Lithographic studies of chemically-amplified

resist using soft x-rays from dense plasma focus

(DPF) device and by infrared spectroscopy
T. L. Tan, R. Verma, R. S. Rawat, P. Lee, S. V. Springham, D. Wong

Invited talk by Assoc Prof. Augustine Tan T L

International Workshop on Plasma Diagnostics & Applications
Singapore, 2-3 July 2009
 Introduction

Chemically-amplified resist (CAR) such as the negative SU-8 is well suited for
micromachining applications.

The resist has vertical sidewalls, very high aspect ratios up to 100:1. It is chemically
inert, thermal resistant, and mechanically stable. It is sensitive to ultraviolet, x-rays,
proton and electron beams.

The short wavelength and high penetration power of x-rays allow microstructures with
submicron resolution and high aspect ratio. No focusing optics are needed.

Compared to the production of x-rays from synchrotron, dense plasma focus (DPF)
produces x-rays of comparable intensity. DPF is cheap, compact, and easily made.

Fourier transform infrared (FTIR) spectroscopy is useful in monitoring the cross-

linking process in SU-8 by studying absorption peaks at 914 cm-1 and 1128 cm-1.
 Objectives of this research
To study the dependency of thermal-initiated cross-linking in 25-um thick SU-8 on
baking temperatures from 95 to 160 degC.

To study the cross-linking process in SU-8 under different post-exposure bake (PEB)
temperatures (65 to 120 degC).

To optimize the exposure dose of soft x-rays (SXR) from DPF and the PEB times and
temperatures for constructing SU-8 test structures with good resolution and aspect
ratio.

Lithographic process
steps
 Experimental method
O O O O

O O O O
Chemical
structure of
SU-8 H 3C
CH3
H 3C
CH 3
H 3C
CH 3
H 3C
CH3

O O O O

O O O O

SU-8 2010 formulation from MicroChem Corp was used. It is a polymeric epoxy resin dissolved in
an organic solvent, gamma-butyrolactone (GBL). A triaryl sulfonium salt as a photoacid generator
(PAG) is added. The resist has 8 epoxy ring groups in each molecule; hence the name is SU-8.

During exposure, the PAG is converted to a strong acid which is thermally activated during post-
exposure bake to act as catalyst in cross-linking reaction.

In cross-linking reaction, the epoxide rings (CH2CH-O-H2C) in SU-8 are disintegrated to ether
linkages (C-O-C) which is insoluble in the developer.

The extent of cross-linking reactions can be detected by the disappearance of infrared peak at
914 cm-1 of epoxy ring vibrational mode.
Spin-coating and thickness
measurements of resist layer

2717 2232
500 rpm

2638 2123
1000 rpm
2674 2014
A 1500 rpm
2609 2087
2000 rpm

3391 2529
2254
2500 rpm

4000 3600 3200 2800 2400 2000 1600

Wavenumber (cm-1)
The SU-8 resist was spin-coated onto a silicon wafer for 50 s, followed by 15-min
soft-bake at 95 degC.

The infrared spectrum of SU-8 shows oscillating patterns due to interference of

multiple reflection. Thickness of resist, t = no. of oscillations/[2n(v1-v2)] where n =
1.67 (refractive index of SU-8)
 Spin-coating and thickness
measurements of resist layer

30

25
Resist Thickness/m

20

15

10

0
0 500 1000 1500 2000 2500 3000
Spin Speed (rpm)

The spin speed of 500 rpm for 50 s gave resist thickness of 25 um. Error = 5 %
 Thermal-initiated Cross-linking
1128 914

95 C

110 C

120 C

A
130 C

140 C

150 C

160 C

1200 1100 1000 900 800

Wavenumber (cm-1)

Resist samples (25 um) which are unexposed were heated on a hot plate for 15 min from 95 to
160 degC to check for thermal induced cross-linking.

As temperature increases, the intensity of infrared peak at 914 cm-1 (epoxy ring mode)
decreases, while that for 1128 cm-1 (C-O stretch) increases, showing that cross-linking is
happening.
 Thermal-initiated Cross-linking
0.8

0.7
Intensity(914 cm-1)/
Intensity(1608 cm-1) 0.6
or extent of cross-
linking 0.5

0.4
90 110 130 150 170
Temperature/
degC

The extent of thermal-initiated cross-linking was quantified by the ratio of the infrared peak
intensity at 914 cm-1 to peak intensity at 1608 cm-1 (aromatic ring C-C stretch mode which is
unchanged during cross-linking reaction). Cross-linking started to occur at 120 degC.

To reduce the risk of thermally induced unexposed cross-linking, baking temperatures were set
below 120 degC.

At elevated temperatures, cross-linking caused by increased diffusion rate of acid resulted in

lower resolution of line width in lithography.
 Soft x-rays exposure of SU-8 resist
0.12

0.10

0.08
Transmission

0.06
10um
0.04 15um
25um
0.02

0.00
0.8 1.0 1.2 1.4 1.6
Energy (keV)

Soft x-rays (SXR) were emitted from a dense plasma focus (DPF) device (NX2) which operates
with a pulsed capacitive discharge where a magnetically compressed plasma is produced at the
end of the two coaxial electrodes. NX2 is a 400 kA, 15 kV Mather-type. Neon plasma of 7 mbar
was used. An aluminised-mylar filter is used to shield off visible and uv radiation. Magnets are
used to deflect any ion flux. Repetitive firing for neon-filled DPF up to 3000 shots was used.

If the beam line transmission is 0.5, 1 kW of SXR power would deliver 100 mJ/cm2 in 2 seconds,
a typical dosage for SU-8 resist. For 10-um resist, SXR (1 keV) dosage of 250 mJ/cm2 was
needed for lithographic work. At 1.1 keV, transmission was 0.1. For 25-um resist, dosage of 2500
mJ/cm2 was needed.
 Post-exposure Bake (PEB)
0.55

Extent of Cross-Linking
0.50

0.45

0.40

0.35

0.30
60 70 80 90 100 110 120 130
PEB Temperature (C)

After exposure to SXR at 2500 mJ/cm2, the resists were baked at temperatures form 65 to 120
degC to initiate cross-linking.

Extent of cross-linking increases with increasing temperatures as the diffusion rate of the
photogenerated acid responsible for catalyzing the cross-linking reaction increases.

To prevent resist cracking, an initial bake at 65 degC for 5 min, followed by ramping of 20
degC/min to the final desired temperature for constant baking for 5 min, and then gradual cooling
at 2 degC/min to room temperature.
 Resist resolutions using test mask

a b c

SEM micrographs of (a) gold mesh, and SU-8 resist at PEB of (b) 110 degC and (c) 95 degC

Gold mesh of 5 um diameter and 11 um x 11 um square grid was used as test masks [SEM
micrograph (a)].

After SXR exposure at 2500 mJ/cm2, PEB temperature at 110 degC was applied. The resist was
then developed using cyclohexanon for 5 min. The SEM of the resist pattern (b) showed that the
square structures were 13 um wide. The high PEB temperature increased the rate of acid diffusion
and caused it to flow to the unexposed regions of the resist, reducing the resolution.

At PEB temperature of 95 degC, the width of the square structure [SEM micrograph(c)] was 11
um, a direct imprint of the gold mesh pattern. Good resolution of line widths was obtained.
 Optimum lithographic conditions for 25-um thick SU-8

Spin- coat at 500 rpm for 50 s to get thickness of 25 um

Soft-bake at 95 degC for 15 min

SXR exposure at dosage of 2500 mJ/cm2 with x-ray mask on resist

Post-exposure bake (PEB):

(a) Pre-bake at 65 degC for 5 min,

(b) Ramp up to 95 degC for 1.5 min,

(c) Bake at 95 degC for 5 min and

(d) Cool to room temperature at 2 degC/min

Develop in developer: cyclohexanon for 5 min without agitation; final rinse

with fresh developer.
 Results

The optimum conditions were applied to SU-8 resist of 25-um thick using a x-ray mask (0.5 um
thick Au absorption layer on 100 nm thick Si3N4 membrane, fabricated using electron-beam
lithography).

Aspect ratio of 20:1 with good resolution was obtained.

Sharp edges in the SU-8 structures showed that the cross-linking process was optimised.

Fourier transform infrared spectroscopy (FTIR) was an effective tool in monitoring the cross-
linking process.
 Further studies

To study the limit to which the neon SXRs (0.9-1.6 keV) are able to penetrate
through thicker resist layers up to 100 um for micromachine applications. Harder x-
rays of 2 to 5 keV are needed.

The yield for Ar SXRs for NX2 is about 100 times less compared to Ne SXRs. The
transmission of Ar SXRs (about 3 keV) is an order of magnitude larger for a 10 um
thick SU-8 resist than that of Ne SXRs (about 1 keV). A further study to optimise the
x-rays from Ar using different inserts that can produce appreciable amounts of
harder x-rays or with some admixture of gases. Thicker resist can then be fabricated
for micromachining uses.

Improvement of the maintenance cycle and SXR power will make the plasma focus a
viable and attractive commercial source for x-ray lithographic applications.
 Thank you

For further communication:

Assoc. Prof. Augustine Tan T L
augustine.tan@nie.edu.sg
Tel: 65-6790 3872
Fax: 65-6896 9446
National Institute of Education
1, Nanyang Walk, Singapore 637616
Singapore