A new method of hybrid photolithography, Laser Augmented Microlithographic Patterning (LAMP), is described in which direct laser writing is used to define additional features to those made with an inexpensive transparency mask. LAMP was demonstrated with both positive- and negative-tone photoresists, S1813 and SU-8, respectively. The laser written features, which can have sub-micron linewidths, can be registered to within 2.2 µm of the mask created features. Two example structures, an interdigitated electrode and a microfluidic device that can capture an array of dozens of silica beads or living cells, are described. This combination of direct laser writing and conventional UV lithography compensates for the drawbacks of each method, and enables high resolution prototypes to be created, tested, and modified quickly.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
Conventional mask-based lithography and direct laser writing (DLW) can be combined into a hybrid technique, which compensates for the drawbacks of each. Conventional photolithography using a mask to expose a photoresist to UV light is commonly used for the production of microelectronics, but has also expanded to a wide range of applications like microelectromechanical systems , lab-on-a-chip devices , and DNA microarrays . The power of photolithography lies in its ability to pattern large areas at sub-micron resolution in a matter of minutes. Two issues with photolithography are: i) the initial cost of the mask, which can be substantial if the resolution is high; and ii) the lack of adaptability of the mask since it cannot be altered. These issues can make prototyping a new device by photolithography fairly expensive. An alternative patterning method that does not suffer from these issues is direct laser writing (DLW), in which a laser is focused to a point in a photoresist and moved with respect to the sample to generate the pattern. One issue with DLW is the potentially long time required to cover large areas. The method we describe here is a combined lithographic technique that utilizes the large area patterning abilities of mask lithography and the high resolution of direct laser writing, which together provide sub-micron patterns, in a short amount of time, and at a reasonable cost. This combined lithography system can effectively pattern with much higher throughput than DLW on its own, while offering more pattern flexibility than ordinary mask lithography.
Others have also investigated hybrid lithography schemes such as the pairing of UV lithography with electron beam lithography (EBL) or nanoimprint lithography (NIL). For example, Kristensen et al. report the patterning of SU-8 with EBL followed by UV lithography resulting in features as small as ~100 nm linewidths . While EBL is most frequently used with positive tone photoresists to open areas for depositing metal contacts, this work shows that UV lithography combined with EBL can also be used to fabricate relief structures for molds. NIL has also been used as the first step in a sequential technique to pattern 500 nm features in SU-8 followed by UV exposure to make contacts at the 200 µm scale . These are attractive prototyping technologies that offer rapid pattern generation with fine resolution. The use of DLW as part of a hybrid scheme is less common, but Eschenbaum et. al. devised a hybrid DLW and UV lithography method for multi-scale patterning . They created high resolution patterns in three dimensions using an ultrafast laser for two-photon polymerization. The results are impressive showing the fabrication of a miniature 45° mirror to view particles from the side while traveling in a microfluidic channel, however their method requires two completely separate processing steps for the mask and laser patterning. Shear and associates have also demonstrated a DLW system that uses the mirror array from a projector to quickly transfer high-resolution patterns, thereby increasing the speed of DLW . Such two-photon systems are very expensive and are somewhat difficult to operate compared to a simple continuous wave diode laser reported here and neither examples in  or  demonstrated the ability to pattern positive photoresists as we do here.
In this work, we present a combined DLW and UV lithography method that uses both positive and negative photoresists as a fast and flexible lithographic technique, suitable for wafer scale definition of sub-micrometer features. Our method, which we call Laser Augmented Microlithographic Patterning (LAMP), first exposes a photoresist through an inexpensive transparency mask and then adds to that exposure using a DLW system before finally developing the pattern (Fig. 1). If one already has a microscope, then it can be readily adapted into a DLW system . LAMP is fairly low cost, straightforward to perform, and requires only a single layer of photoresist. We show that we can register the DLW features to within ~2 µm of the mask alignment marks and can achieve sub-micron linewidths. We envision LAMP to be an attractive alternative to expensive masks for researchers seeking to prototype new microtechnologies. We demonstrate two simple proof-of-principle devices, a microfluidic that can trap an array of single cells and an interdigitated electrode (IDE), to show the utility of LAMP.
The optical components of our DLW system are described in depth elsewhere . For the positive photoresist, Shipley S1813 (Microchem) was spin-coated on 2” glass wafers to a thickness of approximately 1.5 µm. For the negative photoresist, we doped SU-8 (Microchem) with fluorescein in a ratio of 1 mg of fluorescein to 1 mL of SU-8 to yield “SU-8F”. The SU-8F was thoroughly mixed before being spun onto 2” silicon wafers to a thickness of 5 µm. The wafers were soft baked according to their datasheets provided by Microchem and exposed through a transparency mask to a 100 W mercury lamp (Blak-Ray) for 60 s. The wafers were mounted onto an inverted fluorescence microscope (IX-71, Olympus), which had a motorized X-Y stage (Proscan III, Prior Scientific) coupled to a manual rotation stage (Thorlabs). A 405 nm continuous wave diode laser beam, (OBIS, Newport Corporation) overfilled the back aperture of a 20 × 0.75 numerical aperture (NA) objective and was focused onto the sample. The X-Y axis of the exposed mask pattern and the microscope stage were made parallel using the rotation stage, and the laser spot was aligned to a registration point on the exposed mask pattern. The DLW of the photoresist for S1813 was performed using 300 nW while the SU-8F required 500 µW. These powers were measured before the microscope, but at the sample they were 13% of these values due to reflective loss at the mirrors and because the laser overfilled the back aperture of the objective. Following the exposure of the resist in the desired pattern, the sample was submerged in the appropriate developer; S1813 was developed in Microposit 351 for one minute and SU-8F was post baked and developed in propylene glycol monomethyl ether acetate for one minute. Linewidths and registration of patterned features were characterized by scanning electron microscopy, SEM, (MIRA 3, Tescan). DLW of S1813 and SU-8F were tested at various speeds and powers to determine optimal conditions for augmenting mask features. Registration between the mask exposed pattern and the DLW pattern was tested using a mask having a series of rectangles regularly spaced over 25 mm. The rectangles were connected by both horizontal and vertical lines by DLW. Following development, the patterns were imaged by SEM and the position of the laser augmented lines was measured with respect to the center of the target rectangles.
The fabrication of the interdigitated electrode (IDE) began with an inexpensive transparency mask that had 250 µm wide contact leads that came within 1.3 mm of one another. In the gap between the leads, the IDE comprised of 2 µm lines was defined in S1813 by DLW. Following development, the samples were placed in a thermal evaporator (Edwards) and 2 nm of Cr was evaporated followed by 50 nm of Au. Liftoff was performed by soaking the sample in acetone for 1 h.
The microfluidic designed to trap an array of cells was also made using a Mylar mask with a pattern similar to an IDE. A Si wafer was spin coated with SU-8F, prebaked, and exposed through this mask. Next, a series of lines were drawn perpendicular to the “IDE” pattern. After post-baking and developing, this master structure was molded in PDMS . After curing, the PDMS was cut off the master, holes were punched, and the PDMS mold and a glass slide were treated in a plasma cleaner (Harrick) for 1 minute and then bonded together in a 110°C oven for 10 minutes . 1 mm diameter tubes were then fitted into the holes and a dilute solution of S. cerevisia cells or 5 μm silica beads were flowed into the device and images were captured on an inverted microscope.
3. Results and discussion
3.1 The LAMP method is dependent on the photochromism of the photoresist
When exposed to UV radiation, both S1813 and SU-8F have photochromic properties resulting in contrast between the exposed and unexposed regions. This contrast enables the DLW system to be aligned with the mask exposed patterns and is critical for LAMP. The photos in Fig. 2 show both SU-8F and S1813 before and after UV exposure. SU-8 generates a photoacid upon exposure, but shows no color or refractive index change. To visualize where the pattern had been exposed, we doped SU-8 with various concentrations of fluorescein, which is a pH indicator. We found that at low concentration of fluorescein the SU-8F was slightly more yellow in the exposed regions but there was not enough contrast to see the difference between exposed and unexposed regions. If the fluorescein concentration were too high, then the developed sample would not adhere well to the substrate, presumably because the fluorescein interrupted the epoxide polymeric network. We found 1 mg/mL to be ideal for providing enough contrast while maintaining the material properties of SU-8. The SU-8F, which is clear and colorless, turns bright yellow upon exposure due to the generation of acid and subsequent protonation of the fluorescein molecule . The S1813 can be seen to photobleach upon exposure turning from clear orange to clear colorless. To enhance this color change we used a blue LED as an illumination source on our DLW microscope.
3.2 Sub-micron wide lines can be made by DLW
Using a 20 × 0.75 NA objective, sub-micron features can be created by our DLW system. Figure 3 shows electron micrographs of some typical lines for S1813 and for SU-8F. The narrowest lines we could reproducibly write were 780 ± 140 nm wide for S1813 and 950 ± 90 nm for SU-8F. Other speeds were tested for both photoresists ranging from 5 - 50 μm/s. Faster speeds gave smaller lines that were irreproducible and that sometimes did not develop completely down to the substrate (S1813) or resulted in wavy lines that partially delaminated (SU-8F).
3.3 Patterns can be registered by DLW to within ~2 µm on existing patterns
When performing DLW, the first step is to register the position of existing features that were made during the mask exposure. This is possible because of the contrast between the exposed and unexposed regions and because the DLW system is itself a microscope where we can directly image the sample while simultaneously exposing it. We tested how accurately and precisely we could position the laser beam on the DLW system using a test pattern thatcontained dozens of rectangles spanning the length of the mask. By drawing lines from the center of one rectangle to the next (Figs. 4(a), 4(b)) and measuring the distance of the lines from the center point, we obtained measurements for Δx (horizontal) or Δy (vertical). Figure 4 shows a schematic of a portion of the mask and how we define Δx and Δy. Δx (or Δy) is calculated by measuring the distance of the line to both edges of the rectangle then the difference between these numbers is divided by two (Fig. 4(c)). If the line is perfectly centered then both distances to the edge will be the same and Δx will be 0. We define the radius, Δr, within which we can position the laser focal point as Δr = (Δx2 + Δy2)½. Hundreds of lines were measured on more than a dozen different wafers to give and an average Δr of 1.6 ± 1.4 μm for S1813 and 2.2 ± 1.5 μm for SU-8F. These numbers are reasonable given the accuracy and precision of the alignment of the transparency mask. We believe the Δr is slightly smaller for S1813 because its contrast is more pronounced than that of SU-8F, making it easier to pinpoint the edge of an exposed area. The use of confocal microscopy would likely decrease Δr, but would add to the complexity of the system. Since the use of a mask in LAMP is intended to pattern large features quickly, it is likely the case that registering smaller features to within ~2 µm is sufficiently accurate; if it is not, then more intermediate sized features can be made by DLW to better marry the large scale pattern to the smaller scale.
3.4 Proof of principle structures
3.4.1 Interdigitated electrode
Interdigitated electrodes (IDEs) are planar capacitors with high surface area and are useful for electrochemical impedance spectroscopy . The IDE we fabricated by LAMP occupies an area of approximately 200 × 200 µm and took only a couple of minutes to pattern. The IDE was connected to leads that were made by a mask exposure. Had these leads been made with the laser it would have taken hours to expose them even if the speeds and powers were doubled. Following LAMP, the IDE sample was gold coated, liftoff was performed, and the IDE was imaged by reflectance microscopy (Fig. 5(a)). The inset image in Fig. 5(a) shows that the individual lines are 3 µm wide, 200 µm long, and spaced 12.5 µm apart.
3.4.2 Cell trapping microfluidic
Another application of LAMP using the negative tone resist SU-8F is for the creation of microfluidic masters. A cell trapping array was fabricated by using a mask for an IDE-like pattern and the laser was used to define 2 µm channels between the prongs (Fig. 5(b)). Following development and PDMS molding, the mold was bonded to a glass slide. 5 µm silica beads were flowed in an aqueous solution through the channels and were trapped at the intersection points between the laser drawn channels and the larger mask patterned channels (Fig. 5(d)). The master could be used repeatedly to generate new microfluidic molds. We used another mold to trap S. cerevisia cells in a similar device (Fig. 5(c)). Devices that create arrays like this could be useful for multiplexed single-cell assays, for size sorting particles or cells, or for mechanical deformability assays of cells . The benefit of LAMP is that making these types of devices is very quick because the variable features, which are written with the laser, can be made in minutes and the lead-in channels, which are made with the mask, need not change between prototypes. Thus LAMP enables prototyping with the “fail fast” philosophy to quickly troubleshoot a design before committing to an expensive mask for mass production.
4. Summary and conclusion
In summary, we demonstrated a new hybrid lithography technique that combines conventional mask-based UV lithography with DLW to compensate for the drawbacks of each. Using Laser Augmented Microlithographic Patterning, LAMP, we showed that sub-micron laser written features could be registered to larger mask patterns to within ~2 µm in both positive- and negative-tone photoresist. We demonstrated LAMP by making an interdigitated electrode and a microfluidic device. We hope that others will use DLW systems to augment conventional lithography to increase the speed and efficiency with which they can generate their samples.
Bard Research Fund; Bard Summer Research Institute.
References and links
2. Y. Xia and G. M. Whitesides, “Soft Lithography,” Angew. Chem. Int. Ed. 37(5), 550–575 (1998). [CrossRef]
4. M. Gersborg-Hansen, L. H. Thamdrup, A. Mironov, and A. Kristensen, “Combined electron beam and UV lithography in SU-8,” Microelectron. Eng. 84(5-8), 1058–1061 (2007). [CrossRef]
5. L. H. D. Skjolding, G. T. Teixidor, J. Emnéus, and L. Montelius, “Negative UV–NIL (NUV–NIL) – A mix-and-match NIL and UV strategy for realisation of nano- and micrometre structures,” Microelectron. Eng. 86(4-6), 654–656 (2009). [CrossRef]
6. C. Eschenbaum, D. Großmann, K. Dopf, S. Kettlitz, T. Bocksrocker, S. Valouch, and U. Lemmer, “Hybrid lithography: Combining UV-exposure and two photon direct laser writing,” Opt. Express 21(24), 29921–29926 (2013). [CrossRef] [PubMed]
8. C. N. LaFratta, O. Simoska, I. Pelse, S. Weng, and M. Ingram, “A convenient direct laser writing system for the creation of microfluidic masters,” Microfluid. Nanofluid. 19(2), 419–426 (2015). [CrossRef]
9. C. N. LaFratta, T. Baldacchini, R. A. Farrer, J. T. Fourkas, M. C. Teich, B. E. A. Saleh, and M. J. Naughton, “Replication of two-photon-polymerized structures with extremely high aspect ratios and large overhangs,” J. Phys. Chem. B 108(31), 11256–11258 (2004). [CrossRef]
10. D. C. Duffy, J. C. McDonald, O. J. Schueller, and G. M. Whitesides, “Rapid prototyping of microfluidic systems in poly(dimethylsiloxane),” Anal. Chem. 70(23), 4974–4984 (1998). [CrossRef] [PubMed]
11. R. Sjöback, J. Nygren, and M. Kubista, “Absorption and fluorescence properties of fluorescein,” Spectrochim. Acta A Mol. Biomol. Spectrosc. 51(6), L7–L21 (1995). [CrossRef]
12. R. Ohno, H. Ohnuki, H. Wang, T. Yokoyama, H. Endo, D. Tsuya, and M. Izumi, “Electrochemical impedance spectroscopy biosensor with interdigitated electrode for detection of human immunoglobulin A,” Biosens. Bioelectron. 40(1), 422–426 (2013). [CrossRef] [PubMed]
13. D. Wlodkowic, S. Faley, M. Zagnoni, J. P. Wikswo, and J. M. Cooper, “Microfluidic single-cell array cytometry for the analysis of tumor apoptosis,” Anal. Chem. 81(13), 5517–5523 (2009). [CrossRef] [PubMed]