Jtam-A4.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 50, 4, pp. 1025-1036, Warsaw 2012 50th Anniversary of JTAM MECHANICAL MODELLING AND LIFE CYCLE OPTIMISATION OF SCREEN PRINTING Eszter Horvath, Gabor Harsanyi Budapest University of Technology and Economics, Department of Electronics Technology, Budapest, Hungary e-mail: horvathe@ett.bme.hu; harsanyi@ett.bme.hu Gabor Henap Budapest University of Technology and Economics, Department of Applied Mechanics, Budapest, Hungary e-mail: henapg@mm.bme.hu Adam Torok Budapest University of Technology and Economics, Department of Transport Economics, Budapest, Hungary e-mail: atorok@kgazd.bme.hu The application of thick film pastes and adhesives is a screen printing method in the mass production of thick films and low temperature co-fired ceramic circuits. During this process, the paste is printedbya rubber squeegee onto the surfaceof the substrate througha stainless steelmetal screenmaskedbyphotolithographic emulsion.We consider the off-contact screen printingmethod in this paper, because it is now the standard printingmethod in themicro- electronics industry. In our research a Finite ElementModel (FEM)was created in ANSYS Multiphysics software to investigate screen deformation and to reduce stress in the screen in order to extend its life cycle. An individual deformation measuring setup was designed to validate the FEM model of the screen. By modification of geometric parameters of the squeegee, the maximal and the average stress in the screen can be reduced. Furthermore, tension of the screen decreases in its life cycle, which results in worse printing quality. The compensation of this reducing tension and themodified shape of the squeegee are described in this paper.Using this approach, the life cycle of the screen couldbe extendedbydecreased mechanical stress and optimised off-contact. Key words: screen printing, FEM, thick film, optimisation, life cycle 1. Introduction The screenprinting is themostwidespreadand commonadditive layer deposition andpatterning method in thick film technology. A thick film circuit usually contains conductive lines, resistive and dielectric layers. Screen printing technology provides themost cost effective facility of applying and patterning different layers for hybrid electronics industry. Due to its simple technology and relative cheapness, it is still widely used in the mass assembly of recent electronic circuits. The screen-printing is ideal as a manufacturing approach for microfluidic elements also used in the field of clinical, environmental or industrial analysis (Albareda-Sirvent et al., 2000), in sensors (Viricelle, 2006) and in solar cells (Krebs, 2007). The paste printing is carried out by a screen printermachine. The screen is strained onto an aluminium frame and the thick film paste is pressed onto the substrate through the screen not covered by emulsion by a printing squeegee. The squeegee has a constant speed and pushes the screen with contact force. The material of the screen can be a stainless steel or polymer. The design of the printed layer is realized with a negative emulsion mask on the screens. There are two main techniques of the screen printing: 1026 E. Horvath et al. • off-contact, where the screen is warped with a given tension above the substrate; • contact, where the screen is in full contact with the substrate. The contact screen-printing is less advantageous in general, because due to the lift off of the screen, damage of the high resolution pattern is often detected. In the case of the off-contact screen-printing, somepaste is applied on the top of the screen in the front of the polymer squeegee. While the squeegee is moving forward, it deforms the screen downwards until it comes into contact with the substrate beneath. The paste is pushed along in front of the squeegee and pushed through the screen not covered by the emulsion pattern onto the substrate. The screen and substrate separate behind the squeegee. The off-contact screen printing process is demonstrated in Fig. 1. Fig. 1. The squeegee pushes the paste on the screen and presses it through the openings Since the ’60s several experiments and models of the printing process have been evaluated. The optimisation of screen printing was mainly achieved by experimental evaluation without the advantage of numericalmodels. That empirical optimisationmethodwas described byKobs and Voigt in 1970, they appointed more than 50 variables and combined the most important ones almost 300 different ways and compared the effects of them. Those investigations offer a enormous empirical database but general rules for screen-printing cannot be created from this without models. Miller (1969) investigated the amount of paste printed on the substrate in function of paste rheology, mesh size and line width. Others examined the influence of squeegee angle and characteristic on the thickness of the deposited paste (Benson, 1969) and the effect of the screen on fine scale printed patterns (Bacher, 1986). The first efforts to achieve a theoretical description of the screen printing process weremade byRiemermore than 20 years ago (1988, 1989). Hismathematical models of the screen-printing process based fundamentally on a Newtonian viscous fluid scraping model (Taylor, 1962). This model was extended by others (Riedler, 1983; Jeong and Kim, 1985), although they did not take into account the flexibility of the screen, which is an essential feature of the process. None of these models deals with the effect of geometry in the screen printing process. The repetitive characteristic of the printing process requires taking into consideration the effect of the cyclic load. In this work, a mechanical model is presented with similar geometry to the off-contact screen printing process. In this model, the mechanical behaviour of the screen is in the focus instead of the paste deposition phase. Furthermore, ourmodel effectively concerns the geometry of the knife. 2. Experimental 2.1. Material parameters of screen Thefirst step of themodel constitution is to define the geometries and obtain themechanical parameters of the screen. The geometric features and the initial strain - which warps it onto an aluminium frame – are realised by themanufacturing process. Mechanical modelling and life cycle optimisation... 1027 In order to decrease screen tension deviation during the printing, the screen is tightened onto the aluminium framewith the thread orientation of 45◦ to the printing direction. Therefore, the load distribution is more homogeneous between the threads. The elastic (Young)modulus of the screen is determined using themodifiedVoigt expression Ec = ηEfVf +Em(1−Vf) (2.1) where Em = 690MPa is the elastic modulus of the emulsion, Ef = 193GPa is the elastic modulus of the stainless steel, Vf is the volume fraction of the stainless steel and η is the Krenchel efficiency factor (Cox, 1952; Krenchel, 1964). In the case of Θ1 =Θ2 =45 ◦, the thread orientation in the frame is η= 1 2 cos4Θ1+ 1 2 cos4Θ2 = 1 4 (2.2) The Poisson ratio can be expressed as νxy = νfVf +νmVm (2.3) where νf is the Poisson ratio of the stainless steel (0.28) and νm is the Poisson ratio of the emulsion (0.43) (Irgens, 2008). In our study, SD75/36 stainless steel screen was utilised with the mesh number of 230 and open area of 46%. The schematic view of the screen cross section is shown in Fig. 2, where d is the diameter of the thread, α is the bending angle, x is the element length of the thread x= d sinα (2.4) Fig. 2. A sketch of the SD75/36 screen cross section with the main parameters Using Eq. (2.4), the volume fraction of the stainless steel can be calculated Vf =2 d2π 4 l l l+d d sinα =0.27 (2.5) Substituting Eq (2.5) into Eq. (2.1), the elastic modulus of the screen turned out to be 13GPa. The sizes of the screen are 298mm inwidth, 328mm in length and the thickness of it was 72µm. These parameters were utilised in the finite element model. 2.2. Measuring the friction force between the screen and the squeegee The paste we have applied in our experiment was PC 3000 conductive adhesive paste from Heraeus. In the process of screen printing, the friction force between the screen and the squeegee playsan important role.While the squeegeepasses the screendueto the friction force theposition of the mask shifts. The individual friction force, measuring setup is shown in Fig. 3. By thismeasurement, the relationship between the friction force Ff and theprinting speed v and squeegee force Fs was estimated. Every thick filmpaste is viscous and has a non-Newtonian 1028 E. Horvath et al. Fig. 3. Measurement setup for determining the friction force between the squeegee rheology suitable for screen printing. The shear stress τ for this kind of fluids can be described by the Ostwald deWaele relationship τ =K (dv dx )n (2.6) where K is the flow consistency coefficient [Pa·sn), ∂v/∂x is the shear rate or the velocity gradient perpendicular to the plane of shear [s−1], and n is the flow behaviour index [–] (Scott Blair et al., 1939). Thick filmpaste is a shear-thinning fluid, thus n is positive, but lower than 1. In addition, the elongation of the screen – which is greater if the off-contact is greater – results in an image shift as well (Hohl, 1997). The effect of these lateral shifts demonstrated in Fig. 4 has to be taken into account. Fig. 4. Deformed paste deposition as a result of screen elongation The image shift was examined, where the screen tension was in the region of 2-3.3N/mm, the off-contact was 0.9-1.5mm, and the applied friction force was based on the measurement. The reduction of screen tension can affect the quality of the printing in other aspects. The deflection force of the screen is decreasing, so the separation of the substrate and the screen can not start right after the squeegee passes on the screen. This off-contact distance has to be modified in function of screen tension to keep the screen from sticking to the substrate during printing because adhesion causes many separation problems that damage the quality of the printed film. 2.3. Principles of the mechanical model Equations formechanical simulations are based onHook’s law for isotropicmaterials (Bathe, 1996) σ= E 1+ν ( ε+ ν 1−2ν εII ) (2.7) where σ, ε, εI, I, E, ν parameters are the stress tensor, the strain tensor, the first scalar invariant of it, the identity matrix, Young’s modulus and Poisson’s ratio, respectively. Due to Mechanical modelling and life cycle optimisation... 1029 the spatial problem of a large thin plate, the general stiffnessmatrix (used by the finite element software) can be reduced to a simpler form, because the material is symmetric in the x and y direction. The elastic stiffness matrix – according to the circumstances of the given problem – has the following form D= E (1+ν)(1−2ν)     1−ν ν 0 ν 1−ν 0 0 0 1−2ν 2     (2.8) where E is the Youngmodulus, and ν is the Poisson ratio. As a consequence of the squeegee load, the screen bends and gets large displacement, so geometric nonlinearity has to be taken into account. Green-Lagrangian strain components Eij can be expressed as Eij = 1 2 (ui xj + uj xi + uk xi uk xj ) (2.9) where u is the displacement vector. In thenumerical computation,Cauchy stresswas calculated. 2.4. Constructing and verifying the Finite Element Model For the screen model, the shell element was selected because it handles nonlinear geometry in large strain/deflection and in stress stiffening. These two types of geometric nonlinearities are playing a significant role in the modelling of mechanical description of the screen. SHELL93 element has 6 DOF (degrees of freedom) at each node. The first step in thrmodelling of the screen was to determine the initial strain in the screen without an additional load (without the squeegee load), due to the fact it is tighten on the frame. These loads were applied on two perpendicular edges of the screen, while the two others were fixed in the direction of the load acting on the opposite edge. The schematic view of the sizes (x=328mm, y=298mm), edge loads (σx,σy) and constraints only for pretension can be seen in Fig. 5. Fig. 5. Layout of the pre-stress condition In the second model – in which the bending of the screen (resulted by the squeegee load) is calculated in function of the load value and position – the screen tightness is given by a displacement constraint calculated in the model before. As boundary conditions, fixed screen edges (inall directions thedisplacementsandrotations arezero)with thecalculateddisplacement conditions were given. Taking into consideration that the printing process is slow enough, it can be handled stationary in each moment while the screen is in force equilibrium. The width of 1030 E. Horvath et al. the squeegee was 180mm. Rectangular elements were used and themesh density was gradually increasing only towards the load area of the squeegee for faster convergence. The aim of this simulation was to examine how themodel describes the real process. In order to compare the FEM calculation to the real situation, a measurement set-up was designed and realised (Fig. 6). Fig. 6. A sketch of the equipment for measuring deformation of the screen The screen was loaded at 11 different positions, where the distances from the centre are from 0mm to 100mmwith 10mm step size represented in Fig. 7. The load was 40-80N in 10N step sizes. These parameters give 55 different measurement points. At one measurement point, 5 measurements were recorded. Fig. 7. Squeegee line locations on the screen during the measurement In themodel of screen-printing, the displacement of the screen at the load place in the z di- rection was maximised according to the industrial standards of distance (about 1mm) between the screen and the substrate (off-contact) (White et al., 2006). The original construction of the screen printing is shown on Fig. 8. As boundary conditions, ?xed screen edges (motion is zero in all possible directions) were set with the displacement load in the x and y direction, which corresponds to screen tension. In order to simplify the model and to reduce the run time, the contact problem was avoided by using a prescribed displacement load of the screen at the load line. This could be utilized because the screen takes the shape of the squeegee (Fig. 9a). A finer mesh was created in that area where the squeegee acts, and a coarser mesh for the rest of the model (Fig. 9). The finer-meshed area ensures the accuracy, and the coarser-meshed area provides a faster run time. Mechanical modelling and life cycle optimisation... 1031 Fig. 8. The original construction of screen printing in the FEMmodel with constraints Fig. 9. Meshed screen in FEM; the mesh is densified at the load area of the squeegee 3. Results and discussion 3.1. Modelling of stress distribution in the screen In the first model of the screen, the initial strain – occurred by the stretching on the frame – was determined. For the initial stress of σx = σy = 2.65N/mm, displacements in the x and y directions were −0.6209mm and 0.5641mm, respectively. In the second model, the screen tightness was given by this displacement constraint calculated in the model before. In this model, the screen was loaded at 11 different positions and 5 different loads according to the measurements. Compared to the simulation results and measurements the screen deformation can be seen on Fig. 10 for 55 different conditions. In the model of screen printing – where the maximum displacement of the screen in the z direction was 1mm – the stress was concentrated at the ends of the load area (Fig. 11). The maximum stress in the screen (105MPa) appeared around the load edge, while the average stress in the screenwas only about 38MPa. This phenomenon occurred due to the point ending of the squeegee shape, so the corners of the squeegee generate stress concentration in the screen. The surface quality of the used (5000 cycles) screen was examined by an optical microscope to detect the damage resulted by these high stress peaks (Fig. 12). The investigation shows that the screen area, where the edges of the squeegee passed the filaments, are abraded, however, the middle part is intact. 1032 E. Horvath et al. Fig. 10. Measured and simulated bending of the screen at the load line Fig. 11. Mises equivalent stress distribution in the screen along the line to which the load is applied Fig. 12. Themiddle part of the screen is intact, but where the squeegee edges act, the filaments are abraded In order to reduce this relative high stress peak in the material, the shape of the squeegee was modified. The two parameters of the round off are R and f, the radius of the circle and the width of the rounded squeegee segment, respectively (Fig. 13). In the finite element model related to the squeegee round off (further on: fillet), the displa- cement was prescribed only in the line segment, where the squeegee contacts the screen. This could be utilized here as well because the screen takes the shape of the squeegee to the point P, where the deformed screen profile is tangent to the fillet curve. Here, the condition α = β is satisfied (Fig. 14). Mechanical modelling and life cycle optimisation... 1033 Fig. 13. A scheme of the squeegee round off with the radius of the circle and the width of the rounded squeegee segment Fig. 14. The prescribed displacement condition of the screen at the line segment where the squeegee contacts the screen The optimal radius can be obtained from the extrema (in this case, the minimum) of the σ(R) stress-radius function (Fig. 15). Fig. 15. Process flow of the squeegee – round off optimisation As a result of rounding squeegee ends for f = 40mm with the optimal R of 1900mm, the maximum stress in the screen reduced to halves (Fig. 16). The value of f should be as high as the screen mask allows, because a larger rounded area results in lower stress concentration. 3.2. Effect of the friction force and screen tension on the quality of screen printing The model was extended with the friction force (see Section 2.2) in order to determine the shift of the pattern of the screen. Table 1 summarises the friction force between the screen and the squeegee in function of the squeegee force and speed. Evaluating the results fromTable 1, it can be determined by regression of the least squares method that n is between 0.2 and 0.4 in Eq. (2.6) for this type of adhesive paste. Even if the applied friction force was 8.4N, the off-contact was 1.5mm and the tension of the screen was reduced to only 2N/mm, and the resulting shift was less than 2.7µm. 1034 E. Horvath et al. Fig. 16. Mises equivalent stress distribution in the screen along the line of load action Table 1.The friction force between the screen and the squeegee∗ Squeegee Speed [mm/s] pressure [N 20 40 60 80 100 120 140 160 10 2.2 3.4 4 4.6 4.6 5.2 5.6 5.6 20 2.6 4 4.4 5.2 5.4 5.6 6 6.2 30 3.4 4 4.6 5.2 5.6 5.8 6.8 6.6 40 3.6 4.2 5 5.4 5.8 6 6.8 7 50 3.6 4.6 5 5.4 5.8 6.2 7 7 60 3.8 4.6 5.2 6 5.8 6.4 7.2 7.4 70 4.6 5 5.6 6.2 6.2 6.6 7.4 8 80 4.8 5.6 5.8 6.6 6.6 6.9 8.4 8.4 ∗at different squeegee forces and speeds The image deformation arises from the elongation of the screen that is less than 0.5µm in the printing area of the screen in the case of 1.5mm off-contact. Obviously, it is lower if the off-contact is lower. Accordingly, the deposition shift is negligible under 1.5mmoff-contact, and if the friction force is in this region. However, if there is not enough paste on the screen, the friction force can bemultiplied, so the shift can reach 10µm. On the other hand, the quality of printing ismaintainable if the reduction of screen tension is compensated. The screen tension is reduced in the screen caused by repetitive printing – which can be handled as a cyclic mechanical load – the elongation of the screen then increases. As the tension is decreasing, the deflection force of the screen is also decreasing, so the screen usually adheres to the substrate and the separation can not start right after the squeegee passes on the screen. The deflection force is maintainable if the off-contact distance is modified. In our study, the initial screen tension was 3N/mm and the off-contact was the industrial standard (1mm), which resulted in the paper printing quality. In order to avoid adhering, the off-contact has to be increased according to Fig. 17. As the squeegee forcehasnotbeenchanged, thepaste isbeingprintedwith the samepressure, and due to the modified off-contact, the elastic force resulting from screen deflection and the paste adhesion has the same force condition as at the initial screen tension and off-contact. 4. Conclusion Afinite element model was created and verified to describe the stress distribution in the screen due to squeegee load. The boundary displacement condition was determined in the first step by a preliminary model. Using these results, a model was constituted to simulate the bending Mechanical modelling and life cycle optimisation... 1035 Fig. 17. Off-contact compensation in function of screen tension of the screen due to different loads acting at different positions. A measurement set-up was designed and realised to verify the model. Comparing the measured and simulated results, it can be clearly concluded that themodel gives good approximation of the bending values. In the model of screen printing – where the maximum displacement of the screen in the z direction was 1mm– the stress was determined. Themaxima appeared at the ends of the load area. The geometric parameters of the squeegee were modified to reduce the stress in the screen in order to extend its life cycle. By this, the maximum and the average stress in the screen could be reduced. Furthermore, the decreasing screen tensionwas compensated bymodifying the value of the off-contact, which resulted in a sustainable screen-printing quality. Therefore, the life cycle of the screen could be extended by decreased mechanical stress and increased off-contact. Acknowledgement This work has been connected to the scientific program of the ”Development of quality-oriented and harmonized R+D+I strategy and functional model at BME” project. This project is supported by the SzechenyiDevelopmentPlan (Project ID: TÁMOP-4.2.1/B-09/1/KMR-2010-0002).This paper has been supported by JanosBOLYAI fellowship ofHAS (HungarianAcademyof Science). The authorswould like to thank Hatvan – Robert Bosch Elektronika Ltd. for funding the project. References 1. Albareda-Sirvent M., Merkoçi A., Alegret S., 2000, Configurations used in the design of screen-printed enzymatic biosensors. A review, Sensors and Actuators B: Chemical, 69, 1/2, 153-163 2. Bacher R.J., 1986, High resolution thick film printing, Proceedings of the International Sympo- sium on Microelectronics, 576-581 3. Bathe K.-J., 1996,Finite Element Procedures, Prentice Hall 4. Benson M.A., 1969, Thick-film screen printing, Solid State Technology, 53-58 5. Cox H.L., 1952, The elasticity and strength of paper and other fibrousmaterials,British Journal of Applied Physics, 3, 72-79 6. Hohl D., 1997, Controlling off-contact, Specialty Graphic Imaging Association Journal, 4, 5-11 7. Irgens F., 2008,Continuum Mechanics, Springer 8. Jeong J., Kim M., 1985, Slow viscous flow due to sliding of a semi-infinite plate over a plane, Journal of Physics Society, 54, 1789-1799 9. KobsD.R., VoigtD.R., 1970,Parametric dependencies in thick film screening,Proc. ISHM, 18, 1-10 1036 E. Horvath et al. 10. Krebs F.C., 2007, Large area plastic solar cell modules, Materials Science and Engineering: B, 138, 2, 106-111 11. Krenchel H., 1964,Fibre Reinforcement, Akademisk Forlag, Copenhagen, Denmark 12. Miller L.F., 1969, Paste transfer in the screening process, Solid State Technology, 46-52 13. Riedler J., 1983, Viscous flow in corner regions with a moving wall and leakage of fluid, Acta Mechanica, 48, 95-102 14. Riemer D.E., 1988a, Analytical model of the screen printing process: part 1, Solid State Techno- logy, 8, 107-111 15. Riemer D.E., 1988b, Analytical model of the screen printing process: part 2, Solid State Techno- logy, 9, 85-90 16. Riemer D.E., 1989, The theoretical fundamentals of the screen printing process,Hybrid Circuits, 18, 8-17 17. ScottBlairG.W.,Hening J.C.,Wagstaff A., 1939,The flowof cream through narrowglass tubes, Journal of Physical Chemistry, 43, 7, 853-864 18. Taylor G.I., 1962, On scraping viscous ?uid from a plane surface, Miszallangen Angewandten Mechanik, 313-315 19. Viricelle J.P., 2006, Compatibility of screen-printing technology with microhotplate for gas sensor and solid oxide micro fuel cell development, Sensors and Actuators B: Chemical, 118, 1/2, 263-268 20. White G.S., Breward C.J.W., Howell P.D., 2006, A model for the screen-printing of New- tonian fluids, Journal of Engineering Mathematics, 54, 49-70 Modelowanie mechaniczne i optymalizacja żywotności siatki w technologii druku sitowego obwodów mikroelektronicznych Streszczenie Upowszechnionąna skalęmasowąmetodęnakładaniawarstwklejów imateriałówadhezyjnychwtech- nologii grubowarstwowej oraz niskotemperaturowo współspiekanych obwodów drukowanych na płytkach ceramicznych jestdruk sitowy.Zapomocą tej technologii, substancja klejącanakładana jest napowierzch- nię substratu gumowym raklem, przeciskającymklej przez siatkęwykonaną ze stali nierdzewnej pokrytej emulsją fotolitograficzną. W pracy omówiono bezstykową wersję sitodruku, ponieważ jest ona obecnie standardową techniką stosowaną w przemyśle mikroelektronicznym. W prezentowanej pracy wygenero- wano w systemie ANSYS model elementów skończonych badanej siatki do określenia jej odkształceń i ocenymożliwości ograniczenia poziomunaprężeń pod kątem zwiększenia żywotności. Opracowano i wy- tworzono indywidualną aparaturę pomiarową do weryfikacji stanu odkształcenia obliczonego modelem MES. W wyniku badań stwierdzono, że zaproponowana modyfikacja geometrii rakla pozwala obniżyć maksymalne i średnie naprężeniaw siatce. Obserwowanymzjawiskiem jest także stopniowa utrata napię- cia siatki w trakcie normalnej eksploatacji, co prowadzi do pogorszenia jakości sitodruku. Zmodyfikowany kształt rakla kompensuje ten efekt i wydłuża żywotność siatki poprzez obniżeniewartości naprężeń i zop- tymalizowanie parametrów geometrycznych druku bezstykowego. Manuscript received December 7, 2011; accepted for print February 2, 2012