Fabrication of multi-layer SU-8 microstructures

Home

Search

Collections

Journals

About

Contact us

My IOPscience

Fabrication of multi-layer SU-8 microstructures

This content has been downloaded from IOPscience. Please scroll down to see the full text. 2006 J. Micromech. Microeng. 16 276 (http://iopscience.iop.org/0960-1317/16/2/012) View the table of contents for this issue, or go to the journal homepage for more

Download details: IP Address: 138.38.64.230 This content was downloaded on 22/07/2016 at 09:26

Please note that terms and conditions apply.

INSTITUTE OF PHYSICS PUBLISHING

JOURNAL OF MICROMECHANICS AND MICROENGINEERING

doi:10.1088/0960-1317/16/2/012

J. Micromech. Microeng. 16 (2006) 276–284

Fabrication of multi-layer SU-8 microstructures Alvaro Mata1,2, Aaron J Fleischman2 and Shuvo Roy2 1 Department of Chemical and Biomedical Engineering, Cleveland State University, OH, USA 2 Department of Biomedical Engineering ND-20, The Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, OH 44195, USA

E-mail: [email protected]

Received 28 July 2005, in final form 23 September 2005 Published 9 January 2006 Online at stacks.iop.org/JMM/16/276 Abstract The fabrication of multi-level SU-8 microstructures using multiple coating and exposure steps and a single developing step has been achieved for up to six layers of SU-8. Alternating layers of SU-8 2010 (thin) and SU-8 2100 (thick) photoresist films were spin coated, followed by soft-bake, ultraviolet (UV) exposure and post-exposure bake steps. The multiple SU-8 layers were simultaneously developed to create patterned microstructures with overall thicknesses of up to 500 µm and minimum lateral feature size of 10 µm. The use of a single developing step facilitated fabrication of complex multi-level SU-8 microstructures that might be difficult, or even impossible, to achieve by sequential processing of multiple SU-8 layers that are individually coated, baked, exposed and developed.

1. Introduction In recent years, MEMS (microelectromechanical systems) devices have increasingly required the development of thick structures, often with high height-to-width aspect ratio [1–15]. These microstructures have enabled the fabrication of novel components for a variety of applications including micromixers, microelectrophoresis chips, micronozzles and lab-on-chip devices [5, 16] for microfluidic systems; microneedles [2] for transdermal drug delivery patches; micromolds [17, 18] for cell and tissue engineering substrates and various micromachines such as magnetic microactuators and micromotors, microtransformers and comb-drive actuators [8, 15, 19–21]. Traditionally, thick (>100 µm), highaspect-ratio microstructures have been developed using LIGA [12, 13], microstereolithography [14, 22] and deep reactive ion etching (DRIE) processes [23, 24]. However, these techniques often require complex and sophisticated equipment that are generally expensive, relatively slow and not readily accessible [5, 6, 11, 25]. The application of NANOTM SU-8 (MicroChem Corp., Newton, MA) photoresist to develop thick microstructures for MEMS has attracted great interest [5, 8, 25]. SU-8 is a high contrast, negative, epoxy-based line of near-ultraviolet (UV) radiation-sensitive photoresists with suitable chemical and mechanical properties, capable of developing thick photoresist 0960-1317/06/020276+09$30.00

structures in a single photolithographic step [25–27]. SU-8 photoresists have yielded aspect ratios of over 25 [28–32] and microstructures up to 1.5 mm high using a single coating step [5]. In addition, some research groups have reported gradient-height SU-8 microstructures through grayscale lithography [6], hollow SU-8 features using backside exposure [2] and configurations with tilted SU-8 features using inclined/rotated UV lithography [33]. Other investigators have achieved complex SU-8 microstructures through the combination of microstereolithography and UV lithography [22], the implementation of SU-8 as a sacrificial layer [34] and the use of antireflective coatings to better control UV exposure of SU-8 [4]. Another approach to develop complex SU-8 features is the formation of multi-level microstructures by performing multiple photolithographic steps of exposure and development [1, 35]. For example two- and threelevel microstructures have been achieved by spinning and exposing multiple layers of SU-8 sequentially followed by a simultaneous (single) developing step [2, 7, 36, 37]. This process has allowed for the fabrication of unique complex features while reducing the overall processing time. Recently, we utilized the fabrication of three-level SU-8 microstructures to create molds for tissue engineering scaffolds [38, 39]. These molds were fabricated by combining layers of SU-8 2010 and SU-8 2100, which are two recent SU-8 versions that

© 2006 IOP Publishing Ltd Printed in the UK

276

Fabrication of multi-layer SU-8 microstructures

10 µ µm holes column

150 µm hole

Figure 1. SEM image showing a three-level SU-8 microstructure similar to that used to create tissue engineering scaffolds with an overall height of ∼260 µm. The cross section (inset) shows the 300 µm diameter, 100 µm high columns and 200 µm diameter, 150 µm deep holes as well as 10 µm diameter and 10 µm deep holes.

are designed to produce thin (∼10–20 µm) and thick (∼100– 250 µm) films, respectively [26, 27]. The final resulting structure was 200–300 µm thick, which was defined primarily by the SU-8 2100 layers, while the SU-8 2010 layer made it possible to achieve a minimum feature size of ∼10 µm (figure 1). The use of multiple coatings and exposures coupled with a single development step to pattern additional SU-8 layers could increase three dimensionality and achieve even more complex microstructures with greater overall thickness while maintaining a fine minimum feature size. Despite its popularity, SU-8 photoresists pose multiple processing challenges that hinder reproducibility and limit potential applications [5, 10, 11]. Differences in exposure (exposure non-uniformities) throughout the photoresist layer can lead to partial cross-linking (curing) of the resist, residual stress build-up and poor adhesion to the underlying silicon wafer, which can subsequently result in feature degradation, crack-like distortions and lift-off of the entire SU-8 pattern [1, 2, 5, 7, 10, 31, 37]. Uniformity of exposure provided by an appropriate contact between the mask and photoresist is especially critical for thicker SU-8 layers and highly dependent on other processing steps [5, 32]. First, flowing of the uncrosslinked photoresist during the soft-bake step results in thickness variations of the SU-8 layer across the wafer [1, 5]. Second, formation of an edge bead during the spin coating procedure also produces differences in SU-8 thickness across the wafer. Third, thick or multiple SU-8 layers can increase residual stresses, crack-like distortions and subsequent bowing of the silicon wafer [1, 5, 25, 32]. In addition to exposure nonuniformities, components of the developing step also play an important role in the processing of SU-8 microstructures [25, 30, 32, 40]. Low-aspect-ratio features are generally insensitive to developing parameters, while high-aspect-ratio microstructures often require optimization of development time and the use of agitation [32]. However, vigorous agitation of the SU-8 Developer could result in the destruction of highaspect-ratio structures [30].

The increased interest in SU-8 photoresists has resulted in a number of innovative processing methods as reported by various research groups [2, 4–7, 22, 33, 34, 36, 37]. In order to advance the capabilities of MEMS fabrication technology to construct three-dimensional and complex microstructures with increasing thickness while maintaining fine minimum feature size, this paper reports on the development of a novel process to pattern several (more than three) SU-8 layers using multiple exposures and a single developing step. Our investigation is an extension of the fabrication process for three-level SU-8 microstructures, which were used as molds for tissue engineering applications [38, 39]. First, layers of thin (SU-8 2010) and thick (SU-8 2100) SU-8 photoresists of different thicknesses were coated, patterned and examined to verify the manufacturer’s (MicroChem Corp.) recommended protocols. Next, alternating layers of SU-8 2010 and SU-8 2100 were spin coated, followed by soft-bake, UV exposure and post-exposure bake steps. Finally, the multiple SU-8 layers were developed simultaneously. This multi-level process was refined to generate up to six-level SU-8 microstructures with overall thicknesses of up to 500 µm while maintaining a minimum feature size of 10 µm.

2. Materials and methods 2.1. Fabrication details SU-8 2010 and SU-8 2100 were used for the development of multi-level SU-8 microstructures. SU-8 2010 is designed for thin applications (∼10–20 µm) while SU-8 2100 is designed for thicker ones (∼100–250 µm) [26, 27]. Both photoresist versions were first patterned in single layers to verify protocols recommended by the manufacturer [26, 27]. The protocols for the single-layer SU-8 2010 and SU-8 2100 films were then used as a basis to develop a technique to combine and pattern multiple SU-8 layers. This technique consisted of processing up to six SU-8 layers by spin coating, soft bake, exposure and post-exposure bake of alternating SU-8 2010 and SU-8 2100 films, sequentially, and followed by simultaneous development of all layers. The resulting microstructures comprised up to six levels of complex assemblies of SU-8 features with a maximum overall thickness of up to 500 µm. Both single and multiple layer protocols were processed using different photomasks with layout configurations that incorporated various patterns (circles, squares and rectangles) with lateral dimensions ranging from 5 µm to 2 mm (figure 2). The choice of these particular patterns on the photomasks was determined by the design constraints of the ultimate application, which was the fabrication of scaffolds for tissue engineering. Each SU-8 2010 photoresist layer was exposed with either mask 1 or 2, while each SU-8 2100 layer was exposed with either mask 3, 4, 5 or 6. For the multi-level structures, patterns resulting from the photomask for each SU-8 layer were combined with others to produce complex features of the final SU-8 microstructure. 2.2. SU-8 2010 single layer SU-8 2010 was spin coated on a standard 100 mm diameter, (1 0 0)-oriented silicon wafer using a WS-400A-6NPP/Lite spinner (Laurell Technologies, North Wales, PA). After spin 277

A Mata et al Mask 2

Mask 1 5 µm

10 µm

Mask 3

10 µm diameter and separation

400 µm 200 µm

20 µm

30 µm

Mask 4

Mask 5

Mask 6 400 µm

300 µm

300 µm

2 mm

200 µm

300 µm

Figure 2. Schematic illustration of pattern layout on the six photomasks for the various SU-8 single and multi-level configurations. Dark regions correspond to chrome regions (opaque to UV) on the photomask, and light regions correspond to glass regions (transparent to UV).

coating, photolithography was performed using a VWR 400 hot plate (VWR Scientific Products, West Chester, PA), a C-005 convection oven (Lindberg/Blue M, Asheville, NC) and an MA4/6 IR contact mask aligner (Karl Suss Inc., Garching/Munich, Germany). Thickness measurement of the SU-8 films was performed using a JSM-5310 scanning electron microscope (SEM; JEOL USA, Peabody, MA). The key points of the processing protocol are outlined below. 1. Spin coat (a) Dispense SU-8 2010 photoresist (∼2 ml) to cover ∼2/3 of the stationary wafer surface, (b) 2000 revolutions per minute (rpm), 45 seconds (s), acceleration of 500 rpm s−1, (c) 200 rpm, 20 s, for edge bead removal (EBR), (d) EBR was performed by wetting an alpha-wipe (Texwipe, Upper Saddle River, NJ) with acetone (Mallinckrodt, Phillipsburg, NJ), and pressing it lightly on the edge of the spinning silicon wafer. 2. Soft bake (a) 65 ◦ C for 1 min on the hot plate, (b) 95 ◦ C for 5 min in the oven, (c) 10 min of cool down period (relaxation) at room temperature (∼25 ◦ C). 3. Exposure (a) Vacuum contact between the mask and the silicon wafer, (b) Masks 1 and 2 were used on different wafers (figure 2), (c) 365 nm wavelength UV, (d) 100 mJ (10.0 mW cm−2 for 10.0 s or 15.0 mW cm−2 for 6.7 s). 4. Post-exposure bake (a) 95 ◦ C for 5 min in the oven, (b) 10 min relaxation at room temperature. 5. Develop (a) Immersion in SU-8 Developer (MicroChem Corp.), (b) 15 min at room temperature inside a beaker using a stirring bar for slight agitation. 278

2.3. SU-8 2100 single layer SU-8 2100 was processed with the same equipment (spinner, hot plate, oven, aligner and SEM) as the SU-8 2010. The key points of the processing protocol are outlined below. 1. Spin coat (a) Dispense SU-8 2100 photoresist (∼2 ml) to cover ∼2/3 of the stationary wafer, (b) 1000, 2000, 3000 or 4000 rpm, 45 s, acceleration of 500 rpm s−1, (c) 200 rpm, 20 s for EBR, (d) EBR was performed similarly as for SU-8 2010. 2. Soft bake (a) 65 ◦ C for 5 min on hot plate, (b) 95 ◦ C for 55 min in the oven, (c) 10 min relaxation at room temperature. 3. Exposure (a) Vacuum contact between the mask and the silicon wafer, (b) Masks 3–6 were used on different wafers (figure 2), (c) 365 nm wavelength UV, (d) 300 mJ (15.0 mW cm−2 for 20.0 s) or 375 mJ (12.5 mW cm−2 for 30.0 s). 4. Post-exposure bake (a) 95 ◦ C for 30 min in the oven, (b) 10 min relaxation at room temperature. 5. Develop (a) Immersion in SU-8 Developer, (b) 20 min at room temperature inside a beaker using a stirring bar for slight agitation. A speed–thickness plot was also created for SU-8 2100 in order to generate appropriate spin parameters. Thickness measurements of the SU-8 2100 layers were performed by focusing on the bottom and top of developed SU-8 features through a calibrated Secolux 6 × 6 light microscope (Esselte Leitz GmbH, Stuttgart, Germany). The light microscope was attached to an ID-C225EB digital micrometer (Mitutuyo

Fabrication of multi-layer SU-8 microstructures

Table 1. Fabrication protocol for six SU-8 layers using a single developing step.

SU-8 Levels and Order of Processing

Layer SU-8 resist

Spin coat

Soft bake

Exposure

Post-exposure bake

Develop

65 °C for 1 min on plate 95 °C for 5 min in oven 10 min relaxation at 25 °C

Mask 1 6.7 s 15.0 mW cm-2

95 °C for 5 min on oven

No developing

1

SU-8 2010

2000 rpm for 45 s 200 rpm for 20 s 500 rpm s-1

65 °C for 5 min on plate 95 °C for 55 min in oven 10 min relaxation at 25 °C

Mask 4 20.0 s 15.0 mW cm-2

No developing

SU-8 2100

2000 rpm for 45 s 200 rpm for 20 s 500 rpm s-1

95 °C for 30 min on oven

2

65 °C for 1 min on plate 95 °C for 7 min in oven 10 min relaxation at 25 °C

Mask 1 6.7 s 15.0 mW cm-2

No developing

SU-8 2010

2000 rpm for 45 s 200 rpm for 20 s 500 rpm s-1

95 °C for 7 min on oven

3

65 °C for 5 min on plate 95 °C for 60 min in oven 10 min relaxation at 25 °C

Mask 5 20.0 s 15.0 mW cm-2

No developing

SU-8 2100

2000 rpm for 45 s 200 rpm for 20 s 500 rpm s-1

95 °C for 30 min on oven

4

65 °C for 1 min on plate 95 °C for 12 min in oven 10 min relaxation at 25 °C

Mask 1 6.7 s 15.0 mW cm-2

No developing

SU-8 2010

2000 rpm for 45 s 200 rpm for 20 s 500 rpm s-1

95 °C for 14 min on oven

5

SU-8 2100

2000 rpm for 45 s 200 rpm for 20 s 500 rpm s-1

65 °C for 5 min on plate 95 °C for 60 min in oven 10 min relaxation at 25 °C

Mask 6 20.0 s 15.0 mW cm-2

95 °C for 35 min on oven

6

80 min Agitation with stirring bar

All exposures performed in contact vacuum mode. EBR: Edge bead removal performed in similar manner as described for each individual layer.

Corp., Aurora, IL), which was displaced as the microscope moved from a focused position on the top to a focused position at the bottom of the SU-8 features. This technique could reliably measure thickness within ∼ ±5%. Features were measured at nine different locations around the wafer in a region spanning 3 mm from the edge to the center of the wafer. The accuracy of thickness measurements was subsequently verified using SEM.

(a)

(b)

(c)

2.4. Multi-level SU-8 2100 and SU-8 2010 SU-8 2100 and SU-8 2010 were combined to create microstructures of up to six levels using a single photolithographic developing step. The coating/exposure/ bake procedure was performed in a sequential manner whereby each subsequent SU-8 layer was coated on top of a previous one. In this manner, a thicker layer (SU-8 2100) would be ‘sandwiched’ between two thinner (SU-8 2010) layers, or vice versa. All SU-8 levels were subsequently developed simultaneously (table 1 and figure 3). This single developing step process provides important advantages over a process where each coated layer is developed prior to coating of subsequent layers. For example, simultaneous development of multiple SU-8 layers considerably reduces the processing time. In addition, coating uniformity is increased compared to a process that coats over the topography of previously patterned layers.

(d)

(e)

(f)

(g)

3. Results and discussion 3.1. Single layers of SU-8 2010 and SU-8 2100 Figure 4 presents a graph of SU-8 2100 thickness as a function of spin speed. At 1000 rpm, the spin-coated SU-8 2100 was thickest at 225 µm. As the spin speed increased, the SU-8 film became thinner, with the biggest reduction taking place

Figure 3. Cross-sectional schematic diagrams of the six SU-8 layer process depicting coating, baking, exposure and post-exposure baking of SU-8 2010 (a); SU-8 2100 (b); SU-8 2010 (c); SU-8 2100 (d); SU-8 2010 (e) and SU-8 2100 ( f ). All six layers were developed simultaneously (g). SU-8 Developer dissolves all the areas that were not exposed (blocked by dark regions of the photomasks) to UV light during each of the exposure steps (a)–( f ). Features are not drawn to scale.

279

A Mata et al

speed) that are 30 and 5 µm in diameter with aspect ratio of up to ∼3:1 (height:width). SU-8 2100 was exposed using masks 3–6, and produced features varying in thickness from 64 µm (4000 rpm) to 226 µm (1000 rpm) with a ∼16:1 maximum aspect ratio (figures 5(c) and (d)). Thin and thick SU-8 microstructures with high aspect ratio have been previously reported [5, 25, 30, 32]. A reproducible process that allows the fabrication of multi-level SU-8 microstructures using a single photolithographic developing step would enhance feature complexity while decreasing overall processing time. 3.2. Multi-level SU-8 2010 and SU-8 2100

Figure 4. Graph of speed versus thickness for SU-8 2100 spin coated at 1000, 2000, 3000 and 4000 rpm. SU-8 thickness decreased considerably from 226 µm (1000 rpm) to 64 µm (4000 rpm).

from 1000 to 2000 rpm (225 µm and 115 µm, respectively). At 3000 and 4000 rpm, the SU-8 layer thickness decreased further down to 80 and 64 µm, respectively. These values are generally consistent with those reported by the manufacturer [26, 27]. The thicker spin-coated SU-8 films were relatively more sensitive to flow due to gravity. Consequently, if the wafer is not horizontal during soft bake, there is increased likelihood of thickness non-uniformity due to SU-8 flow [5, 41]. Accordingly, the films produced at lower spin speeds exhibited greater thickness variations (±34% for 1000 rpm) across the wafer compared to films produced at higher spin speeds (±4–5% for 2000–4000 rpm). The protocols for both SU-8 2010 and SU-8 2100 produced features with vertical sidewalls and of similar size and shape to the features of the photomask. Figures 5(a) and (b) present SEM images of SU-8 2010 features patterned with mask 1 consisting of 13 µm deep holes (at 2000 rpm spin

Figure 6 presents two-level features from a wafer coated with both SU-8 2100 and SU-8 2010. Figure 6(a) presents an SU-8 2100 layer (coated at 1000 rpm) patterned with mask 3 underlying an SU-8 2010 layer (coated at 2200 rpm), which was exposed using mask 2. This protocol resulted in an SU-8 microstructure comprising 10 µm high posts with 10 µm diameter and 10 µm separation along with 200 µm deep holes that are 200 µm in diameter and with 400 µm separation. Figure 6(b) also presents a two-level SU-8 microstructure constructed from an SU-8 2100 layer (coated at 1000 rpm) patterned with mask 4 underlying an SU-8 2010 layer (coated at 1800 rpm), which was patterned with mask 1. This protocol produced an SU-8 microstructure comprising 15 µm deep holes with 30 µm diameter and 30 µm separation along with 200 µm deep holes that are 300 µm diameter and with 300 µm separation. Close examination of figure 6(b) reveals that the simultaneous developing step removed the unexposed photoresist from the second (top) SU-8 layer (exposed using mask 1) as well as the underlying unexposed photoresist from the first (bottom) SU-8 layer (exposed using mask 4). It appears that the SU-8 Developer removed the unexposed

(b)

(a)

SU-8 2010

SU-8 2010

Si Si

10 µm

10 µm (c)

(d )

SU-8 2100

SU-8 2100

Si

100 µm

Si

100 µm

Figure 5. SEM images of patterned SU-8 films showing various features including: 13 µm deep holes (a) and (b) patterned in SU-8 2010 with mask 1, which are 30 and 5 µm in diameter with aspect ratio of up to 3:1 (height:width); and 300 µm diameter and 200 µm deep holes (c) patterned in SU-8 2100 with mask 4. SU-8 2100 features exhibited 16:1 maximum aspect ratio (d).

280

Fabrication of multi-layer SU-8 microstructures

(a)

(b)

100 µm

300 µm

Figure 6. SEM images of two-level SU-8 microstructures processed using a single developing step showing: (a) an SU-8 2010 layer with 10 µm diameter and 10 µm high posts (patterned with mask 2) overlying an SU-8 2100 layer with 200 µm diameter and 200 µm deep holes (patterned with mask 3) (inset: close up view showing posts near the edge of a hole) and (b) an SU-8 2010 layer with 30 µm diameter and 15 µm deep holes (patterned with mask 1) overlying an SU-8 2100 layer with 300 µm diameter and 200 µm deep holes (patterned with mask 4) (inset: zoom out view showing suspended membrane and sacrificial material behavior of the unexposed SU-8 2100).

6 5 4 3 2 1 100 µm Si wafer

20 µm

100 µm

15 µm

Figure 7. SEM image of a six-layer SU-8 structure constructed from alternating SU-8 2010 (10 µm diameter and 10 µm high posts (inset)) and SU-8 2100 layers. Note the slight feature distortion in the regions where patterns on the corresponding photomasks for two adjacent layers overlap. Each layer is numbered in the order it was coated.

Figure 8. SEM image of a four-layer SU-8 structure showing 15 µm high posts on two SU-8 2010 layers overlying corresponding SU-8 2100 layers. The 10 µm diameter posts exhibit narrower bases in the second (upper inset) SU-8 2010 layer (∼5 µm diameter) compared to the first (lower inset) SU-8 2010 layer (∼7 µm diameter).

SU-8 from the second layer (SU-8 2010), which permitted subsequent penetration into the first layer (SU-8 2100). This infiltration of the SU-8 Developer removed the unexposed SU-2100, which therefore, acted as a sacrificial material. The exposure dosage for the second SU-8 layer (SU-8 2010) was not sufficient to fully cross-link the regions of the first SU-8 layer (SU-8 2100) that were previously blocked by the chrome (opaque) regions on mask 4. Qualitative observations revealed that the crack-like distortions, which are generally observed in thicker photoresist layers due to residual stresses, were not as noticeable on the second SU-8 layer. Figure 7 presents a SEM image of a six-level SU-8 microstructure constructed from SU-8 2010 and SU-8 2100 layers that were sequentially coated and exposed, but simultaneously developed as outlined in table 1. There were minor observable differences in feature quality between those on multiple SU-8 layers and those on corresponding single layers. Specifically, there was a slight narrowing of the 10 µm diameter posts at the base of upper SU-8 2010 layers (figure 8). In addition, there were slight distortions present in the regions where patterns on the various photomasks overlap (e.g. sides of the 300 µm columns in figure 7). Alignment of the photomask to the wafer was achieved using alignment marks on the photomask and corresponding features that were etched

into the silicon substrate. Examination of the final SU-8 microstructure revealed that features were aligned to within 5 µm between adjacent SU-8 layers. Similar to the twolevel SU-8 microstructures (figure 6), subsequent exposure of SU-8 2010 on the six-level microstructure did not fully cross-link the regions of the underlying SU-8 2100 layer that was previously blocked by the opaque regions of the corresponding photomask. Moreover, crack-like distortions were not observed on the upper layers of the multi-level SU-8 microstructure. The multi-level SU-8 process demonstrates that SU-8 can function as both structural and sacrificial materials [42]. This dual functionality is revealed using photomasks with patterns that are not necessarily designed to overlap. Such is the case for the microstructures shown in figure 6(a), where the second SU-8 layer (SU-8 2010 patterned with mask 2) consists of features that are not connected (freestanding 10 µm posts). In this case, removal of the unexposed regions on the first layer (SU-8 2100 patterned with mask 3) will automatically remove the overlying SU-8 2010 regions (whether crosslinked or not). However, if the features on an overlying layer (SU-8 2010) are connected such as in figure 6(b), then removal of the unexposed regions from the underlying layer (SU-8 2100) will not necessarily remove the exposed SU-8 from the 281

A Mata et al

overlying layer. Consequently, the underlying and overlying layers function as sacrificial and structural layers, respectively. Although the scope of our investigation was limited to the current photomask set, it is conceivable that this strategy could enable the fabrication of more common microstructures including cantilevers and bridges. It is interesting to note that the sacrificial material behavior was only observed for SU-8 microstructures where the SU-8 2010 was coated over SU-8 2100, and not for the reverse case (SU-8 2100 over SU-8 2010). The reason for this asymmetrical behavior is likely due to the fact that cross-linking of SU-8 2010 requires considerably less exposure dosage (∼100 mJ cm−2) compared to SU-8 2100 (∼300 mJ cm−2). When SU-8 2100 is exposed over previously unexposed regions of underlying SU-8 2010, the exposure dosage is so high that it fully cross-links the SU-8 2010. Overall, the processing of multiple SU-8 layers using a single developing step offers great potential to advance the capabilities of UV photolithography to construct threedimensional (3D) and complex structures, while maintaining dimensional precision and decreasing the total processing time. 3.3. Processing limitations and approaches The fabrication of the six-level microstructures required overcoming a number of challenges. These process complications resulted in feature alterations and can be classified into problems associated with the exposure or the developing steps. Feature geometry was highly sensitive to variations in exposure uniformity, which arose from SU-8 thickness non-uniformity as well as non-uniform contact between photomask and the SU-8 layer. Consequently, it was critical to ensure that the wafer was level (horizontal) during the initial soft-bake period, especially for thicker SU-8 layers (>100 µm). Furthermore, it was essential to minimize edge bead formation during spin coating of the SU-8 layers. Thickness non-uniformity arising from edge bead formation during spin coating increased with more SU-8 layers. Qualitatively, the thickness of the edge bead increased for any SU-8 layer when it was coated on the top of an already existent edge bead resulting from the previous SU-8 coating. Therefore, it was critical to remove the edge bead from each SU-8 layer prior to coating of the next SU-8 layer. The contact between photomask and SU-8 was also degraded by wafer bowing, which resulted from the residual stress buildup due to mismatch in thermal expansion coefficients between the Si wafer and the SU-8 layer. Wafer bowing increased with the number of SU-8 layers. The use of the aligner’s vacuum contact mode enhanced contact between photomask and SU-8, which, in turn, considerably improved subsequent feature resolution. However, vacuum contact mode increased the likelihood of contamination of the photomask with SU-8 residue from the wafer. The total thickness of the SU-8 microstructure was another important variable in exposure uniformity. In general, exposure non-uniformity increased with the height of the SU-8 layer above the surface of the silicon wafer. An example of this feature distortion is evident in figure 8 where 15 µm high posts are patterned on separate SU-8 2010 layers. In this case, the 10 µm diameter posts exhibit a more cone-like shape on 282

the second (upper) SU-8 2010 layer (∼5 µm diameter at the base) compared to the first SU-8 2010 layer (∼7 µm diameter at the base). This non-uniform exposure through the SU-8 layer could result from increased absorption of ultraviolet (UV) light with increasing thickness of the SU-8 layer, which, in turn, would result in intensity variations and associated differences in reflection characteristics from the underlying wafer surface during exposure [4, 29, 31, 43]. In addition to maximizing contact quality between the photomask and the wafer, optimization of the exposure dosage (power and time) would also improve exposure uniformity. Alignment of patterns on the photomasks to features on the wafer was not a critical issue for the designs in this investigation. However, misalignment of the photomasks could become a potential complication for other designs during the exposure step. When a photomask used for one exposure is not perfectly aligned with the one for the previous (underlying) SU-8 layer, distortions could result in the regions where both patterns overlap. This distortion is visible in figures 6(a) and 7, where some 10 µm posts near the edge of the holes on the underlying SU-8 were not completely removed. Crack-like distortions (wrinkles or cracks) that manifest during the post-exposure bake can result from underexposure of the SU-8 due to exposure non-uniformity or excessive residual stresses in the SU-8 [5, 29, 32, 37]. Nonetheless, it was qualitatively observed that the extent of crack-like distortions on the SU-8 surface decreased considerably on upper SU-8 layers. This decrease might be explained by the similarity in thermal expansion coefficients of adjacent SU-8 layers, which, in turn, redistributes or dissipates the residual stress build-up. Gradual stepping (to simulate ramping) of bake temperatures was critical to minimizing residual stresses and appearance of wafer bowing and crack-like distortions. The developing step also posed potential challenges to the fabrication of multi-level SU-8 microstructures. In this study, developing time at room temperature increased with the number of SU-8 layers ranging from ∼15 min for a single SU-8 2010 layer to ∼80 min for six SU-8 layers. However, this extended developing time did not noticeably limit feature quality. Agitation was also an important parameter during the developing step to ensure penetration of the SU-8 Developer into the narrower spaces in the lower layers of multi-level SU-8 microstructures. However, vigorous agitation could result in mechanical degradation of high-aspect-ratio features on the upper SU-8 layers [30]. Finally, temperature elevation to 30–35 ◦ C enabled superior feature development, while maintaining a low-to-medium agitation level to minimize feature destruction.

4. Conclusion SU-8 photoresists have enabled the fabrication of novel components for a variety of MEMS applications. The processing of multiple SU-8 layers using a single developing step offers great potential to advance the capabilities of MEMS fabrication technology to construct 3D and complex microstructures, while decreasing the overall processing time. A multiple layer technique to create SU-8 microstructures with up to six levels using a single developing step has been

Fabrication of multi-layer SU-8 microstructures

established. The protocol consists of a combination of SU8 2100 and SU-8 2010 photoresists that are sequentially processed and simultaneously developed. The resulting microstructures comprise features ranging from 5 µm to 2 mm in lateral dimensions and 10 to 500 µm in height. The use of a single developing step enables the fabrication of complex multi-level microstructures that may be difficult, or even impossible, to achieve by coating SU-8 layers over existing high-aspect-ratio topographies. This approach extends an inherently two-dimensional (2D) technology such as photolithography, to create complex 3D structures, while maintaining dimensional precision.

Acknowledgments The authors thank the following people for their assistance: Anna Dubnisheva, MS, of the Department of Biomedical Engineering, Lerner Research Institute, The Cleveland Clinic Foundation; Andrew Resnick, PhD, of the Department of Physiology and Biophysics at Case Western Reserve University; Derek Hansford, PhD and Nick Ferrell of The Ohio MicroMD Laboratory, Ohio State University and the Microfabrication Laboratory at Case Western Reserve University. This work was partially supported by a grant from Rockefeller Brothers Foundation.

References [1] Peterman M C, Huie P, Bloom D M and Fishman H A 2003 Building thick photoresist structures from the bottom up J. Micromech. Microeng. 13 380–2 [2] Kim K, Park D S, Lu H M, Che W, Kim K, Lee J-B and Ahn C H 2004 A tapered hollow metallic microneedle array using backside exposure of SU-8 J. Micromech. Microeng. 14 597–603 [3] Tay F E H, van Kan J A, Watt F and Choong W O 2001 A novel micro-machining method for the fabrication of thick-film SU-8 embedded micro-channels J. Micromech. Microeng. 11 27–32 [4] Chuang Y-J, Tseng F-G, Cheng J-H and Lin W-K 2003 A novel fabrication method of embedded micro-channels by using SU-8 thick-film photoresists Sensors Actuators A 103 64–9 [5] Lin C-H, Lee G-B, Chang B-W and Chang G-L 2002 A new fabrication process for ultra-thick microfluidic microstructures utilizing SU-8 photoresist J. Micromech. Microeng. 12 590–7 [6] Waits C M, Modafe A and Ghodssi R 2003 Investigation of gray-scale technology for large area 3D silicon MEMS structures J. Micromech. Microeng. 13 170–7 [7] Carlier J, Arscott S, Thomy V, Fourrier J C, Caron F, Camart J C, Druon C and Tabourier P 2004 Integrated microfluidics based on multi-layered SU-8 for mass spectrometry analysis J. Micromech. Microeng. 14 619–24 [8] O’Brien J, Hughes P J, Brunet M, O’Neil B, Alderman J, Lane B, O’Riordan A and O’Driscoll C 2001 Advanced photoresist technologies for Microsystems J. Micromech. Microeng. 11 353–8 [9] Yu H, Gruntzig A, Zhao Y, Sharon A, Li B and Zhang X 2004 Direct writing of 3D microstructures using a scanning laser system Materials Research Society Symposium Proceedings vol 782 pp A11.5.1–A11.5.6 [10] Chan-Park M B, Zhang J, Yan Y and Yue C Y 2004 Fabrication of large SU-8 mold with high aspect ratio microchannels by UV exposure dose reduction Sensors Actuators B 101 175–82

[11] Chuang Y-J, Tseng F-G and Lin W-K 2002 Reduction of diffraction effect of UV exposure on SU-8 negative thick photoresist by air gap elimination Microsyst. Technol. 8 308–13 [12] Malek C K and Saile V 2004 Applications of LIGA technology to precision manufacturing of high-aspect-ratio micro-components and -systems: a review Microelectron. J. 35 131–43 [13] Solf C, Janssen A, Mohr J, Ruzzu A and Wallrabe U 2004 Incorporating design rules into the LIGA technology applied to a Fourier transformation spectrometer Microsyst. Technol. 10 706–10 [14] Bertsch A, Jiguet S and Renaud P 2004 Microfabrication of ceramic components by microstereolithography J. Micromech. Microeng. 14 197–203 [15] Roy S, Furukawa S and Mehregany M 1996 Determination of Young’s modulus and residual stress of electroless nickel using test structures fabricated in a new surface micromachining process Microsyst. Technol. 2 92–6 [16] Thorsen T, Maerkl S J and Quake S R 2002 Microfluidic large-scale integration Science 298 580–4 [17] Motlagh D, Senyo S E, Desai T A and Russell B 2003 Microtextured substrata alter gene expression, protein localization and the shape of cardiac myocytes Biomaterials 24 2463–76 [18] Tan J L, Tien J, Pirone D M, Gray D S, Bhadriraju K and Chen C S 2003 Cells lying on a bed of microneedles: an approach to isolate mechanical force Proc. Natl Acad. Sci. 100 1484–9 [19] Roth S, Dellmann L, Racine G-A and de Rooij N F 1999 High aspect ratio UV photolithography for electroplated structures J. Micromech. Microeng. 9 105–8 [20] Brunet M, O’Donnell T, O’Brien J, McCloskey P and Mathuna S C O 2002 Thick photoresist development for the fabrication of high aspect ratio magnetic coils J. Micromech. Microeng. 12 444–9 [21] Williams J D and Wang W 2004 Microfabrication of an electromagnetic power relay using SU-8 based UV-LIGA technology Microsyst. Technol. 10 699–705 [22] Bertsch A, Lorenz H and Renaud P 1999 3D microfabrication by combining microstereolithography and thick resist UV lithography Sensors Actuators A 73 14–23 [23] Clerc P-A et al 1998 Advanced deep reactive ion etching: a versatile tool for microelectromechanical systems J. Micromech. Microeng. 8 272–8 [24] Hanein Y, Schabmueller C G J, Holman G, Lucke P, Denton D D and Bohringer K F 2003 High-aspect ratio submicrometer needles for intracellular applications J. Micromech. Microeng. 13 S91–5 [25] Lorenz H, Despont M, Fahrni N, Brugger J, Vettiger P and Renaud P 1998 High-aspect-ratio, ultrathick, negative-tone near-UV photoresist and its applications for MEMS Sensors Actuators A 64 33–9 [26] NANO SU-8 2000 Negative Tone Photoresist Formulations 2035–2100 (Newton, MA: MicroChem Corporation) [27] NANO SU-8 2000 Negative Tone Photoresist formulations 2002–2025 (Newton, MA: MicroChem Corporation) [28] Vora K D, Shew B Y, Harvey E C, Hayes J P and Peele A G 2005 Specification of mechanical support structures to prevent SU-8 stiction in high aspect ratio structures J. Micromech. Microeng. 15 978–83 [29] Jiang K, Lancaster M J, Llamas-Garro I and Jin P 2005 SU-8 Ka-band filter and its microfabrication J. Micromech. Microeng. 15 1522–6 [30] Williams J D and Wang W 2004 Using megasonic development of SU-8 to yield ultra-high aspect ratio microstructures with UV lithography Microsyst. Technol. 10 694–8 [31] Zhang J, Chan-Park M B and Conner S R 2004 Effect of exposure dose on the replication fidelity and profile of very high aspect ratio microchannels in SU-8 Lab on a Chip 4 646–53

283

A Mata et al

[32] Chang H-K and Kim Y-K 2000 UV-LIGA process for high aspect ratio structure using stress barrier and C-shaped etch hole Sensors Actuators A 84 342–50 [33] Han M, Lee W, Lee S-K and Lee S S 2004 3D microfabrication with inclined/rotated UV lithography Sensors Actuators A 111 14–20 [34] Seidemann V, Rabe J, Feldmann M and Buttgenbach S 2002 SU-8-micromechanical structures with in situ fabricated movable parts Microsyst. Technol. 8 348–50 [35] Tien J, Nelson C M and Chen C S 2002 Fabrication of aligned microstructures with a single elastomeric stamp Proc. Natl Acad. Sci. 99 1758–62 [36] Anderson J R, Chiu D T, Jackman R J, Cherniavskaya O, McDonald J C, Wu H, Whitesides S H and Whitesides G M 2000 Fabrication of topologically complex threedimensional microfluidic systems in PDMS by rapid prototyping Anal. Chem. 72 3158–64 [37] Bohl B, Steger R, Zengerle R and Koltay P 2005 Multi-layer SU-8 lift-off technology for microfluidic devices J. Micromech. Microeng. 15 1125–30 [38] Mata A, Fleischman A J and Roy S 2004 Microfabricated 3D scaffolds for tissue engineering applications Presented at

284

[39]

[40]

[41]

[42] [43]

MRS Fall Meeting: Nanoscale Materials Science in Biology and Medicine Mata A, Fleischman A J, Boehm C, Muschler G and Roy S 2005 Fabrication of 3D micro-textured scaffolds for tissue engineering Presented at ASPE 2005 Spring Topical Meeting: Precision Micro/Nano Scale Polymer Based Components and Device Fabrication (Columbus, OH) Zhang J, Tan K L, Hong G D, Yang L J and Gong H Q 2001 Polymerization optimization of SU-8 photoresist and its applications in microfluidic systems and MEMS J. Micromech. Microeng. 11 20–6 Blanco F J, Agirregabiria M, Garcia J, Berganzo J, Tijero M, Arroyo M T, Ruano J M, Aramburu I and Mayora K 2004 Novel three-dimensional embedded SU-8 microchannels fabricated using a low temperature full wafer adhesive bonding J. Micromech. Microeng. 14 1047–56 Chung C and Allen M 2005 Uncrosslinked SU-8 as a sacrificial material J. Micromech. Microeng. 15 N1–5 Engelke R, Engelmann G, Gruetzner G, Heinrich M, Kubenz M and Mischke H 2004 Complete 3D UV microfabrication technology on strongly sloping topography substrates using epoxy photoresist SU-8 Microelectron. Eng. 73–74 456–62