AD-A265 355

CONTROLLED INTERPHASES IN GLASS FIBER AND

PARTICULATE REINFORCED POLYMERS: STRUCTURE
OF SILANE COUPLING AGENTS IN SOLUTIONS AND ON
SUBSTRATES

FLHTSUO ISHIDA
DTIC
ELECTE
Departmentof MacromolecularScience
dUN2
Case Western Reserve University
Cleveland, Ohio 44106-7202 U 1993

ABSTRACT

The structure of silane coupling agents in solution and on solid

substrate is reviewed with special emphasis on the fundamentals of
structural development. Factors affecting the molecular weight,
adsorption behavior, and chemical bond formation are discussed.
Molecular aspects of the reinforcement mechanisms are discussed in
relation to the interfacial bond formation. Effects of surface treatmen6t-
on the rheological and hydrothermal properties of filled systems and
composites are described.

I DW19=t r; SVLTzii
iI MfLr' i _ _iý

App=oed tfar pubic rele~eE i z a

1696
G. Akovali (ed.), The Interfacial Interactionsin Polymeric Composites, 169-199.
0 1993 Kluwer Academic Publishers.Printedin the Netherlands.

9 6039
170

TABLE OF CONTENTS

1. Introduction to Composite Interfaces and Interphases

2. Introduction to Silane Coupling Agents

2.1. Structure of Silanes
2.2. Hydrolysis Reactions of Silanes
2.3. Condensation Reactions of Silanes
2.4. Silane Treating Solutions
2.4.1, Aqueous Treatment
2.4.2. Integral Blends

3. Structure of Silane on Solid Surfaces

3.1. Adsorption and Deposition of Silanes
3.2. Mode of Adsorption
3.3. Physisorbed and Chemisorbed Silanes

3.3.1. Chemisorbed Silanes

3.3.2. Interfacial Bond Formation
3.3.3. Physisorbed Silanes
3.3.4. Mechanical Properties of Silane Interphase

4. Structure of the Silane/Matrix Interphase

4.1. Copolymer Formation with the Matrix Resin

5. Reinforcement Mechanisms
5.1. Various Reinforcement Theories
5.2. Concentration of Interfacial Bonds
5.3. Adhesion of Thermoplastics to Substrates

6. Effect of Silanes on the Rheological Properties of Composites

7. Hydrothermal Stability of Silane Coupling Agents

8. Summary
9. Acknowledgment

10. References
 171

1. Introduction to Composite Interfaces and Interphases

It is now well-recognized that the interface and interphase play an important role
in the mechanical and physical properties of composite materials (1-9). The number of
studies dealing with the characterization of interfaces and interphases and their
influence on mechanical properties has increased dramatically over the past decade.
Among many areas of composite interface studies, the fundamental understanding of
the application of silane coupling agents has advanced substantially. This paper intends
to describe the structure and role of silane coupling agents on glass fibers and
particulate fillers. A special emphasis will be placed on the systematic description of
the understanding and controlling of the structure of silancs.
The terms interface and interphase have been sometimes used interchangcably.
However, an interface is a hypothetical molecular or atomic plane separating two
dissimilar phases and is seldom associated with an actual material unless there is a bond
passing through the separating plane and the structure differs from either of the phases.
On the other hand, interphaseis defined as an "interfacial region" whose properties are
similar to, but distinguishably different from, the bulk properties. Thus, the interphase
is associated with a finite thickness which depends upon the cause of the structural
difference from the bulk. The thickness of the measured interphase also depends
strongly on the characterization methods used despite studying the same interphase.
Accordingly, one must be careful to mention the characterization methods whenever
interphase thickness measurements are discussed. The definitions of the interface and
interphase are schematically illustrated in Figure 1.

S~ Intelfiae

Fiber

Inte fwse

Figure 1. Conceptual drawing of interface and interphase in fiber reinforced

composites.
172

The nature and thickness of the interphase are influenced by the substrate
surface and its specific interaction with the matrix resin. Representative effects of the
substrate surface are illustrated in Table 1.

Table 1. Effect of Substrate Surface on the Structure of Matrix Interphase

Chemical Effects
t Reaction with matrix
Catalytic activity
- Acceleration of polymerization
- Degradation of matrix
poisoningSCatalyst
Selective reaction with a certain component

Physicochemical Effects
0 Preferential adsorption of a catalyst
- Inhibition of cure
0 Diffusion of a surface treating agent
- Interpenetrating networks
0 Topochemical reaction
- Restriction of reactivity

Physical Effects
0 Nucleation of crystallites
-transcrystallization
0 Chain packing
-epitaxial crystallization
0 Restriction of mobility
-molecular orientation
0 Preferential adsorption of non-reactive component
An interphase may possess a sharp or graded boundary, examples of which are
listed in Table 2. For a sharp boundary, recopizing the filler/matrix interface or matrix
interphase/matrix bulk i: relatively straightforward. For a graded boundary,
determining the thickness of the interphase is not possible unless either an average
quantity or relative concentration reduction is defined to be the boundary of the
interphase. Determination of the concentration profile thus becomes an important task
to understand the structure of a graded interphase. A graded interphase is often useful
for diffzing 3 stress concentration by avoiding a sudden change of material properties.
The concept of an interphase is well-established. Yet, it can only be defined
through a collective description of many complex phenomena, and thus at the present
time, no single explanation or approach exists which enables the complete
understanding of the nature of any particular interphase. Therefore, it is essential to
characterize the interphase using many characterization methods prior to attempting its
structural control. If the interphaw su ucture can be controiied, tauonng the mechanical
 173

and physical pioperties of a composite through the knowledge of the structure/property

relationship becomes feasible. This systematic approach may be termed interphase
engineering. It is the purpose of this article to discuss the principles of silane structure
formation in solutions and on substrates rather than to present a summary of all
published work in this field. Therefore, papers published from our laboratory will be
reviewed whenever applicable, although there may be many other papers on the
subject.

Table 2. Structure of Matrix Interphase

Sharp Boundary

S Epitaxial crystal
Transcrystal
Interfacial reaction with matrix
Totally incompatible filler/matrix combination
- No surface treatment

Graded Interphase Structure

Diffusion of coupling agent
- Interpenetrating networks (IPN)
Preferential adsorption
Restricted mobility

2. Introduction to Silane Coupling Agents

2.1. Structure of Silanes
Silane coupling agents usually possess a dual functionality of the form, YSiX3,
where Y is the organofunctional group and X the hydrolyzable group. X is typically an
alkoxy group and sometimes Cl. Some representative silanes are listed in Table 3. The
structure of an organofunctional group is chosen so as to copolymerize with a
thermosetting resin or to be compatible with a thermoplastic resin.

2.2. Hydrolysis Reactions of Silanes Accesion For

NTIS CRA&I
Upon contact with water, the following reactions take place. DTIC TAB ]
Unannounced Q
RSi(OR')3 H20 RSi(OH) 3 + 3R'OH Justification
By
Distribution I
Availability Codes

. ,. ..... ....... . . .. . 7Dist Avail and /or

Soecwt
174

The hydrolysis of the trialkoxy groups takes place in a stepwise fashion. As the first
alkoxy group is hydrolyzed, the steric hindrance to the other two alkoxy groups
decreases. It is expected that the final alkoxy group is easier to hydrolyze than the first
alkoxy group. The hydrolysis of silanes must be carefully controlled to be effective.
Ordinary neutral silanes form oily droplets upon mixing with water. Thus, the
dissolution of the silane molecules into water is the kinetic limiting step. The formation
of the first silanol group increases the solubility of the silane into water. However, it is
not known whether the unhydrolyzed trialkoxysilane first dissolves into water with the
alkoxy groups hydrolyzed in a stepwise fashion or if the formation of the first silanol
group is the driving force for the silane to dissolve into water.
Table 3. Typical industrially used silane coupling agents

Organofunctional Chemical Formula

Group
Cationic styryl CH2=CHC6H4CH2NH(CH2)2NH(CH2)3Si(OCH3)3.HCI
Chloropropyl CI(CH2)3Si(OCH3)3
Cycloapliphatic epoxide
H2 )2 Si(O CH 3 )3 S(C
Diamine H2N(CH2)2NH(CH2)3Si(OCH3)3
Epoxy /Q,
CH 2CHCH 2 0(CH2 )3 Si(OCH 3 )3
Mercapto HS(CH2)3Si(OCH3)3
Methacrylate CH 3
CH 2 =h-COO(CH 2 ) 3Si(OCH 3 ) 3
Primary amine H2N(CH2)3Si(OCH3)3
Vinyl CH2=CHSi("OCH3)3

Usually, acidic water with a pH ranging from 3.0 to 4.0 is used to catalyze the
hydrolysis reaction. However, a glycidoxy-functional (epoxy-functional) silane readily
opens its oxane ring in this pH range. For this silane, pH 5-6 may be needed to
prevent the ring opening reactions. These neutral silanes may need to be hydrolyzed
for tens of minutes at room temperature for complete hydrolysis, provided that the
concentration of the silane is relatively low.
For an aqueous alcohol solution and other organic solution, the situation
changes due to homogeneous reactions. Because of this, a high concentration of silane
in water changes its behavior after certain degree of hydrolysis due to the increased
alcohol content produced as a byproduct of the reaction. Amino-functional silanes
readily dissolve into water and their alkalinity catalyzes the hydrolysis. For this
reason, aminosilanes utilize neutral water for hydrolysis. Hence, upon contact with
water, the hydrolysis will immediately take place in a matter of seconds. Since silanols
are unstable in an alkaline media, immediate condensation also follows. For a few
percent by weight of neutral silane solution, nearly all of silane molecules are
silanetriol. In contrast, the same concentration of an aminosilane yields a solution with
nearly one hundred percent oligomeric silanes. The content of the monomeric
aminosilanetriol dramatically increases if the silane concentration in water is a fraction
of a percent (10).
 175

2.3. Condensation Reactions of Silanes

Upon drying, or with time in solution, the silanol groups condense to form
siloxane groups, SiOSi:

R R
2RSi(OnH) HO-Si-O-Si-OH
0 0
H H

RSi(OH)3 R R R R R
HO-Si-O-Si-O-Si-O-Si-O-Si-OH
I
0 0 0 0 0
H H I H H
HOSiR
0
H

The simplified structures shown above are only to demonstrate that the molecular size
increases by condensing with silanetriols and oligomeric silanes. The actual structures
may be far more complex since the trifunctionality of the silane can lead to many
possible structures. A similar condensation reaction takes place with the surface silanol
and other hydroxyl groups of a filler and fiber, thus allowing the silane to bind
chemically to the surface. This is a typical reaction path for a treatment using a
prehydrolyzed silane solution. The rate of silanol condensation to siloxane depends
strongly on the pH, concentration and temperature of the solution, but weakly on the
structure of the organofunctional group.
The molecular weight of the condensed silane is also influenced by the same
factors. Another factor that is important for the silane coupling effect is the molecular
architecture of the siloxane networks. Caged structures and ladder structures are the
extreme cases of the different siloxane structures (11,12). The solution pH,
topochemical effects of the solid surface, the structure of the organofunctional group,
and the concentration of the silane are all believed to influence siloxane structure,
although no detailed study has yet been reported.
An indirect observation is shown in Figure 2 where Y-methacryloxypropyl-
trimethoxysilane (Y-MPS) is hydrolyzed and condensed in an aqueous alcohol solution
in the absence of a filler (13). The solvents were evaporated at room temperature to
simulate the typical treatment of a filler. The molecular weight was then measured by
size exclusion chromatography as a function of time. When the solution was adjusted
below pH 4.0, Figure 2 resulted showing a discrete increase of the molecular size until
the column no longer separated individual species. On the other hand, the solution near
neutral pH, as well as alkaline pH's, exhibited a quite different behavior. A
monomeric species existed until the solvents were mostly evaporated and suddenly a
large molecular weight species appeared as shown in Figure 3, unlike stepwise increase
176

of the molecular weight in an acidic media. This may be an indication of a ladder-like

structure or more open structure than the cage-like structure.

E 4ZC II,

120 hr

"0 h,
I h

20,000 4.000 800 150

MOleCulnr Weight Of Polystyrene Sto4ad$e,

Figure 2 SEC chromatogiams of the hydrolyzate of y-MPS after the solution

pH was adjusted to 3.6. Time from pH adjustment: (A) 1 h in solution, (B) 20
h during drying, (C) 40 h, dry, (D) 120 h, dry, (E) 420 h, dry.

4
( 20 hr

] 2.000 ISO

C. 40 hr

C. 40 hr

I II I
20.000 4.000 800 I50
Molecular Weight of Polystyrene Stanedards

Figure 3 SEC chromatograms of the hydrolyzate of y-MPS after the solution

pH was adjusted to 6.9. Time from pH adjustment: (A) 1 h in solution, (B) 20
h during drying, (C) 40 h, dry, (D) 120 h, dry, (E) 420 h, dry.
 177

E pH4O6

'I £.pld 6

0PH 32

PHH 60

2 C0 SpH B pH

A PH t3 A ON 53

390.000 35.000 2,000 i50

394000 .W,000 2.000 150
'dalecuier Weqht of Polystfree SOAnaords
MoI*jCjotr W•ar•h Ot Pe f rsE Stoadardl

A B
Figure 4 SEC chromatograms of ;he pl-adjusted hydrolyzate of y-MPS.
Figure 4A: (A) pH 1.3, (B) pH 2.0, (C) pH 2.6, (C) pH 3.2, (D) pH 4.0.
Figure 413: (A) pH 5.3, (B) pH 6.0, (C) pH 6.5, (D) pH 7.1, (E) pH 8.2, (F)
pH 11.9.

1.20
U

$ 1.00
2
0.80

'• 0.60

S0.40

0.20

S0.00 I I I I I
0.00 2.00 4.00 6.00 8.00 10.0 12.0 14.0
pH
Figure 5. Stability of silanol as a function of the pH of the original solution
from which the precipitates were prepared
 178

A similar solution was prepared with varying pH's and the solvents evaporated.
The molecular weight of the precipitated silane is illustrated in Figure 4 as a function of
the solution pH. The average molecular weight of the silane strongly depended on the
initial pH of the solution. In both acidic and alkaline pH's, the molecular weight was
equivalent to a few thousand of the polystyrene standard. On the contrary, the neutral
pH's yielded quite large molecules.
In summarizing the above two experiments, it is possible to suggest a change in
the condensation mechanism around pH 4 to 5. The mechanisms in the neutral and
alkaline pH's are similar, except that the silanol group in an alkaline media forms a
silanolate, SiO', which is resistant to condensation. Thus as shown in Figure 4. small
oligomers resulted at higher pH's. Similar structures can be observed on a solid
surface due to the pH effect.
The stability of silanol depends highly on the pH of the environment. Figure 5
shows the amount ef the residual silanol of the precipitated silane molecules from
aqueous solutions which are adjusted to respective pH's. It shows that the silanol is
most stable around pH 3 and the stability quickly decreases around pH 4-5 (13).

2.4. Silane Treating Solutions

2.4.1. Aqueous Treatment

Ordinarily, silanes are applied to a substrate as monomeric, hydrolyzcd species.

The effectiveness of a silane is better when applied as monomeric silane rather than
condensed oligomers. These silane films are very thin with a thickness on the order of
several nanometers and, thus the strength of the film is rather insignificant in twrms of
the overall mechanical properties of a composite material. However, a silane primer is
applied as a rather concentrated, condensed species, forming a film of a few hundred
nanometers. Thus, the mechanical properties of the film itself can contribute to the
composite properties. For this type rf silane primer as wedl as the structure of silane on
a solid surface, consideration of a kinetic effect and a taermodynamic effect is very
important.
For example, an aminofunctional silane can condense almost instantaneously
after hydrolysis due to the high pH of tLe amine group, leading to a kinetically
controlled structure which is thermodynamically unequilibrated. Provided that a
sufficiently high concentration was used, the solution t..ms into a gel. With time, in
the presence of water, the catalytic action of the amine reorganizes the siloxane
structure by rehydration and condensation reactions, and redissolves the silane. Thus,
the time factor is also important when silane structures are to be studied.
When an aqueous solution of a silane is to be applied, the concentration used is
on the order of a fraction of a percent. In this range, a concentration exists at which
isolated silan, monomers can form associated molecules through hydrogen bonding of
the silanol groups. This concentration may be termed onset concentration of
association(14). The associated silanes adsorb differently from the monomeric silanes
as described in the following section. Furthermore, the amount of physically adsorbed
silanes, also discussed in the following section, vary depending on the concentration of
the silane treating solution. The solution with a concentration lower than the onset
concentration of association yields less physisorbed silanes by decreasing the defect of
the adsorbed layers.
2.4.2. Integral Blends

In addition to being used as a substrate pretreatment prior to mixing with a

matrix resin, silanes can also be used as integral blends. This method eliminates the
steps of pretreatmenk, drying, and repulverization by incorporating the unhvdroivzed
silane directly into an uncured resin. However. the integral blending method usually
requires a greater amount of silane than direct treatment due to its lower efficiency of
silane utilization. The integral technique rLA.ies on the preferential adsorption of the
silane onto the substrate. Immediateiy following blending, many of t"'- reactive groups
may remain completely or partially unhydrol',zed. Hydrolysis may proceed with time*
if sufficient water is available from either the resin or ,hc substrate surface. -)Ln
additional source of water is the adsorbed water on the substrate surface. Direct
chemical reaction with the substrate can also proceed via the following exchange
reaction.

sgi/" ;iOH iOH

A;;iOH + (CtH30) 3 SiR iiO.SiR + CH3014

/OCH
3
;iOH iO

Substrate
Catalysts such as amines or heat treatment helps accelerate the reaction since the
reactivity of the above reaction is lower than the condensation reaction of silanol
groups. Because there are few silanols available, the silane in this treatment method
tends to be near a monomolecular equivalent. A similar situation is also observed when
a filler is heated in a silane solution of a hydrophobic solvent such as toluene.

3. Structure of Silane on Solid Surfaces

3.1. Adsorption and Deposition of Silanes

The adsorption process of a silane molecule onto a substrate surface is
influenced by the nu nber of functional groups which are capable of strongly interacting
with the surface via hydrogen or ionik bonding. Although it is difficult to distinguish it
from the adsorption phenomenon, deposition of silane molecule will occur when the
solvent is a poor solvent for the silane monomer or oligomers. hIs important to not2
that the solubility changes as a function of the silane molecular weight. For example,
water dissolves the silanetriol of the methacryl-functional silane ('-MPS). However,
the oligomers greater than the trimer precipitate from the solution. Adsorption is
related to the active participation of the adsorbent toward the substrate usually with a
special mechanism to anchor the molecule to the surface. Studies dealing with the
adsorption phenomenon of a silane coupling agent have been reported by Nishiyama et
al. (15,16).
 180

On the other hand, deposition is the process in which the adsorbent molecule is
forced out of the solution and onto the substrate surface. Deposition and orientation of
methacryl-functional silanes have also been studied as a function of the length of the
aliphatic spacer group (17). When the solubility of the silane is poor or has become
poor by the structural or concentration changes due to evaporation of the solvent,
deposition may take place. Although both of these processes lead to the accumulation
of silane layers on the substrate surface, it is important to recognize that the molecular
architecture of the condensed silane is different for each situation.
The orientation of a silane molecule influences the molecular architecture of the
condensed silane and the chemical reactivity of the silane with the substrate surface as
well as the matrix. Silanes usually possess multiple sites with quite different
electronegativity or hydrophilicity on one molecule. The pH of the solution influences
the state of charge densitv on the functional group as well as the substrate surface.
Hence, the pH of the solution strongly influences the orientation of the silane either
attracting or repelling a certain portion of the silane molecule. Under ordinary treatment
conditions, silanols are believed to adsorb onto the hydroxyl-covered surface.
However. if the silane molecule possesses another ionic portion that can compete with
this process, it is possible for the molecule to adsorb upside-down (3).

3.2. Mode of Adsorption

Silanes adsorb onto solid surfaces differently depending upon the roughness of
the surface with respect to the dimension of the silane molecules. For example, a high
surface area silica typically adsorbs a few monomolecular layer equivalents whereas a
smooth surface such as glass fibers adsorb many layers. The thickness of the silane on
the glass fiber increases monotonously until the onset concentration of association at
which the hydrolyzed silane molecules start forming hydrogen bonded species (14).
Thus, the orientation of the associated silanes is disturbed and the thickness buildup
slows. The disturbance of the packing is schematically shown in Figure 6.
The silane molecules tend to adsorb head-to-head in an extreme case where
relatively rigid organofunctional groups such as cyclohexyl and phenyl groups are on
the silane molecule. The crystal structure of cyclohexylsilanetriol exemplifies this
extreme case (18). It is these silanes that form relatively stable silanetriol crystals due
to good packing. In these crystals, silanols are hydrogen bonded in a bifurcated
manner as shown below, meaning that the crystal possesses a rather disorganized
hydrogen bonding structure.

•-Si- ,rd
J0 Si •

The packing of the silane molecules in the coupling agent layers strongly depends on
the structure of the organofunctionality. This trend of organized packing relaxes as the
flexibility of' the organofunctional group increases, resulting in an increased free
volume. An important ramification of this statement is that the degree of penetration
of the matrix molecules into the coupling agent layers is highly influenced by the
molecular packing. It is interesting to note that many useful coupling agents belong to
this flexible organofunctional type.
 I81

00%
00 oo 00
0 00
00 0 00 0 0
0?4F00
0o0-o
/ 0
F40o0o

A B

Figure 6. Schematic representation of the adsorption of silane coupling agent

molecules onto glass fiber surface. A) The silane solution is a low
concentration below the onset of association, and B) the silane solution is a
high concentration above the onset of association. Open circle indicates isolated
silanes whereas the closed circle denotes hydrogen bonded or oligomeric
silanes. Note the disturbance in packing is created at a high concentration.

3.3. Physisorbed and Chemn;.-,rbed Silanes

3.3.1. Chemisorbed Silanes

Silanes can form many structures due to their tri-functional condensation
mechanism. Monomeric silanetriol can directly react with the surface of a filler and
fiber. Silane oligomers can also react with the surface as some unreacted silanol
groups remain in the oligomers. The notion of physisorbed and chemisorbed silanes is
used based on the reaction with the substrate surface. Chemisorbed silanes are the
ones that possess at least one covalent bond with the substrate. Within the
chemisorbed silanes, further distinction may be made between tightly chemisorbed
silanes and loosely chemisorbedsilanes. Tightly chemisorbed silanes use many silanol
groups for covalent bonding with the substrate surface whereas the loosely
chemisorbed silanes use only a few silanols to bond to the surface. An extreme case of
this loosely chemisorbed silane is the oligomers or polymers with one covalent bond to
anchor the entire molecule to the surface. Both silanes do not desorb when washed
with an anhydrous organic solvent.
3.3.2. Interfacial Bond Formation
Although chemical bonding of a silane with the substrate surface is expected
from the knowledge of the traditional chemistry, few direct observations of the
interfacial bonds have been reported. This is because the interfacial bond, Sisubstrate-
O-Sisilane bond, is very similar in nature to the Sisubstrate-O-Sisubstrate and Sisilane-
O-Sisilane. Using a lead oxide as a model filler, this difficultv can be circumventcd
since the interfacial bond is now a unique Pb-O-Si against the Pb-O-Pb bond of the
substrate and the Si-O-Si bond of the silane. Figure 7 shows the infrared spectrum of
the interfacial bond as a function of the surface coverage of the silane from one fifth of
a monolaver to several monolaver equivalents (19). Notice that the intensity reached
the maximum at the third spectrum where the coverage is approximately a
monomolecular equivalent. A theoretical calculation of the Pb-O-Si frequency using a
harmonic oscillator also indicates a good agreement with the observed frequency.
Furthermore, this band was not observed either in the silane condensate or in the
substrate as the absorbance contribution of the organic functionality and the substrate
have been subtracted.

A(R=) - (.OIS

F.

-7N~ A.

WAVENU149ERS (cu ')

1
Figure 7. Diffuse reflectance infrared difference spectra from 100 to 910 cm-
of only the plumbosiloxane contribution near 965 cm- 1 . Spectra were obtained
using the digital subtraction method and are not scale expanded. The adsorbate
concentrations and measured peak to baseline intensity differences in Kubelka-
Munk units are respectively, (A) 0.40 mg/m 2 and 0.005, (B) 0.80 mg/m 2 and
0.009, (C) 1.50 mg/m 2 and 0.018, (D) 2.00 mg/m 2 and 0.015, (E) 4.00
mg/m 2 and 0.016, (F) 6.00 mg/m 2 and 0.014.
 In addition to the interfacial bonds, a quantitative observation of the inter-silane
bonds can be made. Figure 8 exhibits the intensity of the inter-silane bonds as a
function of the surface coverage (20). Two observations are especially noteworthy.
First. the intensity of the inter-silane bonds was near-zero until certain surface coverage
was reached despite the presence of the silane could be observed. This is an indication
that the silane did not adsorb as an island but rather adsorbed separately in a random
fashion so that the inter-silane distance is too far to form a siloxane bond. As the
surface coverage increases, the distance becomes sufficiently short to form the siloxane
bond. The second observation is the presence of a plateau near the surface coverage of
the 0.5 mg silane/g substrate. If the silane molecules are forming the second and
higher layers randomly, instead of first completing the bottom layer and then the
second layer, the intensity increase is expected to be monotonous. It seems that during
the plateau, the same process as the first layer is repeated. The adsorption is random in
position thus inter-silane bond formation is retarded. As the coverage of the second
layer becomes high, the inter-silane bonds again started forming. This process may be
repeated for the third layer and beyond. As the number of layers increases, the
registration of the orientation becomes more di3arrayed, and it becomes more difficult
to observe the plateau discussed above. However, a much smoother surface like mica
showed the same phenomenon to the third and possibly the fourth layers (21).
Understanding the mode of surface coverage is important because the interaction of the
treated filler with the matrix is strongly influenced by the way the silane covers the
surface.

1.40

1.20

S1.00

S0.80

, 0.60

S0.40

S0.20

0.00
0.00 0.500 1.00 1.50 2.00
Concentration (mg/g substrate)

Figure 8. Intensities of the infrared band due to the intersilane bonds as a

function of the amount of silane on a substrate
 184

3.3.3. Physisorbed Silanes

Those silanes which do not possess any covalent bonds with the substrate
surface are termed physisorbed silanes. The physisorbed silanes may be small
oligomers or large polymers. The unreacted silanol groups on the molecules must be
stcrically hindered such that they are not readily available for condensation with the
substrate surface or with the chemisorbed silane. Quite often, oligomers take the form
of cyclic silanes where concentration of the unreacted silanol group in the molecule is
low despite the small molecular size. A perfect cubical octamer for example does not
have any silanol groups although the molecular weight is less than a few thousand.
These physisorbed oligomers can be easily distinguished by washing with an
organic solvent. However, high molecular weight physisorbed silanes are not as easily
distinguished as the small oligomers since the rate of desorption is much slower than
for the small oligomers. Many silanol groups, though they may be sterically hindered
for covalent bond formation, can form hydrogen bonds with the substrate surface or
chemisorbed silanes. These hydrogen bonded silanols are in a dynamic equilibrium in
terms of their interaction with the substrate. Thus, when a driving force to desorb
exists, the hydrogen bonds are gradually broken and, with time, even a large
physisorbed molecule desorbs. If a small amount of water exists in the solvent used,
water can hydrolyze the siloxane bonds that were used to chemisorb the silane molecule
to the substrate. The time scale for desorption and hydrolysis of the siloxane bonds
might be comparable for a large, loosely chemisorbed molecules. For small oligomers,
it is not important since the time necessary to hydrolyze the siloxane bonds is much
longer than the extraction time.
The structure of the physisorbed silane molecules tends to be cyclic oligomers,
and their molecular weight strongly depends on the pH of the filler (22). The pH of the
filler is conveniently measured by the slurry pH method where a certain amount of filler
is boiled in water for a specified time and the pH of the liquid measured. This method
provides composite information about both the solubility of the selective surface
components as well as the concentration of the surface acid/alkaline centers. This
slurry pH directly relates to the molecular weight of the condensed silane rather than the
concentration of the surface acid centers alone, since the solubility of the surface
species and the concentration of acid centers both contribute to the pH of the silane
solution that is on the filler surface.
As in the case of the aforementioned simulation shown in Figure 4, the
molecular weights of the physisorbed silanes are small at both acidic and alkaline pH's
whereas they are large near neutral pH's (23). In an acidic media, the silane oligomers
are small because of the tendency to form cage-like molecules while in an alkaline
media the silane oligomers are small due to the termination of the molecular weight
growth by the silanolate group. Thus, the molecular architecture of the siloxane chains
in these oligomers with similar molecular weight is different. Since the diffusion of the
physisorbed silane molecules into the matrix resin depends on the molecular weight
when a silane treated filler or fiber is mixed with the resin, it is essential to understand
the nature of the physisorbed silane. The physisorbed silanes often contain residual
silanol groups and these groups bridge physisorbed molecules via hydrogen bonding.
Thus a desorption process is dependent not only on the molecular weight but also on
the molecular architecture of the physisorbed silane.
Once the basic molecular architectural trend is predestined by the effect of the
substrate surface, a heat treatment only accelerates the completion of the trend rather
than creating a new class of structure (22). Figure 9 and 10 shows SEC
 185

chromatograms of Y-MPS hydrolyzates collected from the surface of a filler and from
the precipitate of a silane solution without a filler. Both chromatograms of the initial
samples are similar in the molecular weight and its distribution. However, after a mild
heat treatment at 80 °C for 10 hr, the silane from the filler surface stays almost the same
size whereas the silane from the precipitate polymerized to a size such that it can no
longer dissolve in the tetrahydrofuran (THF) carrier liquid. The peak near the
molecular weight of 300 may be an additive present in the initial silane.
Consequently, if the molecular architecture of the siloxane networks needs to be
changed, it is necessary to change the solution pH of the silane, treat the filler by an
acid or alkaline material to control the slurry pH, or add another silane which has
different tendency toward forming siloxane networks. For example, a silane with a
flexible organofunctional group such as Y-MPS tends to form a cyclic structure wherein
a rigid functional group leads to a more open structure. Hence, if a larger molecular
weight Y-MPS condensate is desired, it can be mixed with a rigid silane like
phenyltrimethoxysilane or vinyitrimethoxysilane. The molecular weight changes for
mixed silane systems are plotted in Figure 11 where 10 mol % of the second silane has
been added (22). It seems that dilution of the organofunctional group by the second
silane does not grossly affect the effectiveness of the first silane so long as the
composition of the second silane does not dominate. Localization of silane near the
substrate surface can be achieved either by improving chemical bonding of the silane to
the substrate or by reducing the solubility to the matrix by increased molecular weight.

A. Surface

I I I
9000 800 250
tM t,"A of oy Sinr*
nluW

Figure 9. SEC chromatograms of the condensates of y-MPS: (A) extracted

from the surface of particulate clay, (B) dried in bulk.
 186

8, Bulk 80C OhrS

A. Suface
80C MOtnr

II

Mooma W& of %"vewuyrn

SPsOW'e,

Figure 10. SEC chromatograms of the silane shown in Figure 9 except that the
condensates were heat treated at 800 C for 10 h.

C.

a5.
A,

Figure 11. SEC chromatograms of of the bulk hydrolyzate of ,-MPS

containing 10 mol % of a second species and heated at 80o C for 85 h: (A) 100
mol % --MPS, (B) with 10 mol % tetramethoxysilane, (C) with 10 mol %
vinyltrimethoxysilane, (D) with 10 mol % isopropyltridioctylpyrophosphato-
titanate.
 187

3.3.4. Mechanical Properties of the Silane Interphase

At the present time, no in-situ measurement of the mechanical properties across
an interphase containing physisorbed silane has been made. However, a simulation of
the mechanical properties can be made by preparing copolymers of precipitated silane
oligomers with a matrix resin. Figures 12 and 13 show the flexural modulus and the
flexural strength of the copolymers of Y-MPS and unsaturated isophthalic type
polyester resin as a function of the silane concentration (24). Beyond 50 % silane
content, phase separation was observed and thus excluded from the study since a valid
comparison could not be made. However, in reality, regions with such compositions
do exist and indeed the phase separated area might be observed. Phase separation is
also observed with the copolymers of Y-aminopropyltriethoxysilane (Y-APS) and an
epoxy resin (diglycidylether of bisphenol-A) (25). The flexural modulus of the y-MPS
precipitate/polyester copolymers slightly increased until 20 % by weight silane and
decreased beyond this composition. On the other hand, the flexural strength
monotonously decreases at all silane compositions. This is a typical behavior of anti-
plasticization. It is interesting to note that for the y-MPS/unsaturated polyester, any
region with physisorbed silane is mechanically weaker than the bulk matrix.

4
- S

o 5'

o 5

00

0 , 1

0 20 40 60 80 100
Sliane Concentration (wt%)

Figure 12. Flexural modulus of the copolymers of the precipitates of y-MPS

hydrolyzates and unsaturated polyester resin.

j __ _ _ _ _ _ __ _ _ _ _ _
 188

I2

I00

4 0
-. 0

Ih.

0 20 40 60 s0 t00
Sliane Caoceetrettan (wt %)

Figure 13. Flexural strength of the copolymers of the precipitates of y-MPS

hydrolyzates and unsaturated polyester resin.
' Provided that it is overly simplistic, a conceptual mechanical property profile
across the interphase is schematically illustrated in Figure 14 using the expected
diffusion of the silane and the observed mechanical properties of the materials with
various compositions. A mechanically weak portion in the diffusion region of the
physisorbed silane is illustrated and the gradual increase of the strength over distance to
the bulk matrix is also incorporated. Thus, if the physisorbed silanes are removed from
the treated reinforcements, stronger composite materials can be manufactured. Indeed,
such observations have been reported (26,27).

Chemisorbed Physisorbed
silane si lone

/-- •00o O oo
Glass
fls 0 )0 Resin
matrix

)0o

Distance

Figure 14. Conceptual mechanical property profile across the glass fiber/matrix
interface.
 189

4. Structure of the Silane/Matrix Interphase

4.1. Copolymer Formation with the Matrix Resin

The chemical reaction of the organofunctional group and the matrix is, in
principle, the same as any other organic chemistry. However, the availability of the
organofunctional group may be hindered if the silanes are tightly packed. Restricted
molecular mobility, including the preferential interaction with the substrate surface,
reduces the reactivity of the organofunctional groups. Due to the wide distribution of
possible structures of the silane molecules in the silane layers, a systematic study of the
silane/matrix reactivity is very difficult, although some attempts as follows have been
made. Figures 15 and 16 show NMR Tlo relaxation measurements of the
aminosilane/epoxy copolymers at different silane contents (25). The different slopes of
each carbon relaxation process, and thus different Tip values, indicate that the system
is phase separated. However, the composition shown in figure 16 shows the same Tip
values for all carbon atoms, suggesting that the system is homogeneous. The
heterogeneity detected by the NMR method is said to be smaller than 1 nm in size.

o 158 PPM

60-- "& 144 PPM

x 42 PPM
> 5a 59 Ppm
I-.-I~p
7-n 19 PPM
Z' 5o.

z45-
Z
40-

0
35- ]

:30 - i z z i. .
0 4 6 10' 12' 14 16

CONTACT TIME (MSEC)

Figure 14. Tip measurements of the copolymer of 60 mol % y-aminopropyl-

triethoxysilane and 40 mol % epoxy.
In spite of the altered reactivities of surface species, the general trend of
copolymerization still holds. Thus, it would be helpful to know the copolymerization
parameters, r and Q factors, of the organofunctional group and the matrix functional
group. Depending on the r and Q values, the silane and the matrix interphase may form
an alternating copolymer or terminal reaction, leading to a quite different interphase.
190

75
o 158 Ppm
70 " & 144 ppm
65 Xx 42 PPM

Sao o 59 PPM
I- • 19 PPM
_z55o

_jz 45

40

35

30! I I 1
0 2 4 6 8 iO 12 14 16
CONTACT TIME (MSEC)
Figure 15. Tip measurements of the copolymer of 30 mol % y-aminopropyl-
triethoxysilane and 70 mol % epoxy.

5. Reinforcement Mechanisms

5.1. Various Reinforcement Theories

Historically, many reinforcement theories have been proposed. Those include
the chemical bonding theory (28), restrained layer theory (29), deformable layer theory
(30), and coefficient friction theory (31). However, only the chemical bonding theory
could sufficiently explain the observed results. However, the chemical bonding theory
alone is not adequate to explain the necessity of more than a monomolecular equivalent
of silane for optimum composite strength. Thus, this concept is coupled with
interpenetrating network theory (31,32). These theories have been developed primarily
for thermosetting resin composites. Thermoplastic-matrix composites rely on different
mechanisms.
If no chemical bonds exist, as is often the case in thermoplastic-matrix
composites, it is helpful to have a hydrophobic matrix or semicrystalline matrix to
reduce the number of water molecules diffusing to the fiber/matrix interface. It is
believed that isolated water molecules are not as harmful in causing permanent damage
to the composite properties as the condensed water. Even isolated water molecules can
act as a plasticizer and reduce the strength of the matrix, although such property
changes are often recoverable upon drying. This is in contrast to the interfacial damage
caused by the condensed water where an enormous osmotic pressure causes debonding
at the fiber/matrix interface.
 191

5.2. Concentration of Interfacial Bonds

It is interesting to consider the minimum number of interfacial bonds which is
necessary to provide sufficient composite strength. Very few papers are available on
this subject. Figure 17 shows the strength of a composite with coupling agents
specially designed to evaluate the number of interfacial bonds (33). A methacryl-
functional silane is copolymerized with styrene in different concentrations in order to
vary the average molecular weight, Mn, of the inter-pendant groups. The flexural
strengths of the dry and wet composites are plotted as a function of the Mn. Provided
that this is not exact evaluation of the interfacial bond density, qualitative information
can still be obtained. Both the dry and wet flexural strengths of the composite do not
markedly change until the Mn reaches approximately 1000. The dry strength gradually
reduces as the Mn increases implying that compatibility alone can provide substantial
dry strength. On the other hand, the wet strength decreases suddenly beyond Mn =
1000. Thus, chemical bonds are essential to provide permanence of the bond in a
humid environment.
Chemical bonds can maintain bonding even in the presence of small hydrophilic
molecules such as water and alcohol which compete for the hydrophilic reinforcement
surface with the silane. Physical bonds including van der Waals' force, ionic bonds,
and acid-base interactions are quite adequate to provide good dry strength but these
effects can yield misleading results if the wet strength is also of interest.

0•0

C
"I--
4-

:=3"
0 Wet
U.

0 2 4 6 8 10
Average Molecular Weight
Figure 17. Flexural strength of glass cloth reinforced polystyrene with
polymeric silane-treated glass: the silane to styrene ratio was varied so that the
molecular weight of the polystyrene between two silane molecules become from
500 to 10,000 as shown as average molecular weight.
192

5.3. Adhesion of Thermoplastics to Substrates

Another adbesion mechanism for thermoplastic-matrix composites is the
formation of transcrystalline layers around the reinforcement fibers. A transcrystal is a
collection of columnar crystals growing transverse to the fiber axis. When a massive
nucleation at the fiber surface occurs, the initial spherulitic crystals soon touch one
another leaving only the direction transverse to the fiber axis to grow. This form of
morphology is considered useful for the adhesion of a thermoplastic resin to the
reinforcement fibers. However, the improved adhesion might be offset by the residual
interfacial stress caused by the crystallization shrinkage. At the present time, the
overall benefit of the transcrystalline morphology to composite properties is therefore
poorly understood, although there are some papers showing the benefits of this type of
morphology to the mechanical properties of composite materials.
Transcrystals can be formed by many methods as long as massive nucleation at
the fiber surface is promoted. Some of those methods include an applied stress.
epitaxial crystallization, coating of the fiber by a nucleating agent or liquid crystalline
material. All these variations complicate the study. It is thus important to consider the
thermodynamics of the interphase. Until recently, few quantitative studies of
transcrystallization have been reported due to the difficulty in applying the
heterogeneous nucleation theory. Heterogeneous nucleation theory allows one to study
the interfacial energy difference function, Ao, of the fiber/matrix interface by measuring
the growth rate of the transcrystal layer and the nucleation rate. It is straightforward to
measure the growth rate of the transcrystal with many composite systems. However,
determination of nucleation rate is very difficult, especially for reinforcements such as
carbon fibers. Even for transparent fibers, a massive appearance of nuclei makes such
measurements tedious. Furthermore, the nucleation rate is thermal history dependent.
For these reasons, few of the papers dealing with transciystals reported AO values.
Recently, the induction time approach has been proposed as a new method to study
transcrystallization (34,35). When an isothermal heterogeneous crystal growth takes
place, a threshold period exists when the formation and remelting of the precursors of
nuclei takes place. This threshold period is called the induction time, t i. The rate of
nucleation, 1, is inversely proportional to the induction time, and thus the following
relationship cxists.

I-t i = C (constant)

By using t i, which is determined from the crystal growth kinetics, one can eliminate
the use of I, an aforementioned difficult quantity to measure. At the present time, no
quantitative study of the effect of silane structures on the transcrystal formation has
been reported.

6. Effect of Silanes on the Rheological Properties of

Composites
Although the rheology of silane-treated filled systems has been reported (36-
38), few reports deal with the silane structure/theology correlation.
The viscosity of a filled system is usually much higher than a non-filled system
mainly due to the formation of filler aggregates. The aggregated filler shows a higher
 193

apparent filler fraction, Oa, than the true filler fraction, 0, because the interstices of the
fillers are not easily accessible space for the resin to flow. The viscosity is a nonlinear
function of the filler fraction, and the viscosity increases much more rapidly at higher
filler fractions, especially near the closest packing fracticn, Orm. A drastic viscosity
reduction of a highly filled system can be achieved by breaking up the aggregates of the
filler. This can be achieved by applying a shear or extensional force which
mechanically separate the aggregates. The efficiency of the sepiration depends on the
deformation rate. An alternative method is to change the surface energetics of the filler
so that the filler/matrix interaction force owrcomes the aggregation force.
"Theaggregation force might be due to the bridging effect of water molecules,
direct hydrogen bonding of the surface hydroxyl groups, or ionic interactions. All of
these effects can be reduced by treating the filler surface with silane coupling agents.
Silanes can compete for, and react with, the surface hydroxyl groups and cover the
filler with organo-functional groups. If the organo-functional group has a hydropiilic
group, it tends to bend :uward the hydrophilic filler surface and hence to expose the
hydrophobic portion of the group. Since the van der Waals' force between the organo-
'§unctionai groups is similar to that between the silane and resin, a small shearing force
can easily break up the agglomerated filler particles.
In addition to the effects caused by the chemically bound silanes, a physisorbed
silane can further reduce the viscosity. A silicon fluid is a good lubricant or a release
agent. It has been previously discussed that a filler surface tends to produce cyclic
oligomers of varying molecular sizes. An acidic or alkaline filler especially produces
silane oligomers with a molecular weight of a few thousand. The amount of
physisorbed silane varies depending on the concentration of the silane treating solution.
In general, the higher the concentration of the silane in the treating solution, the more
the physisorbed silane. This is especially true in the treatment method used for
particulate fillers. Fillers tend to form a cake if a sluny of the filler/silane solution was
dried. In order to avoid this additional processing, a treatment method termed an
integral blend is adopted where a rather concentrated alcohol aqueous solution of a
silane is sprayed onto a tumbling filer. Thus, the filler does not completely wet and
immerse itself in the silane solution.
A mixture of a silane-pretreated filler and a polymer exhibits complex viscosity
behavior. In addition to the breakup of agglomerates, silanes can interact physically
and chemically with the polymer matrix. A reactive silane, such as an azide -functional
silane, can form chemical bonds with aliphatic polymers and crosslink the polymer
chain forming a matrix interphase around the filler. Polymer melts are usually kept at
elevated temperatures and thus ordinarily unreactive c lanes can form bonds with the
polymer. Polymers also cleave, forming reactive polymeric free radical chain ends.
These chain ends can graft onto silanes. Thus, the viscosity increases, counteracting
the viscosity reduction by the break-up of the agglomerates. Another influencial factor
is the acid-base interaction of the organo-functional group with the functional group of
the polymer chains. Strong acid-base interactions can act as physical crosslinks
modifying the rheological properties of the interphase as the silanes are multifunctional
by the oligomeric or polymeric structures. Physisorbed silanes can diffuse out of the
filler/matrix interface and the thickness of the physically crosslinked matrix interphase
increases. As a result, the viscosity would also increases.
When the integral blending method was used to treat a filler surface and the
amount of added solution was varied to change the amount of the silane coverage,
peculiar results were observed. As the amount of the added solution was increased, the
quantity of the deposited silane monotonously increased as expected. However, this
 11)4

increase in the deposited silane halted suddenlv for the quantity of silane solution where
the filler was completely wetted by the solution as shown in Figure 18. It is likely that
the silane molecules may have coevaported with the alcohol molecules. Accordingly,
there is an upper limit of silane that can be deposited by the integral blending method.
It is not clear wether this phenomenon applies to a compound such as an amino-
functional sle.ne which tends to form oligomers quickly by the catalytic action of the
amine grour
A highly filled system is of great technological significance in the ceramic,
artificial marble, and other industries. Due to the very high filler contents, the viscosity
of the filled system is usually extremely high to the extent that the mixture of initial
ingredients is dough-like in consistencv, even if a silica filler is mixed with a such a
low viscosity material as methyl methacrvlate. The initial viscosity of the mixture can
be as high as above 100 Pa.s (100,000 cp). If a silane is added and the mixture is ball-
milled, the viscosity drops in time almost three orders of magnitude to below 0.1 Pa.s
(100 cp), and the mixture can be readily poured (39). The viscosity reduction process
is a function of the surface coverage of the filler by the silane. The complete coverage
of the filler surface is necessary to observe low viscosity. Since this type of treatment
is done in an organic solvent without any water added, the rate of hydrolysis is very
slow. The consumption of the silane alkoxy group relies mostly on the water adsorbed
on the filler surface and the aforementioned surface silanol/alkoxy group exchange
mechanism. Thus, the addition of an alkaline catalyst greatly accelerates the chemical
reaction of the silane with the filler.
Ordinarily, in the aqueous treatment of fillers and fibers, silane molecules are
almost completely hydrolyzed in the solution prior to mixing with the filler and fiber.
Thus, the silanol groups of the silane can be a driving force for adsorption onto the
substrate surface. In the case of an anhydrous organic system, this type of adsorption
mechanism may be modified. Due to the limited amount of water molecules, only one
of three hydrolyzable alkoxy groups may be hydrolyzed and the incompletely
hydrolyzed silane may be adsorbed. These incompletely hydrolyzed groups may be
hydrolyzed with time on the filler surface. In the other mechanism, when an
unhydrolyzed silane molecule encounter the surface silanol groups, the aforementioned
exchange mechanism shown below takes place, releasing an alcohol molecule and the
silane becomes part of the solid (40).

"NSiOH + (CH 30)3SiR - = Si-O-Si(CH3 )2R + CH3OH

In either process, the rate of adsorption is expected to be extremely slow. Hence, the
addition of an alkaline catalyst greatly aids in accelerating these adsorption processes.
Heat and/or time can achieve the same viscosity reduction effect, because both
increased temperature and prolonged time lead to more adsorbed sil-nes. Very few
silane oligomers are produced in this method since the hydrolyzed silane molecules do
not exist in quantity at any time of treatment. Unlike the mixing of a silane pretreated
filler, it is essential to consider the kinetics of silane adsorption in this type of treatment
method.
 195

120

100

80
0 /
60

40
20/

20

0 1 2 3 4 5 6 7
% Silane
Figure 18. Amount of total and chemisorbed silane on a filler as a function of
the initial silane concentration with respect to the filler

7. Hydrothermal Stability of Silane Coupling Agent

Siloxane and silanol groups establish an equilibrium depending upon the
humidity level as shown below.
H2o
"aSiOSim - * SiOH + 6 SiOH
Silanol groups are very unstable and the equilibrium is toward forming the siloxane
group. Acidic and alkaline pH's catalyze the hydrolysis and condensation reactions.
The bond energy of the siloxane is very high and thus the thermal stability of siloxane
bonds is excellent, as exemplified by the excellent properties of polydimethylsiloxane at
elevated temperatures. However, in the presence of benzoic acid catalyst, the activation
energy of the hydrolysis becomes as low as 5 kcai/mol as compared to the bond energy
of the siloxane, 89.3 kcal/mol. In spite of this low activation energy of hydrolysis in
the presence of a catalyst, a silane coupling agent is effective even in a humid
environment. This is probably because of the highly skewed equilibrium toward
forming siloxane bonds.
The hydrothermal stability of the siloxane networks is not strongly dependent
upon the structure of the organofunctional group. The hydrothermal stability of a
silane is studied by desorption of the silane from the substrate surface. In spite of the
 196

similar time required to cleave siloxane bonds from different silane coupling agent, the
desorption kinetics differs markedly due to the difference in solubility of the newly
formed oligomers. If an additional mechanism exists, such as surface induced
polymerization of the organofunctional groups, further difficulty for desorption is
encountered.
The hydrothermal stability of the SiOR bonds where R represents C, Si, Al and
other elements depends strongly on the element in the R group. It is believed that
electronegativity difference between the silicon atom and the element in the R group is
one of the important factors determining the hydrothermal stability of the bond. Some
evaluations of the various oxane bonds have been reported (41). Pure silica, alumina,
and titania were treated bv a silane coupling agent and hydrothermal stability was
studied by monitoring the rate of silane desorption as shown in Figure 19. It is clear
from the figure that the stability of the bonds is in the following order: SiOSi > SiOTi >
SiOAL. It is intriguing to speculate that a similar stability difference also exists on the
surface of the glass fiber, as it consists of various surface hydroxyl groups as well as
alkaline and alkaline earth elements.

X1017

, 1.6 ' SiO0

41 Ti O2
.2 1.2 o At 2 0 3
0
E
". 0.8

_ 0.4 - _A A A
0

0 20 40 60 80 t00 120 140

Immersion time(hrs)

Figure 19. The desorption kinetics of y-aminopropyldimethylethoxysilane

from various substrate surfaces: (A)silica, (a) titania, (o) alumina.
When water reaches the glass/silane interface, water molecules might exist in a
separate gas-like state or condensed liquid-like state. It is believed that this liquid-like
water damages the interface more than the gas-like molecules. A recent dielectric study
by Steeman and Maurer shows a rare insight on this subject. They measured the
dielectric constant of glass bead filled polyethylene as a function of the silane coverage
on the glass beads surface and the relative humidity. When the water molecules are
less than a monomolecular equivalent, the effect of the water on the dielectric constant
was small. As the coverage reaches several layers equivalent, the hydrolytic damage
dramatically increased.
 1147

A simple concentration consideration alone is difficult to explain the above

phenomenon. It is possible that the associated, liquid-like water can build a strong
osmotic pressure at the interface that will stress the oxane bonds. The oxane bonds at
the interface will be subjected to stress. When a tensile stress is applied to the bond,
the mechanochemical effect lowers the apparent activation energy of the hydrolysis of
the oxane bonds, making it easier to be hydrolytically cleaved. The aforementioned
differences in the hydrothermal stability of various oxane bonds may become especially
important under such circumstances. Easily cleaved, non-siloxane bonds may become
the source of water collection and exert pressure on the surrounding siloxane bonds,
which will be cleaved easier than the siloxane bonds with pure silica surface.
It is also necessary to consider the dissolution of the alkaline and alkaline earth
elements into the water. If the water is acidic, the dissolution of the surface elements
readily takes place. When large atoms near the glass fiber surface are leached out, the
siloxane and other networks will be placed under tension, making the glass/silane
bonds more vulnerable to hydrolytic attack. All of these phenomena described above
do not exist in pure silica's such as quartz powder or fumed silica. It is known that
pure silica filled polymers exhibit good hydrothermal stability. It is possible that the
lack of these mechanisms to weaken the interfacial bonds may significantly contribute
to the stability of the composite. While there are no rigorous studies to prove the
aforementioned phenomena nor are effort correlating these subjects with the
hydrothermal stability of a composite, these phenomena are well-accepted in the
literature for other materials.

8. Summary
The molecular structure of silanes in solution and on substrates has been
reviewed. Major advancements over the past decade have helped to elucidate the
structure of silane coupling agents. With an improved understanding of the silane
structurs. it is now possible to control some structures of silane on a substrate. Once
structure/property correlation is understood, a specific property of a composite material
can be tailored by controlling the interfacial structure. Our current knowledge of silane
coupling agents allows us to attempt this systematic approach that has been defined in
this paper as interphase engineering.

9. Acknowledgment
This work was in part supported by the grant from The Office of Naval
Research.
 198

10. References
1. E.P. Plueddemann, Ed., "Interfaces in Polymer Matrix Composites," Academic
Press, New York (1974).
2. D.E. Leyden, and W. Collins, Eds., "Silylated Surfaces," Gordon and Breach
Science, New York (1980).
3. E.P. Plueddemann,"Silane Coupling Agents," Plenum Press, New York (1982).

4. H. Ishida and G. Kumar, Eds., "Molecular Characterization of Composite

Interfaces," Plenum Press, New York (198).

5. H. Ishida, and J.L. Koenig, Eds.,"Composite Interfaces," Elsevier Science

Publishing, New York (1986).

6. D.E. Leyden, Ed.,"Silanes, Surfacec, and Interfaces," Gordon and Breach

Science, New York (1986).

7. H. Ishida, Ed., "Interfaces in Polymer, Ceramic, and Metal Matrix Composites."

Elsevier Science, New York (1988).
8. F.R. Jones, Ed., "Interfacial Phenomena in Composite Materials," Butterworths,
London (1989).
9. H. Ishida, Ed.," Controlled Interphaqes in Composite Materials," Elsevier
Science, New York (1990).
10. H. Ishida, S. Naviroj, S.K. Tripathy, J.J. Fitzgerald, and J.L. Koenig, J. Polym.
Sci. -Phys., 20, 701 (1982).

11. J.F. Brown, Jr. and L.H. Vogt, J. Am. Chem. Soc., 84, 4313 (1965).

12. J.F. Brown, Jr., J. Am. Chem. Soc., 84, 4317 (1965).
13. J.D. Miller and H. Ishida, Polym. Composites, 5, 18 (1984).
14. H. Ishida, S. Naviroj and J.L. Koenig, in "Physicochemical Aspects of Polymer
Surfaces," K.L. Mittal, Ed., Plenum, New York (1983) p. 9 1 .
15. N. Nishiyama, K. Horie, R. Shick and H. Ishida, Polymer, 31, 380 (1990).

16. N. Nishiyama, J. Dent. Mat., 5, 519 (1986).

17. J. Jang, H. Ishida and E.P. Plueddemann, Proc. 42nd Ann. Tech. Conf.,
Composites Inst., SPI, Sect. 21-F (1987).
18. H. Ishida, J.L. Koenig and K.Gardner, J. Chem. Phys., 77, 5748 (1982).

19. J.D. Miller and H. Ishida, J. Chem. Phys., 86, 1593 (1987).
 SI9N

20. J.D. Miller and H. Ishida, Langmuir, 2, 127 (1986).

21. B.D. Favis, L.P. Blanchard, J. Leonard and R.E. Prud'homme, Polym. Comp.,
5, 11 (1984).

22. J.D. Miller and H. Ishida, J. Polym. Sci. -Phys., 23, 2227 (1985).

23. J.D. Miller and H. 1shida, Aacromu;ecukes, 17, 1659 (1984).

24. H. Ishida, unpublished results.

25. K. Hoh, H. Ishida and J.L. Koenig, Polym. Comp., 9, 15 (1988).

26. R.T. Graf, J.L. Koenig and H. Ishida. J. Adhesion, 16, 97 (1983).

27. M. Kokubpo, H. Inagawa, M. Kawahara, D. Terunuma and H. Nohira,

Kobunshi Ronbunshu, 38, 201 (1981).
28. J. Bjorksten and L.L. Yaeger, Mod. Plast., 29, 124 (1952).

29. C.A. Kumins and J. Roteman, J. Polym. Sci., IA, 527 (1963).

30. R.C. Hooper, Proc. 11th Ann. Tech. Conf./Reinforced Plast. Div., SPI, Sect. 8-
B (1956).

31. E.P. Plueddemann, "Silane Coupling Agents," Plenum Press, New York (1982)
p. 13 4 .

32. H. Ishida and J.L. Koenig, J. Polym. Sci. -Phys., 17, 615 (1979).
33. E.P. Plueddemann, J. Paint. Technol., 10, 1 (1968).

34. H. Ishida and P. Bussi, Macromolecules, 24, 3569 (1991).

35. H. Ishida and P. Bussi, J. Mat. Sci., 26, 6373 (1991).

36. C.D. Han, C. Sanford and H.J. Yoo, Polym. Eng. Sci., 18, 849 (1978).
37. J.D. Miller, H. Ishida and F.H.J. Maurer, Rheol. Acta, 27, 397 (1988).

38. C. Scott, H. Ishida and F.H.J. Maurer, J. Mat. Sci., 26, 5708 (1991).

39. H. Ishida and T. Gayet, unpublished results.

40. P. Dreyfuss, Macromolecules, 11, 1031 (1978).

41. S. Naviroj, J.L. Koenig and H. Ishida, J. Adhesion, 18, 93 (1985).