Applied Surface Science 257 (2011) 9158– 9163

Contents lists available at ScienceDirect

Applied Surface Science

jou rn al h om epa g e: www.elsev ier .com/ locate /apsusc

An undercutting model of atomic oxygen for multilayer silica/alumina films

fabricated by plasma immersion implantation and deposition on polyimide

Yongxian Huanga,∗, Xiubo Tiana, Shixiong Lva, Shiqin Yanga, R.K.Y. Fub, Paul K. Chub,Jinsong Lengc, Yao Li c

a State Key Laboratory of Advanced Welding and Joining, Harbin Institute of Technology, Harbin 150001, People’s Republic of Chinab Department of Physics and Materials Science, City University of Hong Kong, Tat Chee Avenue, Kowloon, Hong Kongc Center for Composite Materials and Structures, Harbin Institute of Technology, Harbin 150001, People’s Republic of China

a r t i c l e i n f o

Article history:

Received 10 May 2011

Received in revised form 30 May 2011

Accepted 30 May 2011

Available online 6 June 2011

Keywords:

Polymer

Multilayer thin films

Erosion

Plasma immersion

Monte Carlo

a b s t r a c t

Multilayer silica/alumina films were created by plasma immersion implantation and deposition to protect

against atomic oxygen (AO) in low earth orbit environment. The AO erosion mechanism of polyimide

under multilayer silica/alumina films has been investigated using a ground-based AO simulator and

Monte Carlo model. The results demonstrate that protective films are detached and plumped due to

AO undercutting, and the exterior silica film is partly detached proven by chemical composition depth

profile and erosion patterns. The undercutting model involving collision, diffusion, reaction, gas releasing,

and retroaction on films is proposed. Based on the model, scattered impingement has serious erosion,

although AO does not directly attack interior polymer. AO erosion predictions at two neighborhood cracks

are first studied by Monte Carlo model for various incidence angles of AO. The protective film between

cracks hinders the escape of AO, and accelerates the erosion.

© 2011 Elsevier B.V. All rights reserved.

1. Introduction

The low earth orbit (LEO) environment at altitudes of

200–700 km, where satellites orbit the earth, is extremely harsh

due to the presence of atomic oxygen (AO). The exposure of organic

materials such as polymer in AO environment may cause a surface

erosion rate as high as 0.5 �m per orbit [1], which can induce the

deterioration of their mechanical properties as a result of polymer

molecular destruction or polymer surface etching, and influence

service lifetime of space vehicles [2–6]. For such composites to be

used in long term space applications, they must be protected in

some manners. Owing to unique properties such as optical trans-

parency, high thermal, mechanical and high electrical resistivity

[7–10], inorganic oxide protective coatings can be used to pre-

vent or minimize surface degradation of advanced polymers and

polymer composites of LEO spacecrafts [11–14]. However, most

protective films have pinholes and defects caused by the deposition

process or substrate imperfections [15]. And the thermal expansion

coefficients of some coatings differ markedly from those of polymer

substrates, which results in cracks due to the thermal cycling after

long exposure. These cracks may provide pathways for AO to dif-

∗ Corresponding author. Tel.: +86 451 86413951; fax: +86 451 86416186.

E-mail address: yxhuang@hit.edu.cn (Y. Huang).

fuse through and attack at the polymer surface resulting in erosion,

which may lead to the failure of both coating and polymer film.

It is essential to understand and negate the adverse effects of

AO erosion on spacecraft materials in order to predict performance

characteristics such as in-space durability [16–18]. However, the

prediction of in-space durability of protected organic materials has

proven to be difficult due to the lack of information concerning the

rate of in-space undercutting oxidation at defect sites in protective

coatings [16]. Predicting the AO durability in the space environment

is a very complex task complicated by the fact that material may

be sensitive to a different synergistic component in the environ-

ment. The erosion mechanisms of polymeric materials are complex

and involve initial reactions at the gas–surface interface as well

as steady-state material removal processes [19]. In this work, we

report experimental and Monte Carlo model results of AO induced

undercutting, and a key issue is to propose a model of AO induced

erosion of protective films on polymer substrate. The etching mech-

anism is indispensable for application of polymer thin films with

coating as a spacecraft construction material.

2. Experimental

2.1. Materials and preparation of samples

Experiments were performed with 100 �m thick polyimide film

(DuPontTM). Multilayer silica/alumina films were fabricated on the

0169-4332/$ – see front matter © 2011 Elsevier B.V. All rights reserved.doi:10.1016/j.apsusc.2011.05.124

Y. Huang et al. / Applied Surface Science 257 (2011) 9158– 9163 9159

Fig. 1. Characteristic erosion pattern of polyimide covered with multilayer sil-

ica/alumina films exposed to a fluence of 1.527 × 1020 atoms cm−2 AO: (a) before

exposure, and (b) after exposure.

surface of polyimide using plasma ion implantation and deposition

(PIID). The aluminum and silicon plasmas were generated from the

vacuum arc using a high voltage to trigger the metal arc and a low

sustaining arc voltage between the cathode and anode. The poly-

imide samples with the size of 100 mm × 100 mm were positioned

15 cm away from the exit of the curved filtered duct, and treated in

the Al plasma for 60 min and then in Si plasma for 60 min with a neg-

ative pulse bias of 10 kV with pulse width of 150 �s and repetition

rate of 60 Hz.

2.2. AO exposure tests

Polyimide with multilayer silica/alumina films (� 40 mm)

was exposed by AO using a ground-based AO simulator, where

the AO beam with 5 eV energy and 2.02 × 1015 atoms cm−2 s−1

flux was produced using a CO2-laser detonation technology.

The longest exposure time is 21 h and the AO fluence is

1.527 × 1020 atoms cm−2. A vacuum level of ∼10−3 Pa was main-

tained during AO exposure. The mass loss was measured using a

microbalance with the precision of 10−5 g. To obtain accurate mass

loss measurements, the sample was dehydrated and outgassed in

vacuum for 24 h before AO exposure.

2.3. Characterization techniques

X-ray photoelectron spectroscopy (XPS) was performed on the

Physical Electronics PHI-5600 using the monochromatic AlK� radi-

ation operated at 14 kV and 350 W. The photoelectron takeoff angle

was 45◦. The XPSPEAK software was used for peak fitting. This pro-

gram employs Newton’s method for optimization as well as Shirley

background subtraction and Gaussian–Lorentzian function for peak

fitting. The morphology of the samples before and after exposure

was examined by scanning electron microscopy (SEM). The samples

were deposited with Pt for SEM observation.

3. Results

3.1. Erosion of multilayer silica/alumina films

The SEM micrograph of multilayer silica/alumina films on a

polyimide substrate before and after exposure to AO is shown in

Fig. 1. The films were partly detached from the underlying layer

after AO exposure. In the case, the undercutting process led to “lift-

off” of the films. But the coating is crack-free even after exposure

to high fluence level. The sample with multilayer silica/alumina

Fig. 2. Chemical composition evaluated by XPS versus sputter time and thickness

of multilayer silica/alumina films on polyimide before and after AO exposure.

films showed a significantly reduced mass loss (1.06 mg/4.71 mg,

no more than quarter), compared with the polyimide control sam-

ple.

XPS was conducted to gain further insight into the chemical

changes caused by AO. Fig. 2 plots Si, Al, O, and C atomic concentra-

tion (in at.%) for multilayer silica/alumina films on the polyimide

substrate. After AO exposure, the results of higher Al and lower

Si concentrations on the surface demonstrate that the erosion of

the upper silica film has indeed occurred. Part of silica film was

peeled off, and then the downside alumina film was moved upside.

In addition, significant carbon content has been found in the mul-

tilayer films before AO exposure, which can be explained by the

presence of hydrocarbon contamination on the surface [20]. A sig-

nificant decrease of carbon content is observed after AO exposure. A

decrease of carbon concentration from 28.49% to 11.31% is detected

at 42 nm thickness. It is the reason why protective layers without

initial defects are undercut. The volatile fragments, such as short-

chain oxidation products, leave the surface of the protective layers.

As a result, the defects, such as pinholes and micro-cracks, are

formed, and then act as channels for AO to penetrate into polymer

substrate.

Fig. 3 shows high resolution C 1s carbon peaks for unexposed and

exposed polyimide covered with multilayer silica/alumina films.

The C 1s peak is composed of four different components: Al–O–C

at a binding energy of 283.6 eV [21,22], C1 at 284.6 eV (C–C) which

corresponds to carbon atom bonds from aromatic rings not directly

attached to the imide ring, C3 at 285.6 eV (C–N) which corresponds

to carbon atoms bonded to nitrogen, and C5 at 286.3 eV (C–O) which

corresponds to carbon atoms bonded to oxygen related to the ether

group [23–26]. The results show that there exist polymer segments

in the multilayer films. These organic components will react with

AO during exposure. Some volatile products will carry mass away

from the multilayer films, and defects are formed in the protective

film. These defects may provide pathways for AO to diffuse through

and to attack at the underlying polymer resulting in erosion.

3.2. Undercutting at the defects of protective film

Two features of AO erosion should be noticed. Firstly, for the

case without initial defects, the undercutting appears to be depen-

dent upon the organic component in the protective film as volatile

fragments leaving to form the defects, as shown in Fig. 1(b). Sec-

ondly, when some defects are existent, pathways are provided for

9160 Y. Huang et al. / Applied Surface Science 257 (2011) 9158– 9163

Fig. 3. Structure of polyimide and high resolution C 1s XPS spectra: (a) molecular

structure of polyimide with numerically labeled carbon atomic sites, and (b) high

resolution C 1s XPS spectra obtained from the multilayer films before and after AO

exposure.

the undercutting by AO, as shown in Fig. 4. If pinhole defects exist in

protective films, AO erosion can occur primarily due to secondary

impacts. The substrate material is eroded away with the diame-

ter of the eroded substrate area being larger than the diameter

of the original defect [27]. Upon further AO exposure, the erosion

zone spreads beneath the protective coating to greater depths and

widths proceeding through the substrate. Fig. 4(a) shows a picture

of a pinhole defect above an undercutting site [28]. The undercut-

ting area is far wider than the width of the initial crack defects.

The erosion patterns of single crack exposing in AO ground-based

simulator is shown in Fig. 4(b), and inset is erosion patterns of

long duration exposure facility (LDEF) in LEO [29]. AO undercut-

ting with normal incidence ram attack can occur because there are

AO velocity components which are transverse to the ram direction

[16]. These transverse components are the contributions of thermal

velocity and scattering contributions.

3.3. Monte Carlo model simulation at neighborhood cracks

A two dimensional Monte Carlo computational model has been

developed to simulate AO attacking and undercutting at two neigh-

borhood cracks in protective films over polymers. The undercutting

physical process of AO erosion at two neighborhood cracks is shown

in Fig. 5. The physical process of AO undercutting is assumed to be a

particle transport procedure. The reacted AO are absorbed and the

unreacted AO continues to complete their transportation course by

means of diffusion or mirror reflection. In diffuse reflection, thermal

Fig. 4. Erosion patterns of a polyimide with protective film for: (a) pinhole [25], and

schematic diagram of AO interaction, and (b) single crack, inset is erosion pattern

of LDEF in LEO [26].

exchange occurs because kinetic energy of AO is equal to the speed

of thermal AO, and AO will depart from the surface. The distribu-

tion direction is in compliance with the law of cosine distribution. In

mirror reflection, AO maintains the identical energy and the same

impact-reaction probability in the next collision. Until absorption

from erosion hole at cracks, AO would be reflected continually and

react with the polyimide film in accordance with the define reac-

tion probability [30–33]. Two neighborhood cracks were chosen to

simulate the undercutting of polymer underlying protective film by

AO. AO motion direction is limited to the vertical normal plane from

material surface for the two cracks. The velocity vectors of AO have

Fig. 5. Undercutting physical process of AO erosion at neighborhood cracks.

Y. Huang et al. / Applied Surface Science 257 (2011) 9158– 9163 9161

Fig. 6. Monte Carlo computational AO erosion predictions for various attack angles of AO at neighborhood cracks: (a) 0◦ , (b) 45◦ , (c) −45◦ , and (d) 45◦ and −45◦ .

random orientations and their speed distribution is Maxwellian.

Protective film will not take part in the reaction of AO and the

mean free path of AO is too large for recombination and forma-

tion of oxygen molecules. AO are driven to enter cracks and impact

the polymer cells. Oxidization effects occur and the reacted cells are

removed. Optimal values of AO interaction parameters have been

identified by forcing the Monte Carlo computational predictions

to match the results of protected samples retrieved from the LDEF

[34].

Two neighborhood cracks with width of 1 �m and 4 �m and

with distance of 5 �m were chosen to study the interaction of cracks

during AO exposure. Fig. 6 shows the Monte Carlo model computa-

tional erosion cavities, and four kinds of situations of 0◦, 45◦, −45◦

attack angles have been carried on the analysis. When AO enters,

the AO becomes somewhat trapped and has multiple opportunities

for reaction until it either reacts or escapes out the defects. When

AO fluence is low, the cracks are isolated during AO exposure, and

erosion cavities do not connected. With increasing AO fluence, the

polymer just under the protective film between cracks decreases,

and the undercutting cavities have been gradually connected. In

some extent, the protective film hinders the escape of AO, and accel-

erates the erosion. That is to say, materials not only are attacked by

AO in the line of sight of the ram fluence, but also by reflected AO.

Experiments performed by Banks et al. [35] also showed the same

effect. When the angles of incidence are 45◦ and −45◦, the interac-

tion between cracks is different owing to the neighborhood cracks

width. For the angle of 45◦, the erosion is more serious at the side of

wide crack, and erosion cavities have been communicated when the

AO fluence is only 1.923 × 1021 atoms cm−2. As shown in Fig. 6(d),

to simulate in-orbit service process of flight vehicle, the 45◦ angle

of AO incidence is transformed as −45◦, the erosion morphology is

more complex, and “beaker” shape disappears. Due to the interac-

tion of two neighborhood cracks, the erosion cavity width increases

with interior erosion step.

4. Discussion

The AO eroded profiles, which are far wider than the width of

the initial defects, is named the undercutting model. As shown in

Fig. 7, the basic process in the proposed model involves the colli-

9162 Y. Huang et al. / Applied Surface Science 257 (2011) 9158– 9163

Fig. 7. Schematic diagram of interaction between AO and polymer covered with protective film.

sion, diffusion of AO and reaction with polymer, forming bubble at

or below the surface of polymer, breaking of blister with gas release,

and eventual formation of crater at the polymer surface, detaching

and plumping the protective films. Initially, incidence AO enters

from defects and collides with polymer or protective film. Direct

inelastic scattering, in which only a fraction of the initial transla-

tion energy is lost to the surface, is the most probable non-reactive

interaction [19]. The reaction probability for thermally accommo-

dated AO is probably not greater than 0.003. The loss rate of AO on

the surfaces is characterized in terms of the surface loss coefficient,

= 1 − (Fout/Fin), where Fin is the incident fluence of AO onto a sur-

face and Fout is the total fluence of AO atoms leaving the surface,

from both instantaneous reflection and desorption of adsorbed AO

[36]. Greaves et al. has reported values of from 0.0001 to 0.24 for

AO on various oxide surface [37]. Most of AO scatters with sufficient

energy to break organic polymer bonds. Then, sufficient fluence

causes oxidative erosion of polymers, and significantly contributes

to the undercutting. Secondly, the chemisorption AO diffuses in

the polymer. The impinging AO can be captured by a potential

well at or below the surface where it chemically reacts to form an

oxide, which migrates from the surface into the bulk of the poly-

mer [38]. AO sticking to the surface is capable to diffuse into the

polymer to significant depths (more than 5 nm at least) before it

reacts with polymer molecules [2]. Thirdly, AO reacts with some

elements such as carbon, nitrogen, and hydrogen to form bubbles.

While the concentration of the gas builds up, the coalescence of

bubbles forms blisters [39]. Fourthly, in polymer substrate, a crater

is formed and gas is released. Meanwhile, volatile products will

carry mass away from the surface [27]. It is speculated that these

gases are released from the surface by imide ring decomposition

[38]. Lastly, the upside protective film is detached and plumped.

Just so, although AO attack on internal or interior surface may not

have direct exposure to the LEO AO fluence, scattered impinge-

ment can have serious degradation effects where sensitive interior

surface are present [29].

The results of erosion pattern and chemical composition depth

profile make sure that the undercutting model is able to explain the

erosion mechanism of AO. Based on the undercutting model, the

anti-AO erosion protective films should be multilayer and have no

or few defects. Furthermore, interface adhesion of protective film

should be high enough to avoid cracks caused by thermal cycle. Due

to the formation of complexes to increase adhesive strength, the

hybrid PIID process will be promising for fabricating the protective

film for anti-AO undercutting. Prior to ion implantation process,

nano-scale deposition coatings are needed for preventing organic

components entering protective film.

5. Conclusion

In the present study, the undercutting model of polymer covered

with protective film induced by AO was defined. The undercut-

ting model is proposed involving the collision, diffusion of AO and

reaction with polymer, forming bubble at or below the surface of

polymer, breaking of blister with gas release, and eventual forma-

tion of crater on the polymer surface, detaching, plumping and even

bursting the protective film. The established model is supported

by the results of erosion pattern and chemical composition depth

profile, and can be applied to explain the AO erosion mechanism

of polymer covered with protective film. The neighborhood cracks

are not isolated, and the undercutting cavities have been gradually

connected with increasing AO fluence. Furthermore, the protec-

tive film between cracks hinders the escape of AO, and accelerates

the erosion. The results of this study could shed more insights into

predicting the AO durability of existing polymers and fabricating

anti-AO erosion protective film for the next generation of spacecraft

materials.

Acknowledgements

The work was jointly supported by the Science and Technol-

ogy Innovation Research Project of Harbin for Young Scholar (No.

2009RFQXG050), the National Natural Science Foundation of China

(No. 50904020), and the China Postdoctoral Science Foundation

(No. 20090460883).

References

[1] J. Visentine, NASA/SIDO Space Environmental Effects on Materials Workshop,NASA Conference Publication, Hampton, 1988, p. 179.

[2] V.E. Skurat, E.A. Barbashev, Y.I. Dorofeev, A.P. Nikiforov, M.M. Gorelova, A.I.Pertsyn, Appl. Surf. Sci. 92 (1996) 441–446.

[3] R. Verker, E. Grossman, N. Eliaz, Acta Mater. 57 (2009) 1112–1119.[4] M.Z. Wang, X.H. Zhao, Z.G. Shen, S.L. Ma, Y.S. Xing, Polym. Degrad. Stab. 86

(2004) 521–528.[5] H. Shimamura, T. Nakamura, Polym. Degrad. Stab. 95 (2010) 21–33.[6] F. Awaja, J.B. Moon, S. Zhang, M. Gilbert, C.G. Kim, P.J. Pigram, Polym. Degrad.

Stab. 95 (2010) 987–996.[7] S. Mann, G.A. Ozin, Nature 382 (1996) 313–318.[8] J. Li, Y. Han, Langmuir 22 (2006) 1885–1890.[9] M. Deepa, A.K. Srivastava, S. Lauterbach, Govind, S.M. Shivaprasad, K.N. Sood,

Acta Mater. 55 (2007) 6095–6107.[10] Y.H. Choi, X. Bulliard, A. Benayad, Y. Leterrier, J.A.E. Månson, K.H. Lee, D. Choi,

J.J. Park, J. Kim, Acta Mater. 58 (2010) 6495–6503.[11] H. Shimamura, T. Nakamura, Polym. Degrad. Stab. 94 (2009) 1389–1396.[12] I.H. Tan, M. Ueda, R.S. Dallaqua, N.R. Demarquette, L. Gengembre, Plasma Pro-

cess. Polym. 4 (2007) S1081–S1085.[13] Y.X. Huang, X.B. Tian, S.Q. Yang, R.K.Y. Fu, P.K. Chu, Appl. Surf. Sci. 253 (2007)

9483–9488.

Y. Huang et al. / Applied Surface Science 257 (2011) 9158– 9163 9163

[14] M. Ueda, K.G. Kostov, A.F. Beloto, N.F. Leite, K.G. Grigorov, Surf. Coat. Technol.186 (2004) 295–298.

[15] M.D. Groner, S.M. George, R.S. Mclean, P.F. Carcia, Appl. Phys. Lett. 88 (2006)051907.

[16] K.K. de Groh, B.A. Banks, D.C. Smith, Solar Eng. 2 (1995) 939–950.[17] S.K. Rutledge, J.A. Mihelcic, Undercutting of Defects in Thin Film Protective

Coatings on Polymer Surfaces Exposed to Atomic Oxygen, 1990, NASA TM-101986.

[18] A. Snyder, K.K. de Groh, The Dependence of Atomic Oxygen Undercutting ofProtected Polyimide Kapton® H upon Defect Size, 2001, NASA-TM-210596.

[19] T.K. Minton, J.M. Zhang, D.J. Garton, J.W. Seale, High Perform. Polym. 12 (2000)27–42.

[20] G. Dennler, A. Houdayer, P. Raynaud, I. Séguy, Y. Ségui, M.R. Wertheimer, PlasmaPolym. 8 (2003) 43–59.

[21] M.C. Zhang, E.T. Kang, K.G. Neoh, C.Q. Cui, T.B. Lim, Polymer 42 (2001) 453–462.[22] Q.D. Ling, S. Li, E.T. Kang, K.G. Neoh, B. Liu, W. Huang, Appl. Surf. Sci. 199 (2002)

74–82.[23] G. Beamson, D. Briggs, High Resolution XPS of Organic Polymers: The Scienta

ESCA300 Database, Wiley, Chichester, England, UK, 1992.[24] S.H. Kim, S.H. Cho, N.E. Lee, H.M. Kim, Y.W. Nam, Y.H. Kim, Surf. Coat. Technol.

193 (2005) 101–106.[25] S.C. Park, K.J. Min, K.H. Lee, Y. Jeong, Y.B. Park, Met. Mater. Int. 17 (2011)

111–115.

[26] L.J. Matienzo, F.D. Egitto, Polym. Degrad. Stab. 35 (1992) 181–192.[27] K.K. de Groh, J.A. Terlep, T.M. Dever, Atomic Oxygen Durability of Solar-

Concentrator Materials for Space Station Freedom, 1990, NASA-TM-105378.[28] A. Snyder, B.A. Banks, D.L. Waters, Undercutting studies of protected Kapton®

H exposed to in-space and ground-based atomic oxygen, in: Proceedings of the10th ISMSE & the 8th ICPMSE, Collioure, France, SP-616, 2006.

[29] B.A. Banks, K.R.M. Sharon, K.K. de Groh, R. Demko, Atomic Oxygen Effects onSpacecraft Materials, 2003, NASA-TM-212484.

[30] B.A. Banks, K.K. de Groh, B.M. Auer, L. Gebauer, J.L. Edwards, Monte Carlo mod-eling of atomic oxygen attack of polymers with protective coatings on LDEF,AIAA 11 (1993) 282820.

[31] K. Kim, B.A. Banks, J. Spacecraft Rockets 31 (1994) 656–664.[32] B. Lin, S.L. Gao, Y. Zhao, J. Aerospace Shanghai 6 (2001) 55–59.[33] Y. Liu, G.H. Li, Acta Astronaut. 67 (2010) 388–395.[34] B.A. Banks, T. Stueber, M. Norris, Monte Carlo Computational Modeling of the

Energy Dependence of Atomic Oxygen Undercutting of Protected Polymers,1998, NASA-TM-207423.

[35] B.A. Banks, K.K. de Groh, K.R.M. Sharon, MISSE Scattered Atomic Oxygen Char-acterization Experiment, 2006, NASA-TM-214355.

[36] S. Gomez, P.G. Steen, W.G. Graham, Appl. Phys. Lett. 81 (2002) 19–21.[37] J.C. Greaves, J.W. Linnett, Trans. Faraday Soc. 54 (1958) 1323–1330.[38] M.R. Reddy, J. Mater. Sci. 30 (1995) 281–307.[39] D. He, M.N. Bassim, J. Mater. Sci. 33 (1998) 3525–3528.