Journal of Micromechanics and Microengineering

TECHNICAL NOTE

Enhanced adhesive strength between SU-8 photoresist and titanium

substrate by an improved anodic oxidation method for high aspect-ratio
microstructures
To cite this article: Liyan Lai et al 2019 J. Micromech. Microeng. 29 047002

View the article online for updates and enhancements.

This content was downloaded from IP address 144.173.37.102 on 04/11/2019 at 11:49

 Journal of Micromechanics and Microengineering

J. Micromech. Microeng. 29 (2019) 047002 (10pp) https://doi.org/10.1088/1361-6439/aaffe0

Technical Note

Enhanced adhesive strength between SU-8

photoresist and titanium substrate by an
improved anodic oxidation method for high
aspect-ratio microstructures
Liyan Lai1 , Yunna Sun1 , Hao Wu2, Zhuoqing Yang1,3 and Guifu Ding1,3
1
  National Key Laboratory of Science and Technology on Micro/Nano Fabrication, School of Electronic
Information and Electrical Engineering, Shanghai Jiao Tong University, Shanghai 200240,
People’s Republic of China
2
  Huaihai Industrial Group Co., Ltd, Changzhi 046012, People’s Republic of China

E-mail: [email protected] (G Ding) and [email protected] (Z Yang)

Received 7 September 2018, revised 10 January 2019

Accepted for publication 18 January 2019
Published 22 February 2019

Abstract
This paper presents an anodic oxidation treatment method for improving the adhesive strength
between SU-8 photoresist and titanium substrate. The adhesion strengths were measured for
the treated titanium substrates by different anodic oxidation techniques, which increased with
the enhancement of the surface roughness and thorny structures on the treated substrate. It
was found that the adhesion was increased by approximately 86% through anodic oxidation
treatment in 3.0 mol l−1 NaOH solution compared to Si wafer substrate. Additionally, the
influence of UV dosage and hard bake temperature on the adhesion strength was examined
for the anodic oxidation treated substrates. The UV dosage effect, with a range from 0.8
to 2.08 J cm−2, was measured and analyzed for a 300 µm thickness SU-8 cylinder, and the
adhesion strength corresponding to the optimal dosage (1.52 J cm−2) was 45.82  ±  3.08 MPa.
The adhesion was further strengthened by a hard bake at 110 °C for 60 min and its value
was up to 71.64 MPa, which almost approaches the breaking strength of the SU-8 polymer.
Using the optimized anodic oxidation and UV lithography conditions, high aspect ratio
(24) SU-8 micropillar arrays with 600 µm height were fabricated without any distortion,
collapse or delamination. Finally, to evaluate the feasibility of the anodic oxidation treated
titanium substrate to serve as a seed layer, the electrochemical properties were measured by
polarization curves.

Keywords: SU-8, titanium substrate, anodic oxidation method, adhesion strength,

high aspect-ratio microstructure

(Some figures may appear in colour only in the online journal)

3
Authors to whom any correspondence should be addressed.

1361-6439/19/047002+10$33.00 1 © 2019 IOP Publishing Ltd  Printed in the UK

J. Micromech. Microeng. 29 (2019) 047002 Technical Note

1. Introduction qualitatively characterized by scanning electron microscopy

(SEM) and atomic force microscopy (AFM). Based on the
SU-8 is widely used as a photoresist for high aspect ratio AOM-treated substrate, SU-8 micropillar arrays with HAR
(HAR) microstructures in microelectromechanical systems are fabricated without any distortion, collapse and delamina-
(MEMS), such as a mold for electroplating or as a structural tion. Finally, the electrochemical properties were measured
material for MEMS devices. The conventional lithogra- by polarization curves to evaluate the reliability of the anodic
phy sources of the HAR microstructure’s fabrication are oxidation of a titanium substrate serving as a seed layer.
deep x-ray, e-beam and two-photo lithography [1–4]. In
compariso­n, UV lithography has been found to significantly
2. Experimental
reduce cost and eliminate the requirement for complex light
sources. Generally, glass and Si wafer are the conventional 2.1.  Substrate preparation
substrates for UV lithography. However, processing still faces
many challenges: the most significant being the poor adhesion The substrates used for this work included Si wafers and pure
of glass and Si wafer substrates to SU-8, which may result Ti plates. The Si wafer was cleaned according to a conven-
in partial or complete delamination of the SU-8 structure and tional method described in the literature [17]. Commercial
may restrict the fabrication of the HAR microstructure with a pure Ti plates with 76.2 mm diameter and 1 mm thickness
small contact area. In addition, when the subsequent process were prepared. Prior to anodizing, the surface of the Ti
is electroplating, these conventional substrates cannot be used plates were mechanically polished on 1000, 3000, 5000, and
as a seed layer. Therefore, to overcome these problems, many 10 000-grit silicon carbide papers, respectively, followed by
researchers have studied combining SU-8 and metal seed lay- ultrasonic cleaning in distilled water and ethanol for 10 min,
ers on a silicon substrate [5–8]. For example, Barber et  al and then immersion in 0.5 mol l−1 NaOH aqueous solution
[9, 10] utilized an adhesion promoter and different seed layer at room temperature for 2 h to remove surface organic resi-
combinations to improve the adhesion between the substrate dues. After these treatments, the Ti wafers were rinsed with
and SU-8 polymer, both through simulation and experiment. distilled water and air-dried. All wafers were treated by the
In addition, the optimization of SU-8 processing conditions anodic oxidation technique with 3 wt.% H2O2 and NaOH
has been extensively studied. Anhoj et al [11] investigated the solutions (0.5, 1.5, 3.0 and 5.0 mol l−1, labeled as Ti1, Ti2, Ti3
influence of a soft baking temperature on the polymerization and Ti4, respectively) as the electrolyte and platinum mesh
and the cracking of the negative photoresist SU-8. Hwang electrodes as the cathode. Equal surface areas of the anode
et al [12] studied, both theoretically and experimentally, the and the cathode were used for the experiment. According
effect of a Si wafer cleaning method on the adhesion behavior to a number of optimization experiments, the anodic oxida-
of SU-8 microstructures. Carlier et al [13] developed a simple tion treatment process was carried out at a current density of
technique where a thin layer of SU-8 2002 was first coated, 0.5 A dm−2 and a voltage of 10 V. Considering that H2O2 will
allowed to dry and then followed by a thick SU-8 layer on a Si decompose in strong alkali conditions and high temperature
wafer. It has been proved that the bi-layer technique enhances [18], all anodization experiments were conducted at 60 °C for
the adhesion of SU-8 to the substrate. Kim et al [14] improved 15 min with a stirring speed of 300 rpm. Following the above
the adhesion by controlling exposure dosage, soft baking time treatment, a pre-bake process was carried out on a hotplate at
and selecting an adhesion promoting layer. 60 °C for 60 min to dehydrate the substrates. To remove the
The most important characteristic highlighted by these surface organic residues, the treated Ti wafers were cleaned
studies is the adhesion of the SU-8 to a given substrate or seed using reactive ion etching (RIE) with oxygen for 10 min. The
layer. The good adhesion of SU-8 to its substrate will provide surface microtopography of the treated Ti was characterized
an excellent template for electroforming or microstructures. by SEM (ULTRA55, Zeiss, Japan) and AFM (Dimension
However, depositing a metal layer as a seed layer on a glass Icon & FastScan Bio, Bruker, USA). The electrochemical
or Si wafer by magnetron sputtering is expensive and has high properties of the treated Ti substrate, as the seed layer, was
processing requirements. In addition, the adhesion strength demonstrated by polarization curves with a CHI660D electro-
will be affected by the UV lithography conditions. To our chemical workstation.
knowledge, there are very few studies that have reported a
substrate with superior adhesion strength, and which serves
2.2.  SU-8 Processing
as a seed layer simultaneously, and many problems of the
SU-8 pattern, such as distortion, collapse and even total peel- Two types of SU-8 resist 1070 and 1075 (SU-8 from
ing, remain unsolved [15, 16]. Therefore, it is necessary to MicroChem Corporation, MA, USA) were used in this study.
develop a suitable substrate possessing both characteristics, The processing flow of the SU-8 is schematically illustrated
at low cost, for MEMS device fabrication and its applications. in figure 1. The SU-8 process consists of spin coating, soft
In this work, an anodic oxidation method (AOM) for a baking, UV exposure, a post-exposure bake, development
titanium substrate was developed for a HAR microstruc- and a hard bake. Different thicknesses of the SU-8 resist were
ture. In addition, the effects of UV dosage and hard bake spun on the AOM-treated substrates and the Si wafers; the
temperature on the adhesion strength between the SU-8 and coated wafers were placed on a leveled surface for 15 min
the AOM-treated substrates were investigated. The adhe- to allow the SU-8 to spread out. Then, the wafers were soft
sion was quantitatively analyzed with the bonding tester and baked on a leveled hot plate surface where the temperature

2
J. Micromech. Microeng. 29 (2019) 047002 Technical Note

Figure 1.  Processing steps for the SU-8 resist.

Figure 2.  SEM (50.00 K  ×) and AFM ((scanning area 10 µm  ×  10 µm) images of the treated titanium substrate surfaces, (a), (c), (e) and
(g) and (b), (d), (f) and (h) represent the Ti1, Ti2, Ti3 and Ti4 substrates, respectively. (Ti1) 0.5 mol l−1 NaOH, 3% wt. H2O2, 60 °C, 15 min,
(Ti2) 1.5 mol l−1 NaOH, 3% wt. H2O2, 60 °C, 15 min, (Ti3) 3.0 mol l−1 NaOH, 3% wt. H2O2, 60 °C, 15 min; (Ti4) 5.0 mol l−1 NaOH, 3%
wt. H2O2, 60 °C.

was increased stepwise. The UV exposure was performed an increase in NaOH solution concentration. Meanwhile, the
by an SUSS MA6 Mask Aligner UV. The exposure intensity surface area and roughness, as well as the thorny structures of
was 8.0 mW cm−2 at 365 nm wavelength. Post-exposure bak- the treated titanium, greatly increased. When the concentra-
ing was carried out on the hot plate. Afterwards, the samples tion of NaOH solution was lower than 3.0 mol l−1, the surface
were slowly cooled and kept at room temperature for 8 h to of the titanium sheet remained flat and smooth but became
release the residual stress. The samples were developed at irregular at 5.0 mol l−1. At low concentrations, the morphology
the same temperature. Following development, the patterned of the Ti1 specimen manifested fine pores; the Ra, Rq and the
wafers were briefly rinsed with isopropyl alcohol and then surface area were 92.5 nm, 72 nm and 144.0 µm2, respectively
naturally dried in an airing chamber. To further enhance the (figures 2(a) and (b)). Compared with Ti1, the Ra, Rq and the
adhesion, it is necessary to perform a hard baking process. surface area value of Ti2 increased to 122 nm, 93.1 nm and
All the above samples were prepared for adhesion tests and 165.9 µm2 respectively. A uniform porous pattern is observed
the adhesion quality of the SU-8 microstructures was also on the surface of the Ti2 sample (figures 2(c) and (d)). As the
characterized by SEM. concentration of NaOH solution was increased to 3.0 mol l−1,
the titanium oxide thin film of Ti3 was loose, porous and
more uniform (figures 2(e) and (f)). The surface roughness
3.  Results and discussion
Ra significantly increased to 171 nm, and was covered with
numerous fine summits, which were evenly distributed all
3.1.  Surface characterization
over the surface. As the concentration was further increased
Figure 2 shows selected SEM and AFM images of samples to 5.0 mol l−1, the surface of the titanium sheet formed a fine
Ti1, Ti2, Ti3 and Ti4 in different concentrations of electro- groove with a hollow because of the anodic corrosion (figures
lytes. The surface roughness details are given in table 1. The 2(g) and (h)). The surface was dissolved in a high concentra-
pore sizes and porosity of TiOx thin films become larger with tion of NaOH solution with H2O2 at 60 °C of the electrolyte

3
J. Micromech. Microeng. 29 (2019) 047002 Technical Note

Table 1.  The roughness parameters calculated from AFM images of

treated titanium surfaces.
Surface
Sample Ra (nm) Rq (nm) area (µm2)
Ti1 92.5 72 144.0
Ti2 122 93.1 165.9
Ti3 208 149 190.9
Ti4 171 138 177.9

temperatures. Therefore, the selected concentration of NaOH

solution was not higher than 3.0 mol l−1.

3.2.  Adhesion strength of SU-8 to the substrate

The adhesion strength was quantitatively measured by a PTR-

1101 bonding tester (RHESCA, Japan), as seen in figure  3.
The adhesion test specimens were optimized in a cylinder with
600 µm diameter and 300 µm thickness for various substrates
and SU-8 lithography. The substrates were mechanically fixed
by the work-holder. In the shear test, a SU-8 cylinder speci- Figure 3.  The PTR-1101 bonding tester.
men was transversally pushed by the shear test tool, and the
load value was continuously measured until the SU-8 cylinder specimens have no residual SU-8 pattern and only a superfi-
specimen was stripped from the substrates. The bonding test cial remnant vestige was left after the adhesion test due to low
tool was operated at a speed of 0.1 µm min−1 until the SU-8 adhesion strength (figures 4(e) and (f)). Visually, this shows
patterns separated from the substrate surface, and the adhe- that the amount of residual SU-8 pattern was the largest for
sion strength, σ (in MPa), is defined as the breaking load per the Ti3 sample, which means that the adhesion strength of the
unit area, equation (1). Ti3 sample on the treated substrate is much higher than the
other samples; this is followed by Ti2, while the Ti1 sample
σ = F/A.
(1) has the lowest adhesion strength.
Where F is the breaking load (in N) between two surfaces For the quantitative analysis, we obtained the adhesion
which can be measured, and A is the cross-sectional area of strength profile using a bonding tester. The variation of force
the sample. The peeled interface was observed by SEM to (N) with the displacement is showed in figure  5. Table  2
evaluate the fracture patterns. records the adhesion strength values of the SU-8 pattern on
different substrates. It can be seen that the Si wafer has the
3.2.1. Influence of the different substrates on adhesion minimum force (6.11 MPa); this is due to the poor adhesion
strength.  In this section, the adhesion strength for a series of strength that causes the SU-8 pattern to be easily peeled away
SU-8 cylinders, with 300 µm thickness and 600 µm diameter, from the substrate. The strength of pure Ti was 18.57 MPa,
on different substrates, was measured while the UV lithog- which was higher than that of the Si wafer. Thus, the adhe-
raphy conditions remained constant. To examine the frac- sion strength of the pure Ti is stronger than Si wafer, which
ture behavior differences of the substrate types on adhesion is in agreement with previous studies. Furthermore, the adhe-
strength, after performing the shear test, the fracture surfaces sion strength of the AOM-treated titanium was superior to
were observed by SEM, as shown in figure  4. The images that of the pure Ti and Si wafer substrates, especially for the
were used to qualitatively classify the adhesion strength. Ti3 specimen. For Ti3, the adhesion strength reached a maxi-
Figure  4(a) shows the adhesion strength test sample of an mum value of 41.18 MPa, which corresponds to an increase of
SU-8 cylinder. After the adhesion test steps, we noticed that almost 85% compared to the Si wafer substrate. These results
some of the samples partially or totally peeled away from are consistent with the above qualitative test.
the substrate. Figures  4(b)–(f) show the fracture surface The mechanism of adhesion strength was enhanced with
images with different extents of peeloff, which respectively both chemical interactions and physical bonding. On the one
represent the Ti1, Ti2, Ti3, pure Ti and Si wafer substrates. hand, because different metals have different atomic structures
In figure 4(b), most SU-8 patterns were peeled from the Ti1 and properties, titanium ([Ar]3d24s2) has only two electrons
specimen and the residual SU-8 patterns have apparent tear- in its d orbitals, which makes it more susceptible to electron
ing marks. In comparison, for sample Ti2 (figure 4(c)), treated donation from the filled O p  orbitals. Contact between the
with 1.5 mol l−1 NaOH solution, about three fifths of the pat- polymer and the metal will produce a very strong separa-
terns were retained. Moreover, there was nearly no SU-8 pat- tion charge at the interface, which generates an electrostatic
tern peeling in figure 4(d), which showed a clear SU-8 layer attraction and van der Waals interactions. Ti has the smallest
on the peeled side, with the residual SU-8 pattern remain- electronegativity and the least outer layer electrons, as well as
ing on the treated substrate. In contrast, the pure Ti and Si the highest surface energy, compared to Au, Ni, Cr and Cu [5].

4
J. Micromech. Microeng. 29 (2019) 047002 Technical Note

Figure 4.  (a) The SU-8 cylinder of the shear test before testing, at 300 µm thickness and 600 µm diameter, (b)–(f) represent the fracture
surfaces of the SU-8 cylinder after the shear test for Ti1, Ti2, Ti3, pure Ti and Si, respectively.

Table 2.  Test results of adhesion strength, formula: σ (MPa)  =  F

(N)/A.
Pure
Sample Ti1 Ti2 Ti3 Si Ti
Interfacial 6.85 10.19 11.53 1.71 5.20
separation force (N)
Diameter (×10−4 m) 6 6 6 6 6
Adhesion strength 24.26 36.29 41.18 6.11 18.57
(MPa)

decrease the interaction energy and modulate each surface

energy component. Based on equation  (2) [23], a decrease
in the value γ12 and an increase in the value γ2 results in an
increase in Wa. As a result, the adhesion strength between
SU-8 and the treated titanium substrate has been enhanced.
Wa = γ1 + γ2 − γ12 .
(2)
Figure 5.  The shear test results for Ti1, Ti2, Ti3, pure Ti and Si.
Where γ1 and γ2 are the surface energies of the SU-8 and
substrate, and γ12 is the surface energy of the interface per
Thus, Ti has the strongest adhesion to SU-8. Moreover, many unit area once two materials are in contact. Wa is equal to
researchers have proved this point [5, 9, 12, 19–21]. However, the sum of two surface energies (the SU-8 and substrate) and
pure Ti has a compact passive film of TiO2 in air, which will deduction of their interaction energy, which represents the
reduce the adhesion to SU-8 on a Ti substrate [9]. The AOM minimum energy that must be applied to separate a unit area
method obtains a porous film on the treated substrate sur- of the interface.
face with a mixed amorphous/crystalline character, which is On the other hand, when SU-8 is spin-coated on a sub-
mainly composed of TiO2 and the sub-oxides Ti2O3 and TiO strate, a solid–liquid interface is formed and a thin liquid film
at the substrate interface [22]. The treated Ti substrate will be is present on the substrate. A good wetting of the substrate and
able to react with SU-8 because rich oxygen atoms in the seed SU-8 can achieve homogeneous and stable coatings [21]. After
layer participate in SU-8 cross-linking. With the introduction UV exposure and post baking, SU-8 undergoes a cross-linking
of oxygen vacancies, more and more electrostatic attractions reaction, the solvent content is reduced and the solid–liquid
as well as van der Waals interactions are formed, which will interface gradually changes to a solid–solid interface. As

5
J. Micromech. Microeng. 29 (2019) 047002 Technical Note

Figure 7.  The effect of exposure dosage (mJ/cm2) on bonding force

Figure 6.  The effect of the SU-8 pattern thickness on the bonding
force. (MPa) on Ti3 substrates, 365 i-line.

the increased surface roughness with more thorny structures Figure  7 shows the variation in the bonding force with the
enlarged the surface area it also improved the pinning effect UV dosage on the Ti3 substrate. All the test specimens were
on the surface layers [24–27]. The adhesive strength between an SU-8 cylinder with 300 µm thickness and 600 µm diam-
the SU-8 photoresist and substrate increases proportionally eter. The bonding force increased from 13.56  ±  2.52 MPa
to the roughness and surface area. In this case, separation of to 45.82  ±  3.08 MPa when the dosage was between 0.8 and
the two solids needs a stronger load to resist the frictional 1.52 J cm−2, then showed a non-linear descending trend.
force, which results in a strengthened mechanical bond being At optimal dosage, the value corresponds to an increase
formed. Based on the discussion above, one obvious conclu- of about 70% compared to the value at under exposure
sion is that the adhesion strength has been enhanced by the (0.8 J cm−2). This is due to the insufficient photo-initialization
improved AOM. to polymerize the SU-8 at a low dosage, especially at the bot-
To better understand the AOM-treated substrate, the thick- tom, which results in a lower adhesion strength. On the con-
ness of the SU-8 cylinder on different substrates was varied trary, over exposure could also weaken the adhesion, as shown
from 160 µm to 1050 µm while the diameter was 600 µm; the in figure  7. This might be because larger exposure dosages
UV lithography conditions remained constant. As shown in induced more cross-linking of SU-8, which results in higher
figure 6, it is clear that there is a trend for adhesion strength as internal stress, as reported in the literature. Furthermore, over
follows: Si wafer  <  Ti1  <  Ti2  <  Ti3, at the same SU-8 thick- exposure results in expanded structures for light divergence,
ness, which correlates with the increase in surface roughness. substrate reflection and diffusion of the photogenerated acid
For the Si substrate, the adhesion strength was between 5.33 [29, 30]. Therefore, it is crucial to choose the appropriate
and 13.85 MPa. With increasing SU-8 thickness between 470 exposure dosage for the adhesion strength. The appropriate
and 600 µm, the adhesion strength increased, then decreased dosage would provide valuable information and a good refer-
beyond 600 µm thickness. The descending rate slows down ence for the future fabrication of HAR microstructures.
and eventually reaches a plateau at 700 µm. These results
are in agreement with the reported values [28]. These results 3.2.3. Influence of the hard bake temperature on adhesion
might be explained by the fact that the applied light was strength.  To further enhance the adhesion strength, a hard
absorbed by the SU-8 resist, and the thicker this layer is, the bake process was applied to increase SU-8 cross-linking. Four
weaker the intensity of the light reaching the substrate will be. identical samples were hard baked at temperatures of 95, 110,
Apparently, the Ti1, Ti2 and Ti3 substrates show a trend simi- 120 and 135 °C for 60 min, respectively. Figure 8 shows the
lar to the Si substrate but the position of the strongest bonding results of the bonding force along with the SU-8 thickness at
force was different. This phenomenon occurred because dif- different hard bake temperatures. An SU-8 pattern with the
ferent substrates have different reflectivity. Moreover, the same thickness was used to analyze and compare the bonding
strongest bonding forces of the Ti1, Ti2 and Ti3 samples were force. The results indicate that the bonding force increased
27.81, 43.13 and 68.63 MPa, respectively, which are obvi- from 55.32 to 71.64 MPa when the hard bake temperature
ously higher than that of the Si substrate. Because the Ti3 ranged from 95 °C to 110 °C. The adhesion strength at 110 °C
sample had a superior bonding force with the SU-8, it was was about 53% higher than the ultimate strength reported [31]
used in the subsequent experiments. (34.0 MPa) and approached the breaking strength of SU-8
[6] (73.1 MPa). This is because hard baking could reduce
3.2.2. Influence of the exposure dosage on adhesion the cracks to some extent [28]. With the elevation of bak-
strength.  Another significant factor that can influence the ing temper­ature, the bonding force tended to decrease, with
adhesion strength of SU-8 coatings is the UV exposure dosage. values of 53.52 and 10.25 MPa for hard baking at 120 and

6
J. Micromech. Microeng. 29 (2019) 047002 Technical Note

array structures. Distortion, collapse and even total peeling off

the micropillar often appears during development. Moreover,
the bending and sticking of neighboring micropillars was
caused by the capillary force during solvent evaporation at
the drying step. Thus, SU-8 with a reliable adhesion strength
to the substrate is very important. To examine the ability and
reliability of the optimized AOM-treated substrate in future
applications, ultra-HAR SU-8 micropillar arrays were fabri-
cated on Ti3, and a Si wafer substrate for comparison. Figure 9
shows the SEM images of micropillar arrays obtained with
optimal UV lithography conditions. For the Si substrate, most
of the pillars were significantly delaminated during the devel-
opment (figure 9(a)) and the ones that survived ultimately
leaned towards each other, resulting in twisting and clamping
for the capillary force during the subsequent rinsing or drying
step (figure 9(b)). This was mainly caused by the weak adhe-
Figure 8.  The effect of different hard bake temperatures on the sion strength of the SU-8 to the Si wafer substrate. Notably,
bonding force. SU-8 micropillar arrays (aspect ratio  =  24) demonstrated
good reproducibility on the Ti3 substrate without any dist­
135 °C, respectively. Therefore, 110 °C was selected as the ortion, collapse or delamination (figure 9(c)). Furthermore,
optim­ized hard baking temperature. its enlarged SEM image exhibited high physical integrity
Figure 8 also reflects the influence of hard baking temper­ and sidewall quality, as shown in figure 9(d). The micropillar
ature on the cross-link density by evaluating the adhesion showed considerable smoothness with no T- profiles on the
strength. The relationship between the bonding force and the top or tapered at the bottom. The results indicated that the Ti3
thickness of the SU-8 from 200 to 420 µm is approximately substrate has strong adhesion strength, enough to resist the
linear when the hard baking temperatures were at 95, 110 rinsing liquid and the capillary force, resulting in a straight
and 120 °C. The value initially increased from 26.28 MPa at micropillar. Consequently, the optimized conditions are reli-
a thickness of 200 µm to 55.32 MPa at 420 µm for the hard able for the fabrication of the ultra-HAR microstructure.
baking temperature of 95 °C, but decreased at a thickness
above 420 µm. A similar variation was observed for the sam-
ple subjected to a 110 °C hard bake. However, a maximum 3.4.  Measurement of the electrochemical properties
for different substrates
bonding force was reached at a critical thickness for the bak-
ing temperature of 120 °C. This saturation may be caused by To evaluate the feasibility of the anodic oxidation of a Ti
the further polymerization and increased cross-link density substrate serving as a seed layer, the electrochemical proper-
due to insufficient exposure of the thickness of the SU-8 film, ties of the treated Ti substrate were measured by polarization
which was compensated at 120 °C. The cross-linked network curves. In fact, Au is the best choice for a seed layer because
of the SU-8 resist was built up based on the increased cross- of its high electrical conductivity and optical reflection. Thus,
link density [14, 28, 29, 32], which secures a strong bond deposited Au film was chosen for the comparison experiment.
between the SU-8 and its substrate, while at the 135 °C hard Figure  10 shows the linear sweep voltammogram curves of
baking temperature, the bonding force was very low. This was three substrates in 0.5 mol l−1 CuSO4 solution. It was obvi-
induced by the excessive cross-linking, which reduced the ous that the slope of the AOM-treated sample was close to the
viscosity of the SU-8 and increased the stress at the interface Au sample but significantly higher than the pure Ti sample.
between SU-8 and the substrates. At the optimal hard bak- The anode passivation current density of the AOM-treated
ing temperature, the solid–solid interface will soften and the titanium was significantly higher than pure Ti but lower than
SU-8 will produce penetration on the treated titanium surface Au at a low potential. In addition, the pure Ti surface was
layers [22, 23]. A real area of contact is further increased and easily covered with a compact passive titanium oxide layer,
the interface between SU-8 and substrate was much more which cannot be used as a seed layer for electroforming. This
stable with Ti3, which eventually results in further enhance- is because the pure Ti plate has the feature of a high affinity
ment of the adhesion. These results suggest that increasing the with oxygen and immediately forms a compact passive film of
hard baking temperature to a certain extent would lead to a TiO2 with a spontaneous reaction in the air [33–35]. The pas-
higher adhesion between SU-8 and the substrate for thin films. sive film acts as a reactivity barrier between the electrode and
Meanwhile, the temperature of 120 °C is suitable for thick electroplating bath. Although its thickness is no higher than
films with inadequate exposure dosage. only a few nanometers (thinner than 10 nm) [22], its resistivity
is extremely high due to the continuous compact film of TiO2.
3.3.  Application of the AOM-treated titanium substrate However, the original compact passive films could be broken
down by the AOM [34, 36, 37] and obtain a porous film on
The adhesion strength of the SU-8 is still a challenge for ultra- pure Ti plate surfaces in H2O2 containing electrolyte. The
HAR microstructure fabrication, especially for micropillar oxidation film has a mixed amorphous/crystalline character,

7
J. Micromech. Microeng. 29 (2019) 047002 Technical Note

Figure 9.  SEM images of micropillar arrays on Si wafer ((a) and (b)) and Ti3 substrate ((c) and (d)). (a) Collapsed SU-8 pillar arrays with
25 µm diameter and aspect ratio  >  10. (b) Clamped SU-8 pillar arrays with 20 µm diameter and aspect ratio  >  20. (c) SU-8 pillar arrays
with 24 µm diameter and aspect ratio  =  20. (d) An enlarged image of (c).

dielectric constant may occur with field strength as well [22].

Moreover, the AOM-treated substrate has a porous and loose
morphology; it has a higher surface area compared to the
passively oxidized titanium. As a result, the initial potential
with the passive range of Au and AOM substrates moved in
the noble direction compared to pure Ti substrate. The AOM-
treated titanium was available for serving as a seed layer in
HAR micromachining using an SU-8 photoresist.

4. Conclusions

A super adhesive Ti substrate was demonstrated for an SU-8

photoresist, which was treated by an anodic oxidization
method. The influence of the type of substrate, UV dosage,
as well as hard baking temperature on adhesion strength, was
analyzed. It was observed that the adhesion of the SU-8 photo-
resist to the treated titanium sheet was affected by the surface
Figure 10.  Linear sweep voltammogram curves of three materials
in CuSO4 solution. roughness and the thorny structures. The substrate treated
with NaOH solution at 3.0 mol l−1 had a superior surface
which mainly contains TiO2 and the sub-oxides Ti2O3 and TiO morphology, which provided the strongest adhesion strength
at the substrate interface, and the electronic conductivity of at a value of 41.18 MPa (significantly larger than that of Si
the ‘TiO2 layers’ increases by up to 14 orders of magnitude wafer, 6.11 MPa). Moreover, UV dosage affected the adhesion
with the introduction of oxygen vacancies; a large variation in strength significantly when the dosage was below the optical

8
J. Micromech. Microeng. 29 (2019) 047002 Technical Note

−2 [9] Barber R L, Ghantasala M K, Divan R, Mancini D C and

point of 1.52 J cm . The optimal hard baking temperature was
110 °C when the SU-8 thickness was less than 420 µm, and Harvey E C 2004 An investigation of SU-8 resist adhesion
in deep x-ray lithography of high-aspect-ratio structures
hard baking at 120 °C for 60 min could sufficiently strengthen
Proc. SPIE 5276 85–91
the underexposed SU-8 patterns. Furthermore, the crucial role [10] Barber R L, Ghantasala M K, Divan R, Mancini D C and
of adhesion strength for HAR micropillar fabrication was Harvey E C 2007 Study of stress and adhesion strength in
identified. Pillars with aspect ratios of 24 were successfully SU-8 resist layers on silicon substrate with different seed
revealed on the Ti3 substrate. From electrochemical property layers J. Micro/Nanolithogr. MEMS MOEMS 6 481–4
[11] Anhoj T A, Jorgensen A M, Zauner D A and Hübner J 2006
analysis, we deduced that the AOM-treated titanium was suit-
The effect of soft bake temperature on the
able for serving as a seed layer. To conclude, the AOM-treated polymerization of SU-8 photoresist J. Micromech.
substrate represents a good seed layer for substrates with an Microeng. 16 1819–24
excellent adhesion, and the optimal SU-8 processing condi- [12] Hwang K-Y, Park C-S, Kim J-H, Suh K-Y, Cho E-C and
tions can be used for the fabrication of HAR microstructures. Huh N 2010 The effects of adhesion energy on the
fabrication of high-aspect-ratio SU-8 microstructures
J. Micromech. Microeng. 20 117001
Acknowledgments [13] 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
The authors would like to thank support from the Shanghai spectrometry analysis J. Micromech. Microeng.
Professional Technical Service Platform for Non-Silicon 14 619–24
Micro-Nano Integrated Manufacturing (16DZ2290103). This [14] Kim S J, Yang H, Kim K, Lim Y T and Pyo H B 2010 Study of
work was supported in part by the National Natural Science SU-8 to make a Ni master-mold: Adhesion, sidewall profile,
Foundation of China (No. 61571287), the Advanced Research and removal Electrophoresis 27 3284–96
[15] Agirregabiria M, Blanco F J, Berganzo J, Arroyo M T,
Ministry of Education Joint Foundation (6141A020224) and Fullaondo A, Mayora K and Ruano-Lopez J M 2005
the SJTU-SMC Young Scholar Program (16X100080044). Fabrication of SU-8 multilayer microstructures based
on successive CMOS compatible adhesive bonding and
releasing steps Lab Chip 5 545–52
ORCID iDs [16] Yang R and Wang W 2005 A numerical and experimental
study on gap compensation and wavelength selection
Liyan Lai https://orcid.org/0000-0002-8268-0665 in UV-lithography of ultra-high aspect ratio SU-8
Yunna Sun https://orcid.org/0000-0001-7393-2827 microstructures Sensors Actuators B 110 279–88
[17] Chen Y-T and Lee D 2007 A bonding technique using
hydrophilic SU-8 J. Micromech. Microeng. 17 1978–84
References [18] Sun F, Xie C X, Zhang F, Jin M P and Wang Z G 2012 Effect
of pH value on adiabatic decomposition characteristics of
hydrogen peroxide Chem. Eng. 40 42–5
[1] Auzelyte V, Elfman M, Kristiansson P, Nilsson C, Pallon J, [19] Liu J, Cai B, Zhu J, Ding G, Zhao X, Yang C and Chen D
Marrero N A and Wegdén M 2006 Exposure parameters 2004 Process research of high aspect ratio microstructure
for MeV proton beam writing on SU-8 Microelectron. Eng. using SU-8 resist Microsyst. Technol. 10 265–8
83 2015–20 [20] Lu C H, Yin X F and Wang M 2007 Fabrication of high
[2] Bogdanov A L and Peredkov S S 2000 Use of SU-8 aspect ratio metallic microstructures on ITO glass substrate
photoresist for very high aspect ratio x-ray lithography using reverse-side exposure of SU-8 Sensors Actuators A
Microelectron. Eng. 53 493–6 136 412–6
[3] Juodkazis S, Mizeikis V, Seet K K, Miwa M and Misawa H [21] del Campo A and Greiner C 2007 SU-8: a photoresist for high-
2005 Two-photon lithography of nanorods in SU-8 aspect-ratio and 3D submicron lithography J. Micromech.
photoresist Nanotechnology 16 846–9 Microeng. 17 R81–95
[4] Yang Z Q, Shi J H, Yao J Y, Zhang X J, Ding G F and [22] Marsh J and Gorse D 1998 A photoelectrochemical and
Zhao X L 2018 A laterally driven MEMS inertial switch ac impedance study of anodic titanium oxide films
with double-layer suspended springs for improving Electrochim. Acta 43 659–70
single-axis sensitivity IEEE Trans. Compon. Pack. Man. [23] Zhao Y P, Wang L S and Yu T X 2003 Mechanics of adhesion
8 1845–54 in MEMS—a review J. Adhes. Sci. Technol. 17 519–46
[5] Dai W, Lian K and Wang W 2005 A quantitative study on [24] Hu Y Q, Zhao Y P and Yu T X 2008 Tensile tests of micro
the adhesion property of cured SU-8 on various metallic anchors anodically bonded between Pyrex glass and
surfaces Microsyst. Technol. 11 526–34 aluminum thin film coated on silicon wafer Microelectron.
[6] Feng R and Farris R J 2003 2003 Influence of processing Reliab. 48 1720–3
conditions on the thermal and mechanical properties of [25] Hu Y Q, Zhao Y P and Yu T X 2008 Formation of dendritic
SU8 negative photoresist coatings J. Micromech. Microeng. nanostructures in Pyrex glass anodically bonded to silicon
13 80 coated with an aluminum thin film Mater. Sci. Eng. A
[7] Zhang X, Du L, Zhao M and Song C 2015 Investigation in the 483–4 611–6
interface adhesion strength between SU-8 photoresist and [26] Bhushan B 2003 Adhesion and stiction: Mechanisms,
metal substrate by ultrasonic treatment Microsyst. Technol. measurement techniques, and methods for reduction J. Vac.
22 2425–33 Sci. Technol. B 21 2262–96
[8] Nordström M, Johansson A, Noguerón E S, Clausen B, [27] Williams J D 2004 Study on the postbaking process and
Calleja M and Boisen A 2005 Investigation of the bond the effects on UV lithography of high aspect ratio SU-8
strength between the photo-sensitive polymer SU-8 and microstructures J. Micro/Nanolithogr. MEMS MOEMS
gold Microelectron. Eng. 78–9 152–7 3 563

9
J. Micromech. Microeng. 29 (2019) 047002 Technical Note

[28] Morikaku T, Kaibara Y, Inoue M, Miura T, Suzuki T, Oohira F, [33] Diamanti M V, Bolzoni F, Ormellese M, Perez-Rosales E A
Inoue S and Namazu T 2013 Influences of pretreatment and Pedeferri M P 2010 Characterisation of titanium oxide
and hard baking on the mechanical reliability of SU-8 films by potentiodynamic polarisation and electrochemical
microstructures J. Micromech. Microeng. 23 105016 impedance spectroscopy Corros. Eng. Sci. Technol.
[29] Chan-Park M B, Zhang J, Yan Y and Yue C Y 2004 45 428–34
Fabrication of large SU-8 mold with high aspect ratio [34] Fadl-Allah S A, El-Sherief R M and Badawy W A 2008
microchannels by UV exposure dose reduction Sensors Electrochemical formation and characterization of
Actuators B 101 175–82 porous titania (TiO(2)) films on Ti J. Appl. Electrochem.
[30] Li B 2005 Low-stress ultra-thick SU-8 UV photolithography 38 1459–66
process for MEMS J. Micro/Nanolithogr. MEMS MOEMS [35] Flis J and Ziomek-Moroz M 2008 Effect of carbon on stress
4 043008 corrosion cracking and anodic oxidation of iron in NaOH
[31] Dellmann L, Roth S, Beuret C and Racine G A 1998 solutions Corros. Sci. 50 1726–33
Fabrication process of high aspect ratio elastic structures for [36] Munirathinam B, Narayanan R and Neelakantan L 2016
piezoelectric motor applications Int. Conf. on Solid State Electrochemical and semiconducting properties of thin
Sensors and Actuators, 1997. Transducers ‘97 (Chicago) passive film formed on titanium in chloride medium at
vol 1 pp 641–4 various pH conditions Thin Solid Films 598 260–70
[32] Amato L, Keller S S, Heiskanen A, Dimaki M, Emnéus J, [37] Proost J, Vanhumbeeck J F and Van Overmeere Q 2009
Boisen A and Tenje M 2012 Fabrication of high-aspect ratio Instability of anodically formed TiO2 layers (revisited)
SU-8 micropillar arrays Microelectron. Eng. 98 483–7 Electrochim. Acta 55 350–7

10