使用 UV LED 的无掩模光刻

A UV light emitting diode (LED) with a maximum output of 372 nm was collimated using a pinhole and a small plastic tube and focused using a microscope objective onto a substrate for direct lithographic patterning of the photoresist. Movement of the substrate with a motorised linear stage (syringe pump) allowed lines in SU-8 to be pattered with a width down to 35 lm at a linear velocity of 80 lm s−1 , while in the dry fifilm resist Ordyl SY 330, features as narrow as 17 lm were made at a linear velocity of 245 lm s−1 . At this linear velocity, a 75 mm long feature could be patterned in 5 min. Functional microflfluidic devices were made by casting PDMS on a master made by LED lithography. The results show that UV LEDs are a suitable light source for direct writing lithography, offering a budget friendly, and high resolution alternative for rapid prototyping of features smaller than 20 μm.

Light emitting diodes (LEDs) are an inexpensive but powerful light source that have found widespread use around the world as well as having several scientifific applications, particularly as alternative light sources for flfluorescence and absorption detection in capillaries and microchips.1,2 For absorbance detection, they have become an attractive alternative to conventional lamps in the visible region due to their high light intensity over a narrow wavelength distribution and their lower noise, which translate to an improvement in LODs by a factor of 10. They have also gained widespread popularity as excitation sources for flfluorescence, and while performance is inferior to that achievable with a laser, the cost is substantially less, with LEDs typically costing <$10, while solid state lasers are at least $1000, and gas lasers an order of magnitude more again. Additionally, LEDs have a lifetime of approximately 10 000 h, making them a very cost effective alternative to conventional light sources.

LEDs were mounted inside a 50 mm long black polymer tube (i.d. 5 mm) positioned 10 mm from the end which contained a 3 mm pinhole. This was then collected using a microscope objective (Mitutoyo M Plan Apo 50× long working distance, numerical aperture 0.55) and focused on the substrate. For the creation of dots, a silicon wafer was used as a shutter to control exposure times. For direct writing of lines, the substrate was placed on the plunger of a Harvard 22 syringe pump. For the direct writing of lines in the dry fifilm resist, an additional 1 mm pinhole was placed on top of the microscope objective. The linear rate of the stage was controlled by varying the flflow rate of the syringe pump. A schematic drawing and photograph and of the experimental set-up are given in the electronic supplementary information.

Functional microflfluidic devices were made by casting PDMS using standard methods detailed elsewhere.11,12 Access holes were made at the channel termini using a hole puncher and the channel was sealed using a clean microscope slide.

SU-8 is one of the most commonly used photoresists in microflfluidics and was therefore chosen for initial studies. Light from the LED was focused onto a microscope slide coated with SU-8 photoresist and exposed for intervals of 5, 10, 20, 40 and 60 s. After development, oval dots were observed on the substrates with the size of the dots increasing with exposure time. Relatively small ovals of 180 lm × 160 lm were obtained with a 5 s exposure (shown in Fig. 1a). A 60 s exposure resulted in features approximately 10 times larger, approaching 2 mm in size, which is highly indicative of over exposure of the photoresist. The reproducibility of the system was excellent, with four dots shown in Fig. 1a.

Fig1

Having demonstrated that it is possible to expose SU-8 with the UV LED, it was necessary to determine whether it would be possible to ‘write’ a pattern onto the photoresist using a motorised stage. The recommended exposure for a 100 lm thick fifilm is 250 mJ cm−2 . 14 Assuming only 20% of the output of a 1 mW LED is focused in a spot with a diameter of 100 lm, the power in this spot will be 2.5 W cm−2 , implying that a 0.1 s exposure should be suffificient. It should therefore be possible to conduct direct writing lithography with linear velocity up to 1 mm s−1 . For this purpose, the substrate was placed on top of a motor-controlled stage, the piston of a Harvard 22 syringe pump. The pump was set at flflow rates between 100 and 1200 lL min−1 for a 250 lL syringe providing a linear velocity between 7 and 80 lm s−1 . Inspection of lines exposed at 80 lm s−1 revealed that the initial lines were slightly wavy and showed regularly spaced bulges. The bulges were attributed to changes in linear velocity of the piston due to defificiencies in the stepper motor, bulges resulted from low linear velocities and thin lines from higher velocities. The wavy nature of the lines resulted from non-straight movement caused by the actuation on a thread causing the piston to move along a sinusoidal path. Repeating the experiments using a brand new Harvard 22 syringe pump resulted in the writing of straight lines. A smooth and straight moving platform is therefore considered to be crucial for direct writing LED lithography. Fig. 1b shows a representative photograph of a straight line created by direct writing LED lithography at a linear velocity of 80 lm s−1 . The width of this line was determined to be 35 lm by SEM. Lines patterned using a linear velocity of 40 lm s−1 were found to be approximately twice as wide, with a nominal width of 68 μm.

Finally, to demonstrate the practical potential of this set-up, direct writing lithography was used to create a simple cross template for the construction of a PDMS microchip suitable for electrophoresis. A photograph of the microchip channels fifilled with green dye is shown in Fig. 2a, an optical image and SEM of the cross of the master are given in Fig. 2b and c, respectively, and an SEM of the cross-section of the channel cast in PDMS is given in Fig. 2d. This template was constructed at a linear velocity of 210 lm s−1 and required a little over 10 min to expose. The fifilling of the channels of this PDMS replicate with an aqueous solution was similar to that of PDMS channels replicated using masters made by conventional techniques. The cross geometry, is suitable for microchip electrophoresis, but the technique is certainly not limited to this.

RMG and MCB would like to acknowledge the Australian Research Council for the provision of an Australian Postdoctoral Fellowship (DP0557083 and DP0453223). The authors would also like to thank Mr John Davis and Dr Karsten G¨omann (Central Science Laboratory, University of Tasmania) for assistance with evaluating the exposure system and with the SEM images and Elga Europe for donation of the Ordyl SY330 dry fifilm resist.