230 Thin Solid Films, 236 (1993) 230-235

An overview on metal/PET adhesion

J. F . S i l v a i n
CNRS, Laboratoire de Chimie du Solide, 351 Cours de la Lib&ation, F-33405 Talence Cedex (France)

J. J. Ehrhardt
CNRS, Laboratoire Maurice Letort, 405 Rue de Vandoeuore, F-54600 Fillers-lbs-Nancy (France)

Abstract

Transmission electron microscopy and X-ray photoelectron spectroscopy (XPS) were used to characterise thin metal
films (Mg, AI, Cu, Ag) thermally evaporated onto polyethylene terephthalate (PET) and to study the formation of
the AI/PET interface. The adhesion was measured with a 180° peel test technique. XPS spectra show that the A1
atoms react preferentially with the carboxylic group of the PET and that the A1/PET interface exhibits a pseudo
layer-by-layer growth mechanism. Two factors strongly favour the increase of metal/PET adhesion: (1) a PET
temperature higher than 100 °C during metal deposition (A1, Cu and Ag) and (2) a partial pressure of oxygen higher
than 10-5 mbar for the A1 evaporation. Furthermore, atomic metal diffusion tends to increase the adhesion while
cluster segregation within the PET skin decreases the metal/PET adhesion.

1. Introduction The microstructure of the metal/PET interface and the

Metallised polymers are commonly used for audio
and video recording tapes, and electronic devices, and
in the packaging industry. Knowledge of the properties
of the metal/polymer interface is important, particularly
to enhance the control of the interface chemistry and to
promote better adhesion. In order to improve the adhe-
sion, several techniques are commonly used, before and
during metal deposition. Corona discharge treatment
[1, 2], r.f. ion-plating treatment [3] and chemical treat-
ments (bromosulfonic acid, fluorine gas [4]) have been
used to improve the surface adhesion characteristics of
polymers. Ion-based techniques in thin film deposition
are also used increasingly to control adhesion. These
techniques are: ion plating, ion beam deposition [5],
electron beam deposition, magnetron sputtering and arc
deposition [6]. Oxygen treatment seems also to play an
important role on the improvement of the metal/poly-
mer adhesion. For example, if oxygen is added to a
polystyrene surface, Burstrand [6, 7] finds that the
metal atoms form complexes with the oxygen at the
interface with a subsequent increase in the adhesion.
The binding mechanisms, the composition and the
structure of the interface have been extensively studied
for the deposition of metals on polyimide [8-12] and to
a lesser extent on polyesters [ 13-16].
In this paper, four metals (Mg, A1, Ag, Cu) have
been evaporated on commercial polyethylene tereph-
thalate (PET) (Du Pont de Nemours (Luxembourg)
S.A. (MylarrM)), at different temperatures ranging
from 20 °C to 250 °C and after different treatments.
morphology of the metal have been correlated with the
adhesion of the metallic film. Two different PET are
used; the first one presents a low crystallinity and the
second a high one. High P E T crystallinity was mostly
used except when noted in the text. The degree of
crystallinity or of chain orientation depends on the
elaboration conditions (biaxially stretched polymer).
We report also on the first steps of AI deposition on
PET studied by X-ray photoelectron spectroscopy
(XPS). Then the chemical composition of the interface
AI/PET and the growth process were investigated for
coverage up to 20 monolayers (ML) (1 M L = 3.2 A).
2. Experimental details
Metallisations were performed in an evaporation
chamber. Constant metallic vapour fluxes (3 nm min-~)
were produced by Knudsen cells and monitored with a
quartz balance. The base pressure, measured with an
ionisation gauge, before evaporation and without oxy-
gen introduction is around 10 -7 mbar. A partial pres-
sure of oxygen (10 -6 to 10 - 4 mbar) can be introduced,
through a leak valve, before and during metallisation.
Sheets of polymers of about 10 x 10 cm 2 are fixed
onto an oven heated by a resistive element. The temper-
ature of the polymer was measured via a chromel
alumel thermocouple. The distance between the Knud-
sen cells and the surface of the polymer was about
50 cm allowing a uniform deposition rate on the entire
surface of the substrate.

0040-6090/93/$6.00 © 1993-- Elsevier Sequoia. All rights reserved

 J. F. Silvain, J. J. Ehrhardt / An overview on metal~PET adhesion 231

These metal/PET laminates were observed either by 0.1

, t , i . v. , , I , i , i , i . i ,
0.0 -
conventional or cross-section transmission electron mi- ~. -0.1.
-0.2
croscopy (TEM) in a JEOL 200 Cx microscope. For the "< -0.3'
-0.4
cross-section preparation, the samples were prepared by -0.5"
-0.6"
bonding two AI/PET samples with A1 surfaces in con- -~ -0.7"
-0.B-
tact with one another (face to face) with an epoxy resin -~ -0.9-
-I.0"
and curing the resultant couple under conditions of -1.1-
-1.2
moderate temperature (40 °C) and pressure. Thin slides 1 2 3 4 5 6 7 8 9 10
Thickness (/I.)
were prepared by mechanical grinding followed by
ion milling [ 17]. The plane section samples were sub- Fig. 1. Attenuationcurves(reducedintensityvs. A1coverage)for the
sequently prepared by dissolving the PET in tri- three distinct carbon environmentsof the PET, namelyCI at 285 eV
fluoroacetic acid (5-10 min immersion). (~-), Cn at 287.7eV (42~O-) and Civ at 289.8eV (~O~)--O).
For the adhesion measurements [ 14], test pieces con-
sisted of an AI support (1 mm thick)/double-sided tape I ML AI/PET

(Permacel P-94)/PET (12 prn)/evaporated metal/ (¢/=)

ethylene acrylic acid (EAA) copolymer film. These lam-
inates were prepared for the peel test by compression
under 1.3 x 105 Nm 2 at 120 °C for 10 s. The peel test
was performed by peeling the EAA copolymer sheet
from the laminate in an Instron tensile tester at
180 ° peel angle and 5 cm min -~ peel rate. It is worth
noting that the upper limit of this measurement is
1300 g in-] due to the mechanical properties of the bulk
polymer,
The electron spectroscopy for chemical analysis
(ESCA) studies have been performed in an UHV appara-
tus [ 18] allowing in situ metal deposition [ 19]. XPS spectra
were recorded in the AE mode with an instrumental
resolution of about 1.1 eV. Deconvolutions of the
C~s, Ols and A12plevels were performed using different
components modelled by symmetric gaussian curves.
3. Experimental results
3.1. X P S study o f the A I / P E T interface
The Cls spectra were first recorded for a virgin PET
surface. They exhibit the fine structure usually exhibited
by this polymer [20-23]. Measured lines have been
decomposed into three peaks identified as Cr at
285.7 eV (-C-), Cn at 287.7 eV (-C--O-) and Cw at
289.8 eV (-O-C--O). The chemical shifts between the
three carbons are in reasonable agreement with litera-
ture values [22]. The behaviour of the intensities of the
three functional groups of carbon has been investigated
during A1 deposition at room temperature (Fig. 1) on
biaxially stretched PET. It is worth noting that:
(a) the attenuation of C~ and CII components obeys a
quasi exponential law (almost parallel straight lines in
the logarithm representation used here), indicative of a
pseudo layer-by-layer growth mechanism,
(b) the intensity of the C1v peak decreases much faster
than the intensity of C~ and Cn peaks. This is evidence
of the drastic chemical change of the carboxylic group
induced by A1 deposition.
NeE) o""xta¢ AI 2p N(E) mcud
Co/s) i!l
io'
: : ......... '?~:',C
86 82 78 74 78 EeKCeV) 86 82 78 74 7~l E~(.V)
Fig. 2. AI2pXPS spectra of A1 deposited on PET at 300 K (1 and
20 ML).
The Al2p levels have been recorded after in situ A1
deposition (1, 10 and 20ML) on biaxially stretched
PET kept at room temperature. In our experimental
conditions, reactions between molten A1 and the A120 3
crucible do occur giving an unexpected flux of "oxi-
dised" AI (as high as 40% in some circumstances) and
we have never been able to produce a pure metallic
aluminium film whatever the thickness or the substrate.
On non-treated PET (Fig. 2), for deposition at room
temperature, it has to be mentioned that the first A1
monolayers are fully oxidised. This is strong evidence for
the formation of AI-O bonds with the oxygen atoms of
the surface of the polymer. After deposition of about
10 ML, the composition of the film is close to that of the
incoming vapour. Therefore, the oxidation process in-
volves only a few layers and the interface would appear
to be quite sharp. At low temperature, this reaction is
strongly inhibited and the composition of the film can be
considered to be independent of its thickness.
3.2. Influence o f the P E T temperature on the
metal~PET adhesion
The adhesion strength of the various metal/PET sam-
ples is reported in Fig. 3 for magnesium, aluminium,
silver and copper films deposited at different substrate
temperatures. The adhesion of the magnesium films is
232 J. F. Silva&, J. J. Ehrhardt / An overview on metal~PET adhesion

The mean grain size also depends strongly on the

i ature, grain size increases with temperature for both
<
Mg
Fig. 3. Adhesion measurements of thermally evaporated metallic thin
films (magnesium, aluminium, silver and copper) at different temper-
atures of the polymer.
close to the highest measurable value; it does not de-
pend on the substrate temperature and is always much
higher than the observed adhesion of the other metals
at the same temperature. The three other metals behave
similarly in the sense that adhesion increases with the
surface temperature of the polymer, and that cohesive
failure inside the PET is always found.
Due to the technical importance of the Cu and A1
films on PET, the deposition of these two metals has
been studied in more detail; the A1/PET adhesion and
the mean grain size are reported in Figs 4(a) and 4(b),
as a function of the temperature of the substrate during
metal deposition. It appears that metal adhesion im-
proves with increasing temperature. Furthermore,
above a critical temperature around 100-120 °C, it is
observed that the adhesion increases very rapidly up to
the upper limit value of 1300 g in -1 for both metals.
1400...,...,...,...L...i...,...=...~...,...
~" 1200:
1000 -
800 -
polymer surface temperature. Below the critical temper-
aluminium and copper grains. Above this temperature,
the grain size in the AI films drops by one order of
magnitude and then becomes independent of the tem-
perature. In contrast, the grain size in the copper film
A1 Ag Cu
increases above this temperature and reaches a constant
value. The aluminium and copper grain sizes are nearly
identical for PET at temperatures above 120 °C.
3.3. Influence of the oxygen partial pressure during
metal deposition on AI/PET adhesion
For the A1/PET couples (Table 1), the increase of
oxygen partial pressure is directly correlated to an
increase of the adhesion and a decrease of the mean
grain size of the aluminium films. It has to be noticed
that for an oxygen partial pressure equal to 10 -4 mbar,
the aluminium film is amorphous and transparent. In
contrast, for a pressure lower than 10 -4 mbar the alu-
minium films exhibit a metallic appearance and no
aluminium oxide has been detected by electron diffrac-
tion in TEM studies. XPS results show that the failure
always takes place inside the PET skin (first 10
nanometres); this result indicates that the cohesive bulk
properties of the metal layer is not concerned during
the pealing test. Figure 5 shows the plane view TEM
micrographs of aluminium films evaporated under
different oxygen partial pressures.
TEM cross-sections have been prepared in order to
observe the metal/PET interface and the morphology of
1400 the aluminium films as a function of the oxygen pres-
/- ~ 1200~ sure. The adhesion of these single and multi layers is
A] /Cu 1000 :~ determined by the adhesion strength of the first evapo-
/ 800 r~ rated layer regardless of the thickness of the layer

=_o - 600 .~ (minimum first layer thickness: 2 nm). Figures 6(a) and
600-
, .... 4o0 6(b) show TEM cross-sections of these films. On these
< 400 - o .."/ - 200 < two micrographs, the (a) and (b) areas of the alu-
200 " 0 minium films are associated respectively with an alu-
70 4'06'o8'0 i6/, i'~o~ i ~ ih;~oo
PET temperature (°C) minium film deposited without oxygen present in the
(a)
metallisation chamber and with an oxygen partial pres-
sure equal to 10 - 4 mbar. Figure 6(a) (aluminium with-
out oxygen) shows the usual column-like structure of
35000 , . . ~, . . =, , . , , , . , . , , ~. , , ~, , , i , . , , . , , ~, . . 2200

/
TABLE 1. Evolution of metal/PET adhesion and the metal mean
grain size with oxygen partial pressure, during evaporation of Al
"~ 20000 -

'& 15000 - Oxygen pressure Adhesion Mean Locus of failure

,oooo / ; \ -,2oo (mbar) (g inch-i) grain size
(nm 2)
0 . 1 ~ ~ - - , ~ ' ~ ' ' , ' ~ , ~ " 800
20 40 60 80 100 120 140 160 180 200 no 0 2 80 4000 cohesive (PET)
PET temperature (°C) 2 x 10 -6 65 2000 cohesive (PET)
(b) 10 -5 1030 800 cohesive (PET)
Fig. 4. Correlation of (a) the dry adhesion and (b) the mean grain 10 -4 1230 amorphous cohesive (PET)
size of thermally deposited aluminium and copper thin films onto a
heated PET. AI (0.6/~ s -1)
 J. F. Silvain, J. J. Ehrhardt / An overview on metal~PET adhesion 233

A1

(a) (b) -

I 100 nm=

(c) (d)
Fig. 5. TEM plane views of AI thin films evaporated onto PET under different oxygen partial pressures (a) no 02, (b) 2 x 10.6 mbar, (c)
10.5 mbar, and (d} 2 x 10-4 mbar.

i~iii!iiiii!~iiii~i!!ii i
i ; i,i ,!iiiiii!i i i !i¸ I I
A1

(a) . . . . .

(a)

(b) iv

Fig. 6. TEM cross-sections of (a) AI thin film (150gin -1) (no

oxygen) and (b) trilayer AI thin film (130gin -1) (without, with
(10 -4 mbar 02) and without oxygen) (scale mark is equal to 50 nm).

the aluminium thin film evaporated onto PET where

the grain boundaries cross the entire thickness of the
metallic film. An example of trilayer aluminium films is
presented in Fig. 6(b). In the PET/a/b/a stacking (Fig.
6(b)), the two morphologies (column-like (a) and
amorphous like (b)) are evident. The interface between
each layer is smooth and sharp.
3.4. Influence of metal precipitation inside the polymer
on the metal~PET adhesion
3.4.1. PET/aluminium interface
Depending on the conditions during the film process-
ing, P E T with different structural properties can be
(b)
Fig. 7. Morphology and structure of (a) a low adhesion (80 g in -1)
aluminium thin film, (b) a low adhesion (50 g in -1) copper film.
produced [24]. Figure 7(a) shows a cross-sectional mi-
crograph of AI/PET samples prepared on a polymer of
low crystallinity (the crystallinity o f the PET depends
on the elaboration mode of this biaxially stretched
polymer). Diminished layer adhesion has been found on
this PET. Some spherical precipitates are observed in
234 J. F. Silvain, J. J. Ehrhardt / An overview on metal~PET adhesion
the polymer in the vicinity of the interface (a micro-
micro diffraction of one grain shows a {112} aluminium
orientation (A)).
3.4.2. PET/copper interface
Figure 7(b) shows the microstructure obtained when
copper is evaporated on a PET sample having a larger
degree of crystallinity. A layer adhesion similar to that
for the A1/PET sample is found. The morphology is no
longer columnar as on the A1/PET films but appears
granular. Extended slabs with a thickness of the order
of 30-40 nm and dimensions between 250 nm and
1250nm are found inside the PET in the Cu/PET
interphase.
4. Discussion
Chemical reactions at the metal/polymer interface are
usually discussed in terms of carbon-oxygen bond-
breaking of the carboxylic group giving rise to
C-O-Metal groups [25] or the formation of a charge
transfer complex, as has been claimed in metallised
polyimide (PI) samples [26]. Concerning the AI/PET
interface formation, the fast attenuation of the car-
boxylic carbon during the build-up of the first layer of
the metallic film suggests the presence of this functional
group at the interface. Furthermore, the first few layers
of the metallic film are oxidised in agreement with
previously published SIMS studies [13]. In fact, most of
the experimental evidence [24, 27] is consistent with the
bond-breaking model for this peculiar metal-polymer
system. With these ideas in mind, the improvement of
adhesion by a controlled oxygen partial pressure during
deposition and by heating the polymer at a temperature
higher than 100 °C can be rationalised. First of all, in
the absence of oxygen in the gas phase, an oxidised
interface will be formed using the oxygen of the poly-
mer, i.e. causing strong modifications of the near sur-
face. The consequence could be some decohesion of the
skin of the polymer and a moderate strengthening of
the adhesion. The presence of oxygen would prevent
this phenomenon and preserve the properties of the
polymer. Improvement in the adhesion coefficient with
the temperature of PET during deposition, can be
correlated with the structural changes in the polymer
occurring above the glass transition temperature of
PET (Tg ~ 80 °C). It has been observed in IR measure-
ments [28] that thermally-induced rearrangement of
polymer chains can occur and particularly reorientation
of the aromatic rings parallel to the surface. Con-
versely, an increasing number of carboxylic groups
which are in the plane of the benzene ring in PET, will
be present in the surface of the sample. This situation
could induce more favourable sites for the initial ad-
sorption of metallic atoms. In this case improved adhe-
sion would result for the metal exhibiting a higher
affinity for oxygen, i.e. for metals with the largest heat
of formation of the oxide. In addition, some other
temperature-dependent mechanism, such as water de-
sorption, which could release available active sites for
metal nucleation, or atomic diffusion of the metal into
the polymer, can also explain this tendency.
The major result of the TEM measurements on
metallisation of PET is that adhesion strength corre-
lates with the interface sharpness. These results are
consistent with the picture given for the metallisation of
PI [ 11, 26, 29]. When the chemical bonding is weak (for
instance, Cu/PET sample), the metal is free to diffuse
into the polymer layer and form clusters. On the other
hand, when the chemical interaction is stronger (for
instance, AI/PET sample), diffusion into the polymer is
hindered and there is little possibility of clustering.
Finally, it is important to note that the structure of the
polymer itself also plays an important role in a sense
that deposition of AI on PET having low crystallinity
gives spherical precipitates; when the same metal is
deposited under the same conditions on a polymer of
high crystallinity a sharper interface results.
Acknowledgment
We would like to thank Dr M. Koenig (Du Pont de
Nemours, Luxembourg) for providing the polymer
samples and for many helpful discussions.
References
1 K. L. Mittal, J. Vac. Sci. Technol., 13 (1976) 19.
2 Y. Novis, M. Chtaib, R. Caudano, P. Lutgen and G. Feyder, Brit.
Polymer J., 21 (1989) 171.
3 T. S. Sun, J. M. Chen and J. D. Venables, Applic. Surf Sci., I
(1978) 202.
4 J. F. Silvain, J. J. Ehrhardt and P. Lutgen, J. Adhesion. Sci.
Technol., 5 (7) (1991) 501.
5 J. S. Colligon and H. Kheyrandish, Vacuum, 39 (7) (1989) 705.
6 J. M. Burstrand, J. Vac. Sci. Technol., 15 (1978) 223.
7 J. M. Burstrand, J. Vac. Sci. TechnoL, 16 (1979) 1072.
8 N. J. Chou and C. H. Tang, J. Vac. Sci. TechnoL, A2 (1984)
751.
9 F. S. Ohuchi and S. C. Freilich, J. Vac. Sci. Technol., A4 (1986)
1039.
10 J. W. Bartha, P. O. Hahn, F. K. Le Goues, and P. O. Ho, J. Vac.
Sci. Technol., A3 (1985) 1390.
11 P. O. Ho, P. O. Hahn, J. W. Bartha, G. W. Rubloff, F. K.
Le Goues and B. D. Silverman, J. Vac. Sci. Technol., A3 (1985)
739.
12 N. J. Dinardo, J. E. Demuth and T. C. Clark, Chem. Phys. Lett.,
121 (1985) 239.
13 Y. de Puydt, P. Bertrand and P. Lutgen, Surf Interface Analysis,
12 (1988) 486.
 J. F. Silvain, J. J. Ehrhardt / An overview on metal/PET adhesion
14 J. F. Silvain, A. Picco, J. J. Ehrhardt and P. Lutgen, in E. Sacher,
J.-J. Pireaux and S. P. Kowalezyk (eds.), ACS Symposium Series,
No. 440, American Chemical Society, Washington, DC, 1990,
Chapter 33, pp. 453-466.
15 Y. Novis, M. Chtaib, N. Degosserie, J. J. Pireaux, R. Caudano, P.
Lutgen and G. Feyder, in E. Sacher, J.-J. Pireaux and S. P.
Kowalezyk (eds.), ACS Symposium Series, No. 440, American
Chemical Society, Washington, DC, 1990, Chapter 33, pp. 288-296.
16 Y. de Puydt, P. Bertrand, Y. Novis, M. Chtaib and P. Lutgen,
Colloque GFP, Le Mans, November 1988.
17 J. F. Silvain, C. L. Bauer, A. M. Guzman and M. H. Kryder,
IEEE Trans. Mag., 22 (5) (1986) 1296.
18 M. Alnot, B. Weber, J. J. Ehrhardt and A. Cassuto, AppL Surf
Sci., 2 (1979) 578.
19 J. F. Silvain, A. Arzur, M. Alnot, J. J. Ehrhardt and P. Lutgen,
Surf. Sci., 251/252 (1991) 787-793.
20 D. Briggs and M. P. Seah, Practical Surface Analysis, Wiley, New
York 1983, Chapter 9.
235
21 Y. Novis, M. Chtaib, R. Caudano, J. Vohs, J. J. Pireaux, P.
Lutgen and G. Feyder, Symposium on Metallized Plastics
Chicago, Illinois, October 9, 1988, p. 521.
22 M. de Koven and L. Hagans, AppL Surf Sci., 27(1985) 199.
23 P. Switft, D. Shuttleworth and M. P. Seah, Practical Surface
Analysis, Wiley, New York 1983, annexe, p. 437.
24 C. J. Heffelinger and K. L. Knox, Science and Technology of
Polymer Films, New York, 1974, Chapter 14.
25 Lj. Atanasoska, S. G. Anderson, H. M. Meyer III, Zhangda
Lin and J. H. Weaver, J. Vac. Sci. Technol., A5(6) (1987) 3325.
26 R. Haight, R. C. White, B. D. Silverman and P. S. Ho, J. Vac.
Sci. Technol., A6 (4) (1988) 2688.
27 N. J. Chou, D. W. Dong, J. Kim and A. C. Liu, J. Electrochem.
Soc., 131 (10) (1984) 2335.
28 G. A. J. Orchard, P. Spiby and I. M. Ward, J. Polymer Sci., 28
(1990) 603.
29 F. K. Le Goues, B. D. Silverman and P. S. Ho, J. Vac. Sci.
Technol., A6 (1988) 2200.