Fong-Cheng Tai
1;*, Shih-Chin Lee
1and Che-Hung Wei
21
Department of Materials Science and Engineering, National Cheng-Kung University, Tainan 701, Taiwan
2Department of Mechanical Engineering, TATUNG University, Taipei 104, Taiwan
The present work describes the hybrid layers consisting of 1-mm-thick hydrogenated diamond-like carbon (DLCH) film deposited by ion beam deposition (IBD) method as the bottom layer and spin coating polyimide (PI) film at a thickness of 6mmas the top layer. Optical microscopy shows that all DLCH film, PI film and PI/DLCH film have smooth surface and are free from cracks or wrinkles. Water contact angle measurement shows that the annealed DLCH film has the higher relative adhesive tension and adhesive work than that of PI film. Raman spectra show that the DLCH film can avoid severe graphitization & degradation at 300C during annealing treatment, and FTIR spectra also exhibit that
the PI film will reach 95% imidization during the curing step at 300C. PI film can act on DLCH film as the stress buffer. This is demonstrate
from warpage measurement showing that top PI film with tensile stress can partially reduce the compressive stress over bottom DLCH film with 6.7% warpage reduction. The electrical resistivity of PI/DLCH/Si is similar to that of PI/Si is resulted from the PI passivation effect. [doi:10.2320/matertrans.47.1869]
(Received February 13, 2006; Accepted June 14, 2006; Published July 15, 2006)
Keywords: hydrogenated diamond like carbon film, Polyimide film, Raman spectra, Fourier transform infrared spectrometer spectra
1. Introduction
It is well known that DLCH film has outstanding mechanical property (high hardness) and chemical property
(high chemical corrosion resistance).1)On the other hand, PI
film has good thermal property (high Tg) and chemical attack
durability due to five-fold type of imide ring.2) Generally,
DLCH film has two major drawbacks: poor adhesion due to higher compressive residual stress during deposition and worse thermal stability due to meta-stable amorphous structure. These can be solved with F or N or Si elements doping into pure DLCH film to form the complex a-C:H:X
film.3–6)Another solution is the annealed treatment on DLCH
film in order to get the transformation from amorphous to graphitic structure. However, such treatments will cause hardness significantly decreasing when annealed temperature
is above certain critical transition point.7–9) Other methods
use the additive layer under the DLCH film, such as the interlayer structure (DLCH/Ti/substrate), multilayer struc-ture (DLCH/TiCN/TiN/Ti/substrate) and special graded
structure to improve the DLCH adhesion.10) In addition,
many studies have been reported about the property in DLCH
over polymer structure.11–21) Several test results show that
DLCH/polymer can significantly reduce the oxygen gas penetration into polymeric film. From our knowledge, no study about spin coating polymer on DLCH structure is reported. The purpose of this work is to evaluate the thermal stability, buffer stress and interfacial adhesion on this PI/ DLCH/Si hybrid structure.
2. Experimental Details
DLCH film was deposited by ion beam deposition method using hydrocarbon gas on 6’’ (100)-oriented p-type silicon
wafer with electrical resistivity 30–60-cm (WAFER
WORKS CORP.). During the DLCH deposition, the base
pressure of the ion beam chamber reached 5104Pa
within 40 minutes by rotary pump and turbo pump, and the working pressure is 0.1 Pa. The hydrocarbon gas is used as the reactive gas, and the DLCH growth temperature was
controlled at 200C and the final deposited thickness was
1.0mm. To evaluate the cured degree of PI film, the liquid
polyimide precursor was synthesized with PMDA and ODA monomers. PMDA-ODA type of polyimide film was fab-ricated on bare Si and DLCH surface with spin-on method. The liquid PI precursor is dispensed at wafer center with the spin speed up to 600 r.p.m to spread the PI precursor over wafer surface within 10 seconds, and then the PI film is operated under 2500 r.p.m for 30 seconds, at last the PI/Si
wafer was cured at 300, 350 and 400C for one hour in the N
2 gas, respectively. The PI/DLCH wafer was also cured at
300C for one hour in the N
2gas. With cured treatment, the
liquid polyimide precursor film was converted into solid polyimide for better electronical, mechanical and chemical properties and the final thickness of cured polyimide film was
6.0mm.22)The chemical structure of PMDA-ODA polyimide
unit is shown in Fig. 1. This type of polyimide (C22H12N2O5)
consists of two major parts: one is the modified PMDA resin
(Pyro-mellitic dianhydride: C10H2O4) containing one
ben-zene ring, four C=O bonds, and two N atoms; the other is due
to modified ODA resin (Oxydianiline: C12H10N2O)
contain-ing two benzene rcontain-ings and one O atom.2)
The thickness of DLCH and PI films were measured by FE-SEM (Philips, XL-40FEG). Water contact angle (OP-TEM) was also measured by using Laplace-Young equation.
Raman spectroscope (488 nm wavelength of Arþ laser and
probe aperture 10mm) was used to identify the amorphous
structural of DLCH film. Fourier Transform Infrared (FTIR) spectrometer was applied to evaluate the curing degree of PI film. The ATR contact probe mode was used to avoid the interference fringes resulted from the uniformity of PI film thickness. IR spectra were taken from Perkinelmer Spectrum
2000 at a resolution of 4 cm1and the scan number was 16
units. Residual stress was determined by the profilometer
with the contact probe and was calculated from the conven-tional Stoney’s formula based on the radius curvature differences of the substrates between before and after coated with DLCH or PI film. The thermal stress of the film was calculated from the following formula:
th¼EFT ðFSÞ ð1Þ
whereEF is the Young’s modulus of the film,T was the
temperature difference between treatment and room
temper-ature,F,Swere the coefficients of thermal expansion of the
film and the substrate, respectively. Tape test used crosscut pattern tool (ASTM D3359-02) was used to evaluate the adhesion strength between PI and DLCH interface. MIS (Metal-Insulator-Semiconductor) sandwich structure was used to measure the electrical resistivity by using Cu as top and bottom electrode, the measuring tool is Agilent 4156 C precision semiconductor parameter analyzer, the test fre-quency is set at 1 MHz.
3. Results and Discussion
3.1 Surface morphology and cross section
The morphology of the 1.0mmthick DLCH film, 6.0mm
thick PI film and PI on DLCH hybrid films was made by optical microscope with a magnification factor 1000X. The DLCH film is free from any surface crack even with 1.67 GPa compressive stress measured by surface profilometer. PI surface is smooth and crack-free with low tensile stress; the stress value for PI film is 90 MPa and is about 5.4% of DLCH
film. Figure 2 shows the cross section of PI/DLCH hybrid structure and there is no crack existed at the DLCH/Si or PI/ DLCH interfaces. The measured thickness of DLCH film is
1.0mmand the thickness of PI film is 6.0mm.
3.2 Contact angle test
According to the Young’s equation on three phase
boundary (air-substrate-liquid), the adhesive tension
(Lcos) can be derived from the simple difference
(S{SL) between the surface tension of the solid (S) and
the interfacial tension at solid/liquid (SL).23)The definition
of adhesive work is L (1þcos). It is well known that
DLCH has the larger contact angle because of the
hydro-phobic C-H or CH3 chemical groups measured by FTIR
spectrum, as a result, DLCH contains high non-polar part.24)
However, in this study, annealed DLCH film seems to be more polar than the cured PI film. Possible reasons are the
loss of hydrogen evolution, graphitization from sp3 to sp2
carbon bonding and oxidized surface after the N2 gas heat
treatment or surrounding contamination. Unlike DCLH film,
PI is a strongly polar polymer due to imide ring.1)Table 1
shows the water contact angle test results based on the surface free energy of distilled water is 72.1 mN/m. The measured contact angle of DLCH film is consistent with the
Ostrovskaya’s result.25) The water contact angle on the
annealed DLCH is less than that of cured PI, but the water contact angle of the PI sample is similar to that of the PI/ DLCH. As a result, the adhesive tension on DLCH surface is
[image:2.595.56.285.69.225.2]Fig. 2 The SEM image for cross section of PI/DLCH hybrid structure.
Table 1 The water contact angles, adhesive tensions and adhesive works.
Sample type DLCH/Si wafer PI/Si wafer
Color bright black dark brown
Water droplet
Contact angle 52.6 61.3
[image:2.595.312.540.71.241.2] [image:2.595.47.550.657.787.2]larger than that on PI surface at a value of 9.2 mN/m. From these observations, we can hypothesize PI on DLCH system is a polar on more polar system, therefore surface roughness and surface chemical groups (hydrophilic or hydrophobic type) will affect the measured values during contact angle test.
3.3 Raman spectra of DLCH film
Figure 3 shows the typical Raman spectra of DLCH and annealed DLCH film. The D band means the disorder carbon
bonding centered at around 1350 cm1and G band means the
graphite carbon bonding located at around 1580 cm1.
Table 2 lists the D band position, G band position, full width high maximum (FWHM) of D band and G band, and
ID=IG ratio measured by Gaussian curve fitting under
different annealed condition. The clear trend is that the D
band peak position, G band peak position andID=IGpeak area
intensity ratio increase with increasing annealed temperature. These annealed characteristics indicate that an annealed DLCH film would lose the hydrogen and induce
graphitiza-tion reacgraphitiza-tion from C-C sp3bonded type transformation to
C-C sp2bonded type and cause a small reduction of the residual
stress.26,27)Table 2 also implies that the DLCH is more stable
and has slight graphitization when heating up to 300C. This
critical temperature is in agreement with the study that thermal annealing treatment would result in the conversion from amorphous carbon into graphite and soften the DLCH
structure.7–9)
3.4 FTIR spectra of PI film
The thermal imidization step of PI film consists of two major reactions: imidization and dehydration. The former is to form the imide ring with five fold type in relative to
general benzene ring (six fold type), and the latter is to evaporate the water molecule because one unit of PMDA and ODA monomer will produce two unit of water molecule during the polymerization reaction. The so-called ‘‘curing degree of PI’’ is the same meaning as conversion percentage of PI. The curing degree can be determined by the
conven-tional conversion equation. If one peak at 1380 cm1band is
assigned to C-N stretching vibration of imide ring, and the
other peak at 1500 cm1 band is assigned to C-C stretching
vibration of benzene,28)then the curing degree can be defined
as:
Curing degree
¼ bðA1380=A1500ÞT=ðA1380=A1500Þ400Cc 100% ð2Þ
whereAIis the intensity of absorbance at I wavelength.
The previous study has shown the PMDA-ODA PI
precursor will begin to convert to full cured PI at 250C
and its initial decomposition temperature (Td) is 470C when
the curing temperature is 400C. Figure 4 shows the FTIR
absorbance spectra of spin on PI film cured at 100, 300, 350
and 400C for one hour, and the relative curing degree is 47,
95, 97 and 100%, respectively. From these results, cured
temperature 300C is the most critical temperature in order to
get enough curing degree. This implies that the imidization reaction has almost finished at this critical temperature. Generally, the higher the curing temperature will result in
higher Tg and higher fracture strength for PI film.29)
3.5 Residual stress
The concave warpage of bare Si wafer is 3mmas shown in
Fig. 5(a). It shows during the polished step of wafer fabrication process, the wafer was in tensile residual stress state. Figure 5(b) shows that the convex warpage of DLCH
on Si wafer is about 30mm, this is caused by the energetic ion
bombardment during the DLCH deposition process. The compressive residual stress of as-deposited DLCH films range from 1.53–1.82 GPa was measured by warpage test. Many studies have reported that the high compressive stress
of DLCH film ranging from 5.0 to 12.5 GPa.30) The
compressive stress can be drawn from the fact that the coefficient of thermal expansion (CTE) of DLCH is smaller than that of Si wafer. It is very difficult to measure the
1000 1200 1400 1600 1800
Intensity (a.u)
Raman shift, R/cm-1
400o
C
350oC
300oC
RT
[image:3.595.57.283.71.236.2]Fig. 3 Raman spectra of annealed DLCH films.
Table 2 Deconvolution for Raman spectra of DLCH films. Annealed
Temp. (C)
D band FWHM (cm1)
G band FWHM (cm1)
D band (cm1)
G band (cm1)
ID=IG
ratio RT 331 142 1367 1555 1.72 300 318 135 1374 1559 1.77 350 319 128 1378 1562 2.02 400 308 123 1385 1566 2.04
800 1000 1200 1400 1600 1800
400O
C
350O
C 300O
C 100OC
Intensity (a.u)
Wavelength, W/cm-1
[image:3.595.315.538.74.237.2] [image:3.595.46.291.292.366.2]thermal stress of DLCH film without accurate measurement of DLCH’s CTE and Young’s modulus. Figure 5(c) shows
that the concave warpage of PI on Si wafer is 4.5mmresulted
from the cured process of PI and the CTE of PI (35 ppm/k) is larger than that of Si wafer (2.3 ppm/k). The measured tensile residual stress of PI film at cuing temperature 300,
350 and 400C are 90, 95 and 114 MPa, respectively, the
calculated thermal stress are 41, 48 and 56 MPa, respectively. The thermal residual stress is estimated near 50% of total residual stress for PI film. Figure 5(d) shows that the convex
warpage of PI/DLCH/Si wafer is around 28.0mm. In
comparison with Fig. 5(b), it has 6.7% warpage reduction, this means that PI has certain effect on DLCH film surface warpage and can be used as stress-buffer layer. The combination of PI’s residual tensile stress and thermal annealing relief effect on DLCH film can partially release
some DLCH’s residual compressive stress.7–9)
Adhesion strength can be evaluated by using the crosscut tool and tape test method. The results show a white and small area, which indicates the crosscut paths for remained DLCH surface due to PI film have been scratched out. On the other hand, the large and gray area shows the non-crosscut for PI/ DLCH combination. Figure 6(a) shows that for the PI on DLCH hybrid films, there is no network-like crack on the surface. The report from G.A. Abbas indicated 148-nm thick DLCH film on PET substrate had cracking all around the coating surface due to high compressive stress with
3.40 GPa.31)Another study from Y. Ohgoe showed that the
0.3mmthick DLCH film on polymeric diaphragm of artificial
heart device will induce severe wrinkling from high compressive stress of DLCH film corrugates the lower
tensile stress of polymeric film.20)In contrast to the easily
crosscut soft PI film, it is difficult to crosscut the hard DLCH film due to its high hardness. Figure 6(b) indicates that at the interface of PI/DLCH hybrid structure, the adhesion strength is high and this phenomenon is confirmed by the tape test.
3.6 Electrical resistivity
Figure 7 shows the typical current-voltage curve for the MIS structures on Cu/DLCH/Si/Cu, Cu/PI/Si/Cu and Cu/
PI/DLCH/Si/Cu system.32)The electrical resistance resulted
from I–V feature curve show that the electrical resistivity of
DLCH is 1.23*1012-cm and is 2.72*1011-cm at 300C
annealing state due to slight graphitization during annealing treatment. These values are in agreement with the previous
reports.33,34)The electrical resistivity of PI film is three orders
of magnitude higher than DLCH and is about 5.30*1014
-cm. The electrical series resistance of the PI/DLCH/Si
0 10 20 30 40 50
-200 -150 -100 -50 0 50 100 150
(d)
(c)
(b)
(a)
Wafer warpage,
Λ
/um (X10)
Scanning length, Λ/mm
Fig. 5 Warpage test for (a) Si wafer, (b) DLCH/Si, (c) PI/Si and (d) PI/ DLCH/Si hybrid structure.
Fig. 6 Tape test for PI/DLCH/Si hybrid structure with (a) before and (b) after test with a 50X magnification.
-10 -5 0 5 10
1E-14 1E-13 1E-12 1E-11 1E-10 1E-9 1E-8 1E-7 1E-6 1E-5 1E-4 1E-3
Current, I/A
Voltage, V/V DLC_RT
DLC_300oC PI_300o
C PI/DLC_3000C
[image:4.595.55.285.72.237.2] [image:4.595.312.541.74.431.2] [image:4.595.314.542.487.657.2]Experimental results show that for PI/DLCH/Si hybrid
structure, the optimal curing step is 300C for one hour with
N2 protective gas. This condition can get enough curing
degree for PI film during thermal imidization reaction and avoid severe graphitization resulted from the conversion of
sp3into sp2within DLCH film. PI/DLCH film have smooth
surface and are free from cracks or wrinkles due to high compressive stress of DLCH. The residual tensile stress of PI film can partially release DLCH’s residual compressive stress at PI/DLCH/Si hybrid structure. No obvious delamination occurs at the PI/DLCH/Si hybrid structure by tape test with crosscut pattern and this implies there is enough adhesion in the PI/DLCH interface. The electrical series resistance of the PI/DLCH/Si has increased significantly due to the PI passivation effect.
REFERENCES
1) A. Grill: Thin Solid Films355–356(1999) 189–193. 2) G. Maier: Prog. Polym. Sci.26(2001) 3–65.
3) L. Valentini, E. Braca, J. M. Kenny, G. Fedosenko, J. Engemann, L. Lozzi and S. Santucci: Diamond Relat. Mater.11(2002) 1100–1105. 4) Y. H. Cheng, Y. P. Wu, J. G. Chen, X. L. Qiao and C. S. Xie: Diamond
Relat. Mater.8(1999) 1214–1219.
5) G. A. Abbas, J. A. McLaughlin and E. Harkin-Jones: Diamond Relat. Mater.13(2004) 1342–1345.
6) C. Donnet: Surf. Coat. Technol.100–101(1998) 180–186.
7) D. R. Tallant, J. E. Parmeter, M. P. Siegal and R. L. Simpson: Diamond Relat. Mater.4(1995) 191–199.
8) B. Oral, R. Hauert, U. Muller and K. H. Ernst: Diamond Relat. Mater.4
(1995) 482–487.
9) R. Wachter and A. Cordery: Diamond Relat. Mater.8(1999) 504–509. 10) D. Y. Wang, C. L. Chang and W. Y. Ho: Surf. Coat. Technol.120–121
(1999) 138–144.
14) K. Donnelly, D. P. Dowling and M. L. McConnell: Diamond Relat. Mater.8(1999) 538–540.
15) V. M. Elinnson, V. V. Sleptsov and A. N. Laymin: Diamond Relat. Mater.8(1999) 2103–2109.
16) S. V. Borucki, W. Jacob and C. A. Achete: Diamond Relat. Mater.9
(2000) 1971–1978.
17) N. K. Cuong, M. Tahara, N. Yamauchi and T. Sone: Surf. Coat. Technol.174–175(2003) 1024–1028.
18) M. Yoshida, T. Tanaka, S. Watanabe, M. Shinohara, J. W. Lee and T. Takagi: Surf. Technol.174–175(2003) 1033–1037.
19) A. R. McCabe, A. M. ones and S. J. Bull: Diamond Relat. Mater.3
(1994) 205–209.
20) Y. Ohgoe, K. K. Hirakuri, K. Tsuchimoto, G. Friedbacher and O. Miyashita: Surf. Coat. Technol.184(2004) 263–269.
21) N. K. Cuong, M. Tahara, N. Yamauchi and T. Sone: Surf. Coat. Technol.193(2005) 283–287.
22) P. Garrou: IEEE TRANSACTIONS ON ADVANCED PACKING23
(2000) 198–205.
23) E. Lugscheider, K. Bobzin and M. Moller: Thin Solid Films355–356
(1999) 367–373.
24) J. Kim, K. Lee, K. Kim, H. Sugimura, O. Takai, Y. Wu and Y. Inoue: Surf. Coat. Technol.162(2003) 135–139.
25) L. Ostrovskaya, V. Perevertailo, V. Ralchendo, A. Dementjev and O. Loginova: Diamond Relat. Mater.11(2002) 845–850.
26) R. Vuppuladhadium, H. E. Jackon and L. C. Wu: J. Appl. Phys.77
(1995) 2714–2718.
27) D. R. Tallant, J. E. Parmeter, M. P. Siegal and R. L. Simpson: Diamond Relat. Mater.4(1995) 191–199.
28) T. Nishino, M. Kotera, N. Inayoshi, N. Miki and K. Nakamae: Polymer
41(2000) 6913–6918.
29) M. Amagai: Microelectronics Reliability40(2000) 2077–2086. 30) X. L. Peng and T. W. Clyne: Thin Solid Films312(1998) 207–218. 31) G. A. Abbas, J. A. McLaughlin and E. Harkin-Jones: Diamond Relat.
Mater.13(2004) 1342–1345.
32) E. Staryga and G. W. Bak: Diamond Relat. Mater.14(2005) 23–34. 33) N. Tomozeiu, A. Har, B. Kleinsorge and W. I. Milne: Diamond Relat.
Mater.8(1999) 522–526.