Reducing SU-8 hygroscopic swelling by ultrasonic treatment

1Key Laboratory for Micro/Nano Technology and System of Liaoning Province, Dalian University of Technology, Dalian 116024, China 2Key Laboratory for Precision & Non-traditional Machining Technology of Ministry of Education, Dalian University of Technology, Dalian 116024, China *Corresponding author. E-mail: duliqun@dlut.edu.cn Reducing SU-8 hygroscopic swelling by ultrasonic treatment Du Liqun1, 2,*, Zhang Benzhuang2 and Liu Yaping2

SU-8 resist has become a prevalent mould material to electroform MEMS devices because of its mechanical and chemical stability and its ability to produce high-aspect-ratio moulds by UV-LIGA technology [1,2]. However, electroforming processes based on SU-8 mould still face serious challenges. It is well known that the dimensions of electroformed structures are usually shrunk compared with the masks. This is mainly due to SU-8 mould distortions caused by thermal expansion and hygroscopic swelling during electroforming [3]. The relative dimensional errors of electroformed structures may reach 23% [4], which are unacceptable in practical applications, and the tapered structures cannot be corrected using a scaled or biased mask pattern. Furthermore, resist displacements will limit the maximum producible aspect ratio of a metal structure when the cavity of the mould will close at the top under the worst conditions [5].
Some works have been published to analyze PMMA swelling during electroforming. Two approaches have been proposed to reduce PMMA swelling. (i) Electroform at room temperature since the lower temperature may decelerate the solvent molecules diffusing rate throughout the resist thickness [6]. However, lower electroforming temperature induces higher internal stress in electroformed metal layer [7]. (ii) Improve the layout design of masks: auxiliary structures are introduced to Nano-Micro Lett. 2, 197-203 (2010) 198 Du Liqun et al DOI: 10.5101/nml.v2i3.p197-203 http://www.nmletters.org form trenches around the part in PMMA [8,9]. Then, the mass of the resist that swells is reduced. While auxiliary features can dramatically decrease tapers for linear structures, they increase skew for curved structures in some cases [5]. Additionally, complicated geometries pose a challenge for designing auxiliary features because the features must follow the perimeter of the part uniformly. Complex non uniform geometries of the auxiliary structures make it difficult to predict the expansion behaviors of the resists [10].
The above two methods are applicable to the processing based on SU-8, but apparently the same problems exist.
Compared with PMMA, the number of publications on SU-8 swelling is very limited. Experimental and simulative studies have been taken on the thermal swelling of SU-8 [3]. And the influences of different post-exposure bake (PEB) temperatures on the thermal swelling of SU-8 were investigated [4]. However, as far as we know, no effective method which is independent of processing parameter and mask pattern has been proposed to reduce SU-8 hygroscopic swelling during electroforming.
In this paper, a novel research for the effect of ultrasonic treatment on SU-8 swelling is presented. The ultrasonic treatment was introduced to the electroformed Ni-structure fabricating process. And the dimensional errors of electroformed structures fabricated with and without ultrasonic treatment were measured respectively. The experimental results are presented and the mechanism of ultrasonic effect on SU-8 swelling is discussed.

SU-8 mould distortions during electroforming
Electroformed structures based on resist moulds exhibit large dimensional errors. These errors result from thermal expansion and hygroscopic swelling of the resist since electroforming is performed in aqueous electrolyte at an elevated temperature [5,6]. Poission"s ratio is 0.22. The temperature load is 26℃ (from room temperature 24℃ to electroforming temperature 50℃).
The simulated result is shown in Fig. 2c. It is known that the top width of SU-8 only increases by 0.27 μm due to thermal expansion. In fact, moisture and thermal diffusion are two interactive processes. When the polymer absorbs external molecules its CET always changes. For SU-8, the value of CET will reduce after absorbing water molecules [11]. Thereby, the On the other hand, the experimental total width increases more than 9 μm (will be shown in Fig. 6, measuring point 1). The analysis results indicate that the dimensional error produced by thermal expansion is less than 3% in the total error.

Hygroscopic swelling
From above analysis, it is known that swelling is the predominant reason for SU-8 mould distortions. The mechanism of moisture diffusion in epoxy has been widely studied. And it has been proved that the hydroxyl groups of epoxy resins play the leading role in moisture uptake process, where water molecules can form strong hydrogen bounds [12,13]. In general, each SU-8 monomer molecule contains eight reactive epoxy groups, and therefore high degree of cross linking can be obtained and form three-dimensional network after photo-thermal activation. During cross linking reaction, hydroxyl groups generated. Hydroxyl groups have significant affinity to polar molecules such as water. Consequently, SU-8 could absorb a lot of water when exposed to aqueous surroundings, which causes swelling.
The water absorption process of epoxy can generally be expressed by Fick"s law [14], Where C is the concentration of water; τ is time; D is water diffusion coefficient; and x, y and z are axes along the concentration gradient.
D varies with temperature, and can be described by the classic Arrhenius function [15], Where D 0 is a constant; E D is the activation energy for diffusion; R and T represent the ideal gas constant and absolute temperature respectively.
The volume expansion with respect to the moisture content has been found to be linear to a good approximation for SU-8 [16]. Therefore, the hygroscopic strain ε induced by swelling can be related to the concentration of water according to C    Table 1.

Ultrasonic treatment
Ultrasonic treatment was performed by a self-designed ultrasonic device as shown in Fig. 3. Its vibrating frequency is 20 kHz. When the uncrosslinked regions of the resists disappear after development, SU-8 moulds become easily damaged.
Therefore the ultrasonic processing was carried out before rather than after development. After PEB, the SU-8 coated Ni-substrates were bolt fixed on the worktable, and then the ultrasonic energy was imposed to the SU-8 resist for 10min at the constant input power (125 W).

Dimension measurements of SU-8 moulds and electroformed Ni-structures
In order to distinguish the different effects of ultrasonic treatment on the mould distortions before electroforming and the swelling during electroforming, the top dimensions of SU-8

Results and discussion
Ultrasonic effect on development process   Fig.1), and the subscripts, "unultrasonic" and "ultrasonic", stand for the samples fabricated by the conventional method described in section 3.1 without ultrasonic treatment and the experimental samples which were subjected to ultrasonic treatment, respectively.
Compared with the mask, the dimensions of the resist moulds are slightly shrunk. This is expected to be a combined effect of diffraction during exposure [17] and swelling due to absorbing developer during development [18]. For the same measuring points of the non ultrasonic and the ultrasonic samples in Table 2, the widths of the cavities in SU-8 moulds are almost identical. This phenomenon reveals that ultrasonic treatment has little influence on the subsequent development process.    Table 3, where δ unultrasonic and δ ultrasonic are the dimensional errors of unultrasonic and ultrasonic samples respectively, and error decreasing: α=(δ unultrasonic -δ ultrasonic )/ δ unultrasonic .

Ultrasonic effect on electroforming process
Take measuring point 1 for example, ultrasonic treatment decreases the dimensional error of Ni-structure by 59.0%. This is a significant improvement to a MEMS device which always requires higher dimensional accuracy.

Simulation
If SU-8 is a simply freestanding film, and if the hygroscopic strain ε in SU-8 is uniform and isotropic, then all of the dimensions would simply grow by the magnitude of ε when SU-8 expands. However, the situations become complicated since the resist is bonded to a rigid substrate. Therefore, ANSYS was adopted to calculate the hygroscopic strain ω y along the direction perpendicular to the substrate surface under a certain hygroscopic strain ε, as shown in Fig. 7, where t 0 is the initial thickness of SU-8 film; Δt is increased value of thickness due to swelling; ω y =Δt/t 0 . The physical properties of SU-8 used in this model are introduced in section 2.2.
Finite element analysis reveals the relationship between ω y and hygroscopic strain ε of SU-8. The results are exhibited in Fig.   8.
For the large plane sheet sample which is confined by a substrate, the hygroscopic strain along the direction perpendicular to the substrate is for both t 0 =56.2 μm(for non-ultrasonic sample)and t 0 =52.7 μm (for ultrasonic sample). When the resist is freestanding, ω y =ε, as shown in Fig. 8.
During experiment, the thicknesses of SU-8 resist have been measured as soon as electroforming finished, Δt =4.7 μm for non ultrasonic sample and Δt =2.0 μm for ultrasonic sample.
As can be seen from Table 4, the hygroscopic strain ε of SU-8, after immersed in electroforming solution for 6.5 h, declined from 6.8% to 3.1% because of ultrasonic.
Furthermore, the hygroscopic strain ε calculated in this way can be used to predict swelling of SU-8 mould by ANSYS. The simulated results for measuring point 1 are presented in Fig.9, where "1-s-ultrasonic" and "1-s-unultrasonic" represent measuring point 1 for ultrasonic sample and non-ultrasonic sample, respectively.
Transient swelling occur throughout the electroforming process, and the convex electroformed metal layer limits lateral SU-8 swelling. Therefore, the experimental errors are lower than

Mechanism of ultrasonic effect on SU-8 swelling
It is known that the value of absorbed moisture in crosslinked epoxy resins is deeply depended on the quantity of hydroxyl groups, in other words, depended on the level of the hydrophilicity [12,13,19]. To deeply understand the ultrasonic effect on the hydrophilicity of SU-8, it is necessary to consider the change of contact angle after ultrasonic processing.
Therefore, the water contact angles of SU-8 were measured by Drop Shape Analysis System (DSA100, KRÜSS GmbH) at 50℃. The results show that the water contact angle of SU-8 increases from 70.8° (before ultrasonic processing) to 74.9° (after ultrasonic processing), which indicates that the hydrophilicity of SU-8 decreases while exposed to ultrasonic.
In view of water contact angle change, the mechanism of ultrasonic effect on SU-8 swelling can be explained as follows. It has been found out that ultrasonic can induce chemical bonds in polymers breaking [20]. For this study, when the samples are exposed to ultrasonic a part of hydroxyl groups may break away from the SU-8 backbones, as shown in Fig. 10. Thereby, the hydrophilicity of SU-8 reduces, which makes the amount of moisture absorbed in the same time decreases. As a result, the swelling potential of SU-8 mould declines and the dimensional errors of electroformed structures diminish. However, SU-8 still keeps three-dimensional networks and cannot be dissolved by SU-8 developer or electroforming solution.

Summary and prospect
1) The finite element analysis results indicate that the dimensional error produced by thermal expansion is less than 3% in the total error, so hygroscopic swelling is the predominant reason for SU-8 mould distortions.
2) The ultrasonic treatment was introduced to the electroformed Ni-structure fabricating process after PEB. For a 400 µm mask, ultrasonic treatment decreases the dimensional error of Ni-structure by more than 50% at 50 μm height.
3) Simulation of hygroscopic swelling is conducted by ANSYS, and the results indicate that the hygroscopic strain ε of SU-8 declined from 6.8% to 3.1% because of ultrasonic.
4) The increased water contact angles of cured SU-8 before (70.8°) and after (74.9°) ultrasonic processing indicates that the hydrophilicity of SU-8 decreases since exposed to ultrasonic.
5) The mechanism of the ultrasonic effect on SU-8 swelling is discussed. When the samples are exposed to ultrasonic a part of hydroxyl groups may break away from the SU-8 backbones, which decreases the hydrophilicity of SU-8 and in turn reduces the mould swelling.
Although this work reveals that ultrasonic treatment can reduce SU-8 hygroscopic swelling, further studies optimizing the ultrasonic process parameters such as ultrasonic frequency, acoustic power and exposure time are ongoing.