Laser prototyping of printed circuit boards

Transcription

1 OPTO ELECTRONICS REVIEW 21(3), DOI: /s Laser prototyping of printed circuit boards M.R. NOWAK *, A.J. ANTOŃCZAK, P.E. KOZIOŁ, and K.M. ABRAMSKI Faculty of Electronics, Wroclaw University of Technology, 27 Wybrzeze Wyspianskiego Str., Wroclaw, Poland This paper describes the application of laser micromachining to rapid prototyping of printed circuit boards (PCB) using nano second lasers: the solid state Nd:YAG (532/1064 nm) laser and the Yb:glass fiber laser (1060 nm). Our investigations included tests for various mask types (synthetic lacquer, light sensitive emulsion and tin). The purpose of these tests was to determine some of the basic parameters such as the resolution of PCB prototyping, speed of processing and quality of PCB mapping with commonly available laser systems. Optimization of process parameters and the proposed conversion algorithm have allowed us to produce circuit boards with a resolution similar to that of the Laser Direct Imaging (LDI) techno logy. Keywords: printed circuit board (PCB), rapid prototyping, laser micromachining, Nd:YAG laser, Yb:glass fiber laser. 1. Introduction The ability to quickly manufacture a small series of printed cir cuit boards designed in a graphic software program may be a very useful skill to have. Laser technology allows one to avoid all the procedures related to known technologies such as, for example, the photochemical processes, which are more suited for longer production series of PCBs. Some PCB em bodiments manufactured using laser technology have already been discussed in the literature [1 3]. The most important techniques used in these examples were: direct exposure of photoresist (for laser direct imaging technology), selective re moval of mask material protecting copper from chemical etch ing and structuring of plastics and ceramics for subsequent electroplating metallization. Applications for which the proto types are intended often require the PCBs to be made in the high density interconnects (HDI) technology, where the elec tric circuit paths and spaces are at most 75/75 μm and the pack aging level is 20 pads per cm 2. Our laser micromachining tech nology presented in this paper is based on the use of solid state lasers (532 nm and 1064 nm) and a fibre laser (1060 nm) with a galvo scanner. These are used to selectively remove the mask material covering the copper surface of the PCB laminate. 2. The method for selective removal of protective mask Before the laser prototyping began, we covered the entire laminate surface with a protective mask (made of various materials). This step was intended to protect the copper sur face from chemical etching. The scanning laser beam re moves the protective layer applied to the surface in those areas where copper is to be etched. A negative of the de signed PBC is thus made. Next, the copper layer covering the PCB is chemically etched. The main differences bet ween the two technologies, namely between the laser direct imaging (LDI) and the laser micromachining, are shown in Fig. 1. In the description of Fig. 1 we emphasized the fact that laser micromachining replaces two LDI processes. The processes of exposure and development are replaced with a one step process of laser micromachining prototyping. The quality of a PCB is determined by the proper selection of mask material. This material should display the following characteristics: lower level of ablation compared to copper, absence of destructive factors (such as melting or burn ing of edges), effective protection of copper during the chemical etch ing process, ease of mask material removal after etching. The basic PCB substrate material which we used for the selective removal technology was the FR4 Bungar glass epoxy laminate covered on one side with a 35 μm copper layer. From a variety of tested materials, we selected three coatings which met the above requirements: tin plated, synthetic lacquer spray, photoresist (POSITIV 20). Selective laser removal of protective mask has one extra advantage: all of the photolithographic processes can be avoided. * e mail: maciej.nowak@pwr.wroc.pl 320 Opto Electron. Rev., 21, no. 3, 2013

2 Fig. 1. Comparison between the diagrams of laser direct imaging (LDI) [4] and laser micromachninig protyping. Processes of exposure and development are replaced with a one step process of laser micromachining. 3. Experimental setup For the purpose of PCB prototyping two galvo scanner sys tems were used (Fig.1). Both systems included pulsed lasers: the first one was equipped with a Nd:YAG 532/ 1064 nm solid state laser and the second one with a Yb:glass fiber laser Laser systems The laser systems we used in our experiments were typical for this kind of applications. Each system consisted of a laser beam source: the first one was the Nd:YAG pulsed laser with two wavelengths of 532/1064 nm and the second one was the fiber Yb:glass laser with a 1060 nm wave length. Both systems were equipped with a galvo scanner (Fig. 2). The z axis was fixed by adjusting. The basic parameters of the Nd:YAG laser were as fol lows: maximum average powers: 10 W for = 532 nm and 25 W for = 1064 nm; pulse repetition rates: khz, spot diameter for both wavelengths: 35 μm, pulse duration: ns, depending on the repetition rate, beam quality factor: M Fig. 2. Block diagram of PCB prototyping process with a laser galvo scanner system. The basic parameters of the fiber Yb:glass laser were the following: average power: 20 W, pulse repetition rate: 2080 khz, pulse duration: ~100 ns, spot diameter: 35 μm, beam quality factor: M In both systems, the focused beam was translated by means of the galvo scanners with a maximum speed of 20 m/s. The laser beam was focused with a 163 mm F Theta lens. The working area of the galvo scanner head was mm. The galvo scanner driver and interface allo wed for working with CAD, pdf and bmp files Testing procedures First, we designed a simple test pattern [(Fig. 3(a)] intended to determine the optimal process parameters. Next, a sec ond, more advanced pattern [(Fig. (3b)] was designed in order to characterize the quality and resolution of proto typed PCBs. After the best process parameters for applied laser sys tems were fixed, we performed another test series with the second pattern [(Fig. 3(b)]. This pattern was divided into five sectors. In the picture the white colour corre sponds to the removal of the layer protecting copper us ing a laser beam. In order to be more understandable for the electronics community, we decided to use mils (one thousand of an inch) as our main unit and to always add metric units in bracket. Our 1st test sector includes seven groups of paths with the following sizes: 3, 4, 5, 6, 8, 10 and 15 mils (76, 102, 127, 152, 203, 254 and 381 μm, re spectively). The paths are mapped out in such a way that the laser line (the straight sequence of overlapping la ser made dots) runs perpendicularly to path direction. The distances between the paths are equal to the path widths; Opto Electron. Rev., 21, no. 3, 2013 M.R. Nowak 321

3 Laser prototyping of printed circuit boards Fig. 3. Test patterns: (a) pattern used to determine the optimal process parameters of the systems, (b) pattern used to obtain the laser system resolution with fixed optimal process parameters, (c) photos of selected test items. In the 2nd sector the laser lines run along the paths. All other parameters are the same as those in sector 1; In the 3rd sector the paths are oriented at ± 45 degrees to the laser line; In the 4th sector the paths are oriented at ± 30 degrees to the laser line; Lastly, the 5th sector demonstrates the ability of microma chining to make bending paths (rings) of different widths: 5, 10, 15, 20 mils, and pads of different diameters: from 10 to 35 mils (254 to 889 μm) with a step of 5 mils (127 μm, 6 groups of pads with 12 pads per group). All investigations were carried out in the natural environment air. 4. Measurement results Taking into account the above test patterns, first we carried out the test presented in Fig. 3(a) in order to determine the optimal process parameters of our laser systems. Next, the resolution tests were performed with the system parameters set at the op timal values. For the laser galvo scanners operating with a solid state laser at the wavelength of = 1064 nm (with pulse duration = 10 ns) the best results amongst all tested masks (tin, lacquer, photoresist) were obtained for the following val ues: power P = 1 W, pulse repetition rate f = 10 khz, hatching H = 0.03 mm and velocity v = 50 mm/s. Similar tests performed for = 532 nm (and =10ns) produced much worse results. For that reason further testing at this wavelength was abandoned. There were two reasons of ignoring = 532 nm. The beam diameters in our system were the same (35 um) for both wavelengths (532 nm and 1064 nm). Hence, there were no differences in resolution. How ever, in the case of = 532 nm, we observed a much stronger interaction (as a result of higher absorption) with the copper substrate, which appeared as copper dust on the copper sur face. In the case of tin, we were unable to effectively remove the tin layer, because absorption for this material at 532 nm is relatively small compared to that at 1064 nm. These were the reasons for not using the 532 nm wavelength. Table 1. Summary of representative measurement results wide tracks performed at different angles (system Nd:YAG/1064 nm). Design path width [mils (μm)] The measured width of the path made with a solid state laser [mils (μm)], Parameters: = 1064 nm, P = 1 W, f = 15 khz, v = 50 mm/s; Mask: tin layer (76) 3.35 (85) 3.58 (91) 3.66 (93) 3.70 (94) 3.62 (92) 3.54 (90) 4 (102) 4.21 (107) 4.45 (113) 4.25 (108) 4.33 (110) 4.49 (114) 4.57 (116) 5 (127) 5.28 (134) 5.67 (144) 5.51 (140) 5.04 (128) 5.20 (132) 5.31 (135) 6 (152) 6.14 (156) 6.61 (168) 6.22 (158) 6.26 (159) 6.30 (160) 6.50 (165) 8 (203) 7.99 (203) 8.39 (213) 8.23 (209) 8.31 (211) 8.27 (210) 8.43 (214) 10 (254) (255) (263) (262) (258) (273) (272) 15 (381) (384) (390) (380) (391) (385) (384) 322 Opto Electron. Rev., 21, no. 3, SEP, Warsaw

4 For the laser galvo scanner operating with the Ytterbium fibre laser (with pulse duration around = 100 ns) the opti mal process parameters were as follows: power P =14W, pulse repetition rate f = 20 khz, hatching H = 0.03 mm and velocity v = 150 mm/s. Representative results of the measurements performed for the case of Nd:YAG laser (1064 nm) are given in Tables 1 and 2. Table 2. Summary of representative measurement results pads (system Nd:YAG/1064 nm). Design pad dimension [mils (μm)] The measured width of the pad made with a solid state laser [mils (μm)], Parameters: = 1064 nm, P = 1 W, f = 15 khz, v = 50 mm/s; Mask: synthetic lacquer 5 (127) 4.80 (122) 10 (254) 9.92 (252) 15 (381) (367) 20 (508) (496) 25 (635) (639) 30 (762) (760) 35 (889) (885) These results allow one to obtain PCBs with the mini mum path width and the minimum space between paths both equal to 3 mils (76 μm). In practice, these values do not depend on the angle between the laser line and the path direction. We also determined that the minimum obtainable pad diameter (see Table 2) was 3 mils (76 μm). However, due to the laser beam diameter (35 μm), the actual accuracy was much lower. 5. Implementation of the designed circuit board Figure 4(a) shows a sample PCB laser machining design made in the Altium Designer. Figure 4(b), in turn, is a photo of the actual manufactured PCB. In order to present the quality of laser micromachining, three separate characteris tic sectors were selected on the PCB. These sectors repre sent typical shapes and elements of paths, pads and solder areas. Each sector is shown in detail in Fig. 5. Figure 5(a) is the designing structure. The three series of photos below corre spond to different mask materials (photoresist, tin and lac quer) these are Figs. 5(b), 5(c) and 5(d), respectively. The designed paths are typical and have 8 10 mils in width. These enlarged sectors illustrate how effectively the laser system mapped the geometric dimensions for a variety of used masks. In order to produce the required PCB, a special algo rithm of file conversion into the laser system was deve loped. This algorithm was created for the Altium/Protel programs and is compatible with the popular Gerber file format. For all the investigated masks the PCBs displayed good quality and high accuracy of mappings. Fig. 4. (a) PCB as seen in Altium TOP layer. White rectangles are sectors for which the geometric dimensions of produced circuits were tested; (b) PCB prototype made with a fibre laser, the pho toresist mask. Depending on the PCB size and the paths and pads den sities, the manufacturing process may take up to a few min utes. We found that for the fibre galvo scanner system, the PCB presented in Fig. 4, with a surface area of 4.4 cm 2, was made in 110 s. Hence, one square centimetre was prototy ped in about 25 s. Electrical circuits covered with a mask layer in the form of tin, photoresist or synthetic lacquer prevent the copper from oxidizing. There is no need for any additional protec tive layers. In the case of PCBs with a photoresist or a lac quer layer, in order to remove the mask for soldering it suf fices to selectively apply either isopropanol or acetone. For the case with a tin mask, however, the areas are, in principle, ready for direct soldering. 6. Conclusions The above presented experiments prove that simple and easily available laser micromachining systems at wave length of around 1 μm, with selected mask materials such as tin, photoresist or synthetic lacquer, can be used for Opto Electron. Rev., 21, no. 3, 2013 M.R. Nowak 323

5 Laser prototyping of printed circuit boards Fig. 5. Elementary examples of designed (a) and micromachined (b d) PCBs for different mask materials. manufacturing of PCBs with path and space dimension of 3/3 mils, and pad diameter of 5 mils. The method is quite fast and practically applicable and can be used for pro totyping of single PCBs, particularly when a sequence of corrections is needed with feedback between the PCB de signing and implementation processes, which usually requires several iterations. 324 Acknowledgements This research was supported by the Wroclaw Research Cen tre EiT+ in the framework of the project The use of nano technology in modern materials NanoMat (POIG /08) financed by the European Regional Deve lopment Fund (Operational Programme Innovative Eco nomy Measure 1.1.2). Opto Electron. Rev., 21, no. 3, SEP, Warsaw

6 References 1. G. Koziol and W. Steplewski, Laser applications in high den sity interconnect PCB technology (in Polish), Elektronika konstrukcje, technologie, zastosowania, 49, (2008). 2. D.J. Meier and S.H. Schmidt PCB laser technology for rigid and flex hdi via formation, structuring, routing, LPKF La ser & Electronics AG Garbsen, ticles/hdi/pcb_laser_technology.pdf. 3. R. Barbucha, M. Kocik, J. Mizeraczyk, G. Kozioł, and J. Borecki, Laser Direct Imaging of tracks on PCB covered with laser photoresist, Bulletin of the Polish Academy of Sciences Technical Sciences, 56, (2008). 4. A. Süllau and A. Wiemers, Laser Direct Imaging, ILFA Pub lications, (1999), _Imaging.pdf Opto Electron. Rev., 21, no. 3, 2013 M.R. Nowak 325