design explains this phenomenon as interference between heating cooling processes. For purposee of better understing reflow, r we put 4 types of resin sheet through stamping rmal nanoimprinting processes. Figure 4 shows ratios of e replicated form heights to that of metal moldd (replication ratios). The mold has an array of hemispheres of f 15 m in height 30 m in diameter. The larger feature is more difficultt to be replicated because it needs larger l heatingg value. As upper figure shows, each type of resin flows better with higher mold temperature higher mold pressure. The lower figure shows that mold temperaturee upon mold release r caused Fig. 3.. Thinking process of decoupling in rolled rmall nanoimprinting. Fig. 4.. Replication ratio (Rr) in stamping rmal nanoimprinting with four types of resin.
214 Masayuki Nakao et al. / Procedia CIRP 34 ( 2015 ) 212 216 Fig. 5. Replication ratio (Rr) in rolling rmall nanoimprinting.. Fig. 7. Process design d for laser heated rolling rmal nanoimprinting. Fig. 6. Ratio of replication in laser heated rolling rmal nanoimprinting. reflow when it was above Tg, resulting in lower replication ratio. This experiment demonstrated that above interference takes place no matter what resin was. Figure 3 (b) shows improved system Ogino developed [5]. The mold was given a belt shape, instead of a roller shape, to have a longer cooling area after heating area. Thiss belt- turned out to be expensive did not meett cost constraint. Figure 3 (c) shows our improving configuration with a pre- shaped thick nickel plated mold for highh tension, however, heating roller to lessen amount of heatt input at roller contact point; so resin sheet can quickly go into its cooling process. This trial, however, resulted in necking with elongation in rolling direction shrinking in lateral by sheet tension of 0.8MPa. We observed this necking when pre-heating temperature was around Tg, 110 C for polymethyl methacrylate (PMMA). Yet anor a attempt we made to cool mold quickly placed a heat insulating polyimide sheet of 250 m thick between a thin nickel film on roller surface roller body inside. For better cooling of resin sheet, we wrapped resin sheet halfway around cooling roller. As Figure 5 shows, replication rate reached as high as 83% with higher mold temperature higher mold pressure; however, when mold temperature exceeded a certainn point, reflow caused a drop in replication ratio (Rr). For this case, replicated ridges had dimples of size about 1μm. These dimples from metal mold were a sign of a 100% Rr when resin contacted metal mold. The lowered Rr after entire process was w a sign that reflow caused ridges to reduce ir heights. The heating cooling processess were still interfering. Figure 3 (d) shows s our last attempt which is subject of this paper. We produced p a glass roller to pass a laser beam to cast on contact area where laser is absorbed to metal mold. The laser wavelength was 1.07μm, absorption rate was w 0% for glass PMMA, 28% for nickel, 50% with a 0.1μm diamond-like carbon (DLC)( layer on nickel, 90% % with a 3μmm DLC layer on it. Figure 6 (a) shows results using 0.23 m height mold with 3 m DLC 4W YAG laser [6]. Thee diagram on left shows replication area with intermittent laser irradiation. The area was larger withh greater laser power density longer irradiation time. The replication area, as t diagram on right shows, accomplished a Rr of 100% free of reflow in center. Figure 6 (b) shows results using 15 m height Ni mold replicate a higher hemisphere. Deformation speed was 10-40 m/msec at 10 MPa. Irradiation time of 3.3msec could make a Rr of 100% without reflow. Figure 7 (a) shows outline of experiment setup. Figure 7 (b) shows how we set process parameters. First heated depthh ( ) had to be larger than metal mold feature height (hm)( for proper replication. Solving 1 dimensional heat transfer problem for finding, we have [12 x (resin rmal t diffusivity : 1.0 x 10-7 m 2 /s) x (irradiation time: 3msec)] ^0..5=60μm. Thee maximum hm for our molds was 15μm, less than 60 m; thus replication was feasible. Also, too prevent reflow, we will keep roll contact length (2. Lr) longer than sum of heating length (Lh) cooling lengths l (Lc). The Lh, for example, was
Masayuki Nakao et al. / Procedia CIRP 34 ( 2015 ) 212 216 215 Constraints were; C 1 : Increase replication speed, C 2 : Decrease replication const. Rough values for se constraints were C 1 : 1m/min= =16mm/s or faster, C 2 : 10$/m 2 or lower. Thesee constraints were both affected by DP 1, DP 2 DP 3 ; replication speed C 1 wass strongly affected with irradiation time DPP 1 coolingg DP 3, production cost C 2 strongly with fiber laser price for DP 1. 3. Discussion This section reports results of applying decoupling method of processs design using same fiber f laser to or processes, that is, laser welding laser cladding. The following results are from interviews we conducted c to designers to verifyy decoupling of interferences. Fig. 8.. Design equation of laser heating in rolling rmal nanoimprinting. laser spot diameter of 0.5mmm with a power densityy of 23kW/cm 2 an irradiation time of 3msec. A temperature transition simulation gave a cooling time of 5msec for this example. Multiplying this 5msec to feed f velocityy of 16mm/s resulted in Lc=0.08mm. The 2. Lc, on or h, for conditions of 10MPa mold pressure, 2GPa Young s modulus of resin, 75 m sheet thickness, 15 m mold height, 100mm roller diameter, came out to be 1.2mm as shown in Figure 7 (b). This length of 1.2mm prevented reflow because this 1.2mmm is larger than sum of Lh=0.5mm Lc=0.08mm. A galvano mirror scans 100W laser in lateral direction. To replicate seamlessly by scanning spot inn range of 100mm, feed velocity (vf) should become slower to 0. 83 mm/s under irradiation time (ti) of 3msec. Iff we can shorten ti to 0.15msec with a higher power laser, we can get a larger vf of 16mm/s a shorter heated depth ( ) of 13 m, requiring a DLC layer for higher absorption. Consequently, vf, ti, have interferes; so we define vf as a constraint in next section. 3.1 Decoupling interference withh laser welding Figure 9 showss design equations for (a) conventional spot welding (b) new laser welding. For realizing functional requirements FR 1 : Make plates contact each or, FR 2 : Melt plates around contact spot, FR 3 : Increase welding area, conventional spot s welding (a) selected design parameters DP 1 : Clamping tool, DP 2 : Electric resistance welding, DP 3 : Pitch of welding spots. Interferences for this case are X21: 2 Requires clamping tool to t pass electricity, X 23 : Narrower pitch of spots disables welding due to current leakage to adjacent welded spots, X 31 : Places that cannot be clamped by tools cannot be welded. Higher rigidity with a narrower spot pitch would cause electricity leakage, i.e., addedd spots electricity leakage interferes.. Laser welding in i case of (b) has same set of functional requirements to realize. Changing design parameters to DP 1 : Crimping or temporary welding DP 2 : Plug welding with laser leads too a design equation that shows s successful decoupling of interference. The clamping tool for (a) occupies areas on both sides of plates; however, tools for laser beaming only sit on one side. Heat input with laser is also 2.2 Decoupling interference with laser heating Figure 8 shows design equations for conventional rolling rmal nanoimprinting in Figure 3 (a) laser assisted heating in (d). We set functionall requirementss to; FR 1 : Heat soften resin sheet, FR 2 : Inversely replicate nano-features to sheet, FR 3 : Cool release sheet from mold. Figure 8 (a) shows design parameterss of conventional one; DP 1 : Heating mold roller, DP 2 : Pair of pressing rollers, DP 3 : Water cooling roller. Interferences were, as we explained above, X 21 : Necking, X 23 : Reflow, X 31 : Heating even after replication, se interferences make DM non-diagonal. For laser heated casee of Figure 8 (b), FRs were same, but with a different DP 1 : Laser absorbing mold, FRs weree decoupled to t make DM diagonal. We succeeded in rolling rmall nanoimprinting with a 100% replication ratio. Fig. 9. Design equation of laser welding.
216 Masayuki Nakao et al. / Procedia CIRP 34 ( 2015 ) 212 216 interferences of, respectively, heating cooling, added spots electricity leakage, seat thickness fixing strength. We showed for se cases that momentary local laser casting succeeded s in n decoupling interferences innovatively. References Fig. 10. Design equation of laser cladding. narrowly localized spot weld pitch is small. The constraints for this case are C 1 : Give high rigidity, C 2 : Lower welding cost. A smaller spot pitch withh DP 3 strongly affects rigidity C 1 fiber laser price for DP 2, cost C 2. [1] Suh, N. P., 1990, The Principles of Design, Oxford University Press. [2] Brown, C. A., 2014, Axiomatic Design of Manufacturing Processes Considering Coupling, C Proceedings of Eighth International Conference on Axiomatic A Design, 149-153. [3] Thompson, M.. K., 2013, Improving requirements process in Axiomatic Design Theory, Annals of CIRP 62(1): 115-118. [4] Chou, S. Y., Krauss, P. R., Renstrom, P. J., 1995, Imprint of sub-25 nm vias trenches in polymers, Applied Physics Letters, 67-21: 3114-3116. [5] Ogino, M., Hasegawa, M., Sakane, K., Nagai, S., S Miyauchi, A.., 2008, Fabrication of 200-nm 2 Dot Pattern on 15-m-Long Polymer Sheet Using Sheet Nanoimprint Method, Japanese Journal of o Applied Physics, 52: 035201. [6] Nagato, K., Takahashi, K., Sato, T., Choi, J., Hamaguchi, T., Nakao, M., 2014, Laser assisted replication of large area nanostructures, Journal of Materials Processing Technology, 214: 2444-2449. 3.2 Decoupling interference with laser cladding Figure 10 shows that conventional valve seats for an aluminum engine block use (a) press inserted sintered rings. For a new design, y are replaced with (b) laser cladding. Thinner this seat is, intake to engine makes a low angle flow that generates better swirling to increase combustion efficiency of engines. The functional requirements are; ; FR 1 : Harden valve seat, FR 2 : Fix valve seat, s FR 3 : Thin valve seat. For case (a), design parameters are; ; DP 1 : Iron based sintered ring, DP 2 : Shrinkagee fitting d DP 3 : Milling valve seat. The DM shows interferences of X 21 : Strong pressing required for compensating difference of rmal expansion, X 23 : Strong fixing requires certain height while a thinner seat is desired. For this case, seat thickness fixing strength interfere. The new design in this case selects laserr cladding. Case (b) has same set of FRs, however, a different set of design parameters; DP 1 : Hard copper powder,, DP 2 : Laser cladding. The change realizes a strong surface yet with a thin seat combustion efficiency goes up by several percent. The DM is diagonal showing a successful decoupling of interference. The constraints are C 1 : Increase combustion efficiency, C 2 : Reduce hardening cost. The swirling of C 1 is affect by seat thickness DP 3, cost C 2 by e fiber laser price DP 2. 4. Conclusion In designing a new process, applying independence axiom of axiomatic design effectively led to t a design solution with decoupled functional requirements. This paper reported our analysis with laser heated roll-to-plate rmal imprinting, laser welding, laser cladding. For each case, re existed