Ian Jones1,* and Jonathan Griffiths2
1TWI Ltd, Granta Park, Great Abington, Cambridgeshire, CB23 6AL, UK
2Cavendish Laboratory, University of Cambridge, 19 J J Thomson Avenue, Cambridge, CB3 0HE, UK
Paper published in Mater.Trans. Vol. 56 No. 7, July 2015
Laser welding can make very precise joints in plastics products, both in terms of joint location and amount of heating applied. These welding methods allow complex products such as microfluidic devices to be made, where channels and structure resolution below 100μm is regularly used. However, to date, the dimension of welds made using lasers has been limited by the focus spot size that is achievable from the laser source. Theoretically, the minimum spot size possible from a good quality laser beam is comparable to the wavelength of the radiation emitted. Practically, with reasonable focal length optics, the spot size achievable is a few factors larger than this. The resulting weld even larger than this. The narrowest welds feasible to date have therefore been 10 to 20μm wide using a near-infrared laser source.
The aim of this work was make welds less than 10μm wide in PMMA thermoplastic, using EB lithography to prepare laser absorber tracks and channels, followed by laser welding to carry out welds of the order of 1 μm wide. This technique should allow welds to be made below the
resolution limit of the near-infrared laser.
Welded joints with a width of 1 μm have been achieved and channels with a width of 5 μm. The procedure was based on the principle of transmission laser welding using a thin coating of infrared absorbent material at the joint interface. The coating was patterned using electronbeam lithography to obtain the required resolution in a reproducible manner, and that resolution was retained after the transmission laser welding process. The joint strength was ratified using larger scale samples. The results demonstrate that plastics products could be made with a high density of structure with resolution below 1 μm, and that welding can be applied without excessively heating regions outside the weld lines. This may be applied to smaller scale sensor and analysis chips, micro-bio and chemical reactors using either liquids or gases, and to microelectronic packaging.
With the drive towards ever smaller scale complex plastic products such as biological analysis chips, chemical microreactors and electronics products in plastics, there is a need for welding processes that meet these challenges. Laser welding developments in recent years have shown that the laser beam size can be used to limit the weld size. Welds of the order of 10μm in width have been demonstrated using a focused infrared laser beam of similar dimension. Welds much smaller than this are not possible using this technique due to the resolution limits inherent in focusing a near infrared laser.
An alternative technique has been investigated in this work, using precise patterning of laser absorber dye on the plastic surface to define the weld position. This has enabled joints to be made an order of magnitude smaller than this. The new method used electron beam lithography to apply laser absorber in precise patterns, mimicking the methods used to build micro-electronic circuits. The lithographic methods were also used to generate the micro-channels with dimension of approximately 5 μm width and depth. The absorber tracks have then been used to generate welds between two plastic parts using laser heating, sealing a lid over the channels.
2.1 Literature survey of high precision laser welding of plastics
To date, laser processing has provided the means for manufacturing using the smallest weld dimensions. By comparison, ultrasonic welding has been applied to welds with dimensions down to 500 μm; whereas laser welding has been reported to produce welds of less than 50μm in width. Manipulation equipment is available to position components to accuracy better than 1 μm.1) The limiting factor in providing precise location of a weld was therefore the localisation of the heat source (i.e., the laser beam).
Nd:YAG, diode and fibre laser sources for transmission laser welding can provide small focused spot sizes (less than 50μm wide), and have been used to make welds of a similar width. A short focal length is required which reduces the flexibility of the laser system is reduced. Attempts have been made to precisely locate the absorber material either using a liquid coating process or a solid form as fine filaments. 2) These techniques have not previously improved the weld precision.
Transmission laser welding of plastics has been used in a variety of methods to create small and complex welded structures in plastics. A spin coating of infrared absorber dye has been used to weld 250μm thick polycarbonate (PC) foils using a Nd:YAG laser. 3) Minimum weld widths of around 150μm were achieved and it was concluded that weld width consistency could be improved by improving the spin coating consistency. It has also been shown that a film could be welded over square channels (50 x 50 μm) in a polymethylmethacrylate (PMMA) base plate using a 15mm wide diode laser line source covering many channels at once. 3) The process was applied to sealing of capillary gel electrophoresis chips made of PMMA containing 300 channels with 50 μm width and depth to be used for low-cost and high-throughput routine analysis of protein mixtures. 4) Low power (<1W) fibre-coupled laser diodes are available which provide a focused spot width of 25μm and have been used to produce weld widths of 50μm in clear to carbon black filled PC. 5) Weld strengths from 616 MPa were measured and thermal degradation of the PC was identified as a problem as well as limitations in process speed. A line source diode laser and mask based system has also been applied to assemble microfluidic devices. Accuracy in positioning of the welding seam in the order of 5-10μm were claimed. 6)
Low power diode laser sources have also been reported giving welding seams as narrow as 10μm in welds in polyethylene terephthalate glycol (PETG). The PETG was coated with narrow band IR absorber. 7) A process using a thin carbon coating (5-20 nm) as an absorber has been developed for welding channels in the 100μm size range. It was also suggested that smaller features (10-30 μm) could be welded in the same way and potentially multi-layer structures. 8)
In summary, the smallest weld width demonstrated and published in welding plastics has been 10 μm, using a low power diode laser source with a small focus spot.
2.2 Electron beam (EB) lithography methods
EB lithography is typically used to generate patterns for electronic circuits in semi-conductor materials. It has been used to produce line widths as small as 20 nm in microelectronics applications. The procedure is based upon deposition of coatings (resists) over a wide area, followed by local sensitisation of the resist material using electron beam patterning and then development of the pattern by flushing away the sensitised material with a solvent wash. The procedure can progress in two general forms (Fig. 1). Positive resist, in which the EB degrades the resist in a predefined pattern such that the EB treated material can be dissolved away by the developer solvent.
Negative resist, in which the EB locally cures the resist and the remaining uncured material is dissolved away.
2.3 Laser welding of tracks patterned using EB lithography
Once the tracks of resist containing IR absorber have been applied, the top can be applied and clamped. An IR laser can be used to heat the absorber to melt and weld in precise locations. The laser can be significantly larger than the absorber tracks (see Fig. 1), but will pass through the plastics that are not coated with absorber to provide welds only in the required positions.
A number of requirements must be successfully demonstrated to allow laser welding of parts with lithographic patterning:
- The absorber material must dissolve in the resist fluid.
- The absorber material must remain active and absorbent in the resist material (no reaction with it).
- The resist properties must be unimpaired by the absorber; i.e., it must cure or be degraded as designed in following EB treatment.
- The absorber must be unaffected by the EB treatment, and remain in place after treatment.
- The absorber must be unaffected by any pre or post processing required on the resist. Some resists require oven drying or curing.
Fig. 1 Procedure for EB lithography using positive or negative resist types and concept for transmission laser welding to seal the edges of a microchannel.
- The absorber/resist combination must not act as a barrier to welding, either as a contaminant or other barrier to polymer chain diffusion.
To simultaneously generate micro-channels and infrared absorber tracks at their edges using electron beam lithography and seal the channels using welds in plastics smaller than any previously achieved (less than 10μm width) through use of transmission laser welding to provide controlled heating and melting of the tracks.
Glass slide (1mm thick) was used as the substrate for coating. 25 x 100mm samples were used. The top sheet (channel lid) was PMMA sheet 250μm thick.
The following resist materials were used:
495PMMA A8 from Micro-Chem. 8% solids of PMMA with a molecular weight of 495,000 in anisole solvent.
AR-N 7520-18 from Allresist. 18% solids of Novolak epoxy resin, naphthochinondiazides and cross-linking compounds in propylene glycol methyl ether acetate.
The resist materials were mixed with laser absorbing dye, type A194 from Crysta-lyn. The following mixes were tested:
Mix C 5mL AR-N 7520-18 + 0.16 g A194 premixed with 1mL acetone.
Mix F 11.5mL 495PMMA A8 + 0.025 g A194 premixed with 1mL acetone.
The most suitable resist materials were selected based on the dissolution properties and the absorption strength of the mixture for laser wavelengths available for welding.
Both the above mixes appeared to dissolve the absorber, giving a dark green colour to the mixture. In a thin coating on glass substrate Mix C absorbed 8.9% of incident radiation at 940 nm and Mix F absorbed 9.7%.
The EB lithography was carried out using a Vistec EBPG5200 series equipment. It has an accelerating voltage of 20100 kV and can provide features as small as 8 nm. Laser welding was carried out using a 200W fibre laser from IPG (1070 nm wavelength), defocused to a 0.5mm wide beam and a 150W diode laser from Laserline defocused to a 3mm wide beam. The beam was manipulated using an optical scanner. The parts for welding were mounted in a clamp under an acrylic cover plate.
4.3 Materials preparation, lithography and welding procedure
The two resist/absorber mixes were applied to the glass substrate by spin coating at 4000 rpm for 1 min. This was expected to provide a 0.4μm thick coating. Two series of trials were carried out using the positive tone resist, one giving 5 μm feature sizes, the second giving 0.5μm feature sizes.
Mix C (Negative tone epoxy based resist) was EB treated to give 5 μm features using doses from 100 to 190 μC/cm2, with a beamstep of 25 nm and current 31 nA. It was developed in Shipley MF322 developer (tetra methyl ammonium hydroxide (TMAH) with a normality of 0.268 N, diluted with DI water in the ratio 0.6 : 0.4).
Mix F (positive tone PMMA based resist) was EB treated to provide 5 μm and 0.5μm features. Doses from 200 to 1032 μC/cm2 at 100 kV were used with a beamstep of 25 nm and current of 34 nA to 50 nA. The resist was developed in MIBK:IPA 1 : 3 for 2 min, rinsed in IPA and blown dry.
Laser welding was carried out using a beam wider than the tracks deposited, over a range of processing conditions. A top part of clear un-patterned PMMAwas placed over the sample parts. The samples were mounted in clamping equipment with the parts pressed up to the lower side of a clear acrylic cover sheet. The beam from a fibre laser was transmitted through the cover sheet and the top part of the samples to heat the patterned absorber deposit.
Welding trials were carried out using standard infrared absorber coating, coated onto PMMA material with no patterning to define suitable welding conditions. This was repeated with resist mixes C and F. Weld strengths were tested manually in these initial trials. The EB patterned samples were processed at a range of different laser powers and beam manipulation speeds. Weld samples were selected for sectioning and examination. Regions of the resist deposits that had not been patterned were also welded at the same conditions to compare directly with the initial welding trials and assess whether strong welds could be made using the spin coated material. Tensile lap-shear tests were carried out with these samples.
The samples were examined by optical and scanning electron microscopy after EB treatment, developing and welding. The welds produced were sectioned and images were taken using optical microscopy. Tensile tests were carried out on large scale samples prepared in the same way as the EB patterned samples.
Fig. 2 Brightfield SEM images showing AR-N 7520-18 resist after EB exposure and developing. The small and large dark specks indicate poor dissolution of the absorber. The lines exposed to the EB are 5 μm wide and separated by 25 μm. The resist has been removed along the lines.
Fig. 3 SEM images showing 495PMMA A8 resist after EB exposure and developing. The lines exposed to the EB are 25μm wide and separated by 5 μm wide tracks. The resist has been removed along the 25μm lines. The left image exposed at 200 μC/cm2, shows no foaming. The right image exposed at 345μC/cm2 shows foaming of the PMMA substrate.
5.1 Negative resist
Samples coated with Mix C, the AR-N 7520-18 negative resist, are shown in Fig. 2 after EB exposure and development. Unexpectedly, the resist film was still present everywhere except for the exposed 5 μm lines, like a positive tone resist. When the initial laser welding trials were carried out on samples coated with the AR-N 7520-18 resist, the joint strength was not good. Further work was not carried out with this resist due to problems with poor dissolution of the absorber and poor weld strength.
5.2 Positive resist, 5 μm feature size
Samples coated with Mix F, the 495PMMA A8 positive resist with a 5μm feature size, are shown in Fig. 3 after EB exposure and development. The samples exposed to high charge density showed foaming in the PMMA substrate. Samples exposed to less than 300 μC/cm2 showed no foaming. Welds were carried out using a clamp pressure of 0.3N/mm2, 0.5mm laser beam diameter, 48W laser power and a process speed of 100mm/min. The same conditions had shown good weld strength in large scale samples. A cross-section of the 5 μm wide welds coinciding with the lithographically formed tracks is shown in Fig. 4.
Fig. 4 Laser welded sample in cross section. The tops of the 5 μm wide tracks have been welded to the top sheet of PMMA. Gaps with a height of a few hundred nanometres have been left on either side of the welded tracks.
Fig. 5 Optical images showing 495PMMA A8 resist after EB exposure and developing. The lines exposed to the EB are 2.5μm wide and separated by 0.5μm wide tracks. It was exposed at 417 μC/cm2.
5.3 Positive resist, 0.5μm feature size
Samples spin-coated with Mix F, the 495PMMA A8 positive resist with a 0.5μm feature size, are shown in Fig. 5 after EB exposure and development. Following further process optimization, welds were carried out using a clamp pressure of 0.3N/mm2, 2.0mm laser beam diameter, 48W laser power and a process speed of 3000mm/min.
SEM images of the welds in cross-section were taken from a surface prepared by microtome. A cross-section of the 0.5μm wide welds with 2.5μm separation, made in a patterned region is shown in Fig. 6.
A set of samples with the same spin coating and welding conditions were prepared for tensile-shear tests. The 25mm wide samples failed at an average of 624 N, with failure in the parent material at the edge of the weld. This is a similar result to that seen in laser welding PMMA with a normal infrared absorbing coating.
5.4 Positive resist (Mix F), 5μm channels with 1 μm welds
EB exposure and development was used to generate 5μm wide channels with 1 μm wide infrared absorber tracks on either side (see Fig. 7). Welds were carried out using a clamp pressure of 0.3N/mm2, 3.0mm laser beam diameter, 50W laser power and a process speed of 4000mm/min. A cross-section image after welding is shown in Fig. 8.
Fig. 6 Laser welded sample in cross section. The tops of the 5 μm wide tracks have been welded to the top sheet of PMMA. Gaps with a height of a few hundred nanometres have been left on either side of the welded tracks. (scale bar 6 μm).
Fig. 7 Sample of PMMA on glass substrate in cross section with 5 μm wide channels and 1 μm infrared absorber tracks at the tops of the channel walls (and some in between) prepared using EB lithography.
Fig. 8 Welded sample in cross section with 5 μm wide channels prepared using diode laser transmission welding.
6.1 Negative resist
The negative tone resist that was selected was successfully patterned using EB lithography, but unexpectedly behaved as a positive tone resist. The EB treated areas that should have been cured to a cross linked epoxy compound, were removed in the solvent developer, whereas the untreated areas remained. This suggests that the epoxy had cured prior to EB treatment, possibly as a result of the presence of the laser absorber dye.
The absorber dye did not dissolve fully in the propylene glycol methyl ether acetate base solvent of the negative resist, even when acetone was used in advance to dissolve the dye. It is not expected that this problem would be insurmountable. However, the nature of the cured cross-linked material acting as a barrier to diffusion of the plastic at the joint surfaces is likely to be a fundamental problem that would make the use of a negative tone resist difficult as a coating to promote heating and welding.
6.2 Positive resist
The behaviour of a positive tone resist is better suited to use as a carrier for absorber dye. After EB treatment, albeit at a lower beam current than usually used, the resist could be preferentially removed by the solvent developer, leaving well defined regions containing absorber dye and through deeper removal to generate channels in the same series of processing activity. The absorber in the remaining tracks of coating was suitable promoting selective heating and welding. There do not appear to be any problems with reactions between the dye and the resist, or from the EB treatment that could have affected the performance of the dye absorber.
6.3 Potential developments
In this work welds with a width of 0.5μm were demonstrated. The size was accurately dictated by the dimensions of the resist pattern and the hence by the EB treatment. Given that it is feasible to pattern the resist to dimensions of 20 nm, it should be possible to provide welds with even smaller dimensions, though there are no applications envisaged for smaller welds.
In theory, it should be feasible to extend the process to polymers other than PMMA. Electron beam processing can be used to promote chain scission in a wide range of polymers, potentially allowing selective removal, using solvents, of EB treated regions of polymer. However, commercial resists are generally only available in PMMA, so procedure development would be required to obtain the most suitable EB treatment, solvent type and dissolution parameters.
Successful demonstration of what is believed to be the smallest welds produced in plastics, allows new domains to be explored in welding of plastics applied to parts with micro- and nano-scale features such as in microfluidic devices manufacture. It follows work which demonstrated that short pulse laser welding could be used to minimise distortion when joining plastic parts. 9)
Welds of 0.5μm width, believed to be the smallest made in plastics, have been demonstrated in polymethylmethacrylate (PMMA) using a combination of electron beam lithography of a positive tone resist material based on PMMA in anisole doped with an infrared laser absorber, and transmission laser welding using a near infrared fibre laser. There is potential for welds in other plastics using similar techniques.
Philip Hoyle and Mike Butler of Vistec Lithography are thanked for their advice concerning EB lithography trials on new materials. Chris Burt of the Faculty of Engineering & Physical Sciences, University of Surrey, is thanked for assistance with SEM imaging. Alex Davis of the University of Cambridge, is thanked for assistance with preparation of absorbers and test pieces.
- H. Klein and E. Haberstroh: ANTEC 1 (1999) 14061409.
- I. A. Jones: Precision Laser Welding of Plastic Components, (TWI members report, Cambridge, 2006).
- T. Klotzbuecher, T. Beaune, M. Ritzi, K.-S. Drese and U. Teubner: Microclear®A Novel Method for Diode Laser Welding of Transparent Micro Structured Polymer Chips, (ICALEO, 2004).
- A. Griebel, S. Rund, F. Schoenfeld, W. Doerner and R. Konrad: Lab Chip 4 (2004) 1823.
- D. Grewell and A. Benatar: ANTEC (2003) 10391044.
- J.-W. Chen and J. M. Zykbo: Proc. SPIE 5718, (2005) pp. 9298.
- T. Ussing, L. V. Petersen, B. Helbo and L. Hoslet: Proc. Int. Conf. on Multi-Material Micro Manufacture, Karlsruhe, (2005) pp. 291293.
- W. Pfleging, O. Baldus, M. Bruns, A. Baldini and E. Bemporad: Proc. SPIE 5713, (2005) pp. 479488.
- I. A. Jones and V. Loyo-Maldonado: Laser User Mag. 62 (2011).