'gﬁiaﬁhﬁlt.

«vi?

 

 

 

 

 

 

.- ...

“—4... ‘
“an...

33.4..

”'59-?-
u .m-

..

a
L hug}? A. ”i

v5m:m\ ..,...

 

- ’
' I
:2-
.1
1"

v.3”. A. i"

‘ . q . .\

‘ 331$ rif‘égﬁg '~ 9’ 3 _
’ Ki 3%!“ , x. ., _

5‘ 4-, . 7' 41'? g: a. “Li:

. ‘; -;| ;-‘; 1'. 32““..‘7

37:1;ri!’ "5“," {#ﬁffl‘f b‘hmi

n.

 

THESR‘.‘

2:00?

This is to certify that the
dissertation entitled

NANOPARTICLE INDUCED WETTING OF POLYMER FILMS
AND SELF-ASSEMBLED MULTILAYERS OF
NANOCOMPONENTS

presented by

 

R.S.KRISHNAN

has been accepted towards fulﬁllment
of the requirements for the

.UBRARY
Michigan State
University

 

 

PhD. degree in Chemical Engineering and
Materials Science

77/7 7%

Major Professor’s Signature

(None 9023 6

Date

 

 

MSU is an Afﬁnnative Action/Equal Opportunity Institution

PLACE IN RETURN Box to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 p:/CIRCIDateDue.indd-p.1

 

NANOPARTICLE INDUCED WETTING OF POLYMER FILMS
AND SELF-ASSEMBLED MULTILAYERS OF
NANOCOMPONENTS
By

R.S.Krishnan

A Dissertation
Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY

Department of Chemical Engineering and Materials Science

2006

ABSTRACT

NANOPARTICLE INDUCED WETTING OF POLYMER FILMS
AND SELF-ASSEMBLED MULTILAYERS OF
NANOCOMPONENTS

By

R.S.Krishnan

The control of dewetting for thin polymer ﬁlms is a technical challenge and of
signiﬁcant academic interest. Although studies have been published on the wetting of
polymer ﬁlms in the presence of nanoparticles, the underlying physics is still a matter
of debate. In this work, we report a systematic study of improved wetting behavior of
thin polymer ﬁlms containing nanoparticles, as a function of nanoparticle size and
concentration. An enthalpy matched system consisting of polystyrene nanoparticles in
linear polystyrene is used to show that nanoparticles are uniformly distributed in the
ﬁlm after spin coating and drying, however on annealing the ﬁlm above its bulk glass
transition temperature the nanoparticles segregate strongly to the solid substrate. We
ﬁnd that for a wide range of ﬁlm thicknesses and nanoparticle sizes, approximately
monolayer substrate coverage of nanoparticles is required for strong dewetting
inhibition. We also show that cadmium selenide quantum dots inhibit dewetting of
both polystyrene and PMMA thin ﬁlms. Moreover, TEM microscopy images indicate
that CdSe quantum dots segregate primarily to the air surface. Gain of conﬁguration

entropy of the melt linear chains promotes segregation of nanoparticles to the

substrate, as occurs for polystyrene nanoparticles. However, for CdSe nanoparticles
this is offset by surface energy terms which promote segregation of the nanoparticles
to the air surface. We argue that this is due to the inert low-energy Oleic acid brush
introduced to promote organic compatibility of the quantum dot surfaces.

Finally, we use the nanoparticle induced wetting of a polymer ﬁlm due to the
self-assembly of nanoparticles at the interface to construct the layered assembly of
polymer-nanoparticle sandwich ﬁlms. We report an alternative route to multilayer
nanostructures where the layered self-assembly of the constituents is driven by the
interplay between entropy, due to architectural differences, and surface energy. The
approach is simple, thin ﬁlm deposition is followed by an aging step to allow
segregation of the components, and after stabilization of this layer, the process is
repeated. We show that this enables fabrication of multilayers using a wide range of
nanoparticles and non-polar polymers. We call this the Self Assembled Multilayers of

Nanocomponents or SAMON process.

Copyright By
lléLPﬂRISIUNAUN

2006

Dedication

This dissertation is dedicated to the Lord Almighty Krsna who has given me

enough strength and courage to face all the good and the bad times.

Acknowledgements

The period of my life here at Michigan State University has evolved me as a
mature human being. It’s hard for me to fathom that the journey that began in 2001
has come to an end. Throughout the duration of this degree, I’ve learnt some of the
very important lessons of life. As a graduate student one is self reliant, yet research is
a collaborative effort and there have been numerous people who have helped me
directly or indirectly with my work during the past few years. Firstly and foremostly I
would like to thank my advisor Professor Michael Mackay for his guidance and
support during the period of this degree.

My research would not have been possible without some of the materials that
I worked with. In this regard, I want to acknowledge Dr. Craig Hawker at UCSB for
supplying us with polystyrene nanoparticles which laid the foundation for my work.
Thanks are also due to Professor Michael Wong and his student Suba Asokan at Rice
University for providing us with the invaluable cadmium selenide quantum dots.

I want to express my sincere thanks to Professor Phil Duxbury for his help
with the theoretical insight on some of the forces involved in stratiﬁed systems. I’m
indebted to Dr. Alicia Pastor for her incredible help with ultramicrotoming and TEM
imaging.

I also want to thank Dr. P.Thiyagarajan and Rick Goyette at Argonne National
Laboratory (ANL) for providing partial ﬁnding for my work and also ample time to

do neutron reﬂectivity measurements. I’m also grateful to Nicholas de Souza at ANL

vi

and Jinbo at University of Massachusetts for suggesting some really useful and timely
tips which reﬁned my sample preparation techniques.

Finally, I want to thank my friends, my parents and my sister for their support
and encouragement throughout the course of my Ph.D. and also for being there

during some of the tough times in my life.

vii

Table of Contents

List of Tables ............................................................................................................... x
List of Figures ............................................................................................................ xii
Chapter 1 : Introduction ............................................................................................ l
1.1 Motivation .................................................................................................... l

1.2 Background .................................................................................................. 4
Chapter 2 : Inﬂuence of Molecular Architecture on the Dewetting of Thin
Polystyrene Films ........................................................................................................ 9
2.1 Introduction ................................................................................................. 9

2.2 Materials and Methods .............................................................................. 12

2.3 Results and Discussion .............................................................................. 15

2.4 Conclusion ................................................................................................. 30
Chapter 3 : Nanoparticle Induced Wetting of Polymer Thin Films .................... 31
3.1 Introduction ............................................................................................... 31

3.2 Experimental .............................................................................................. 35

3.3 Results and Discussion .............................................................................. 38

3.4 Conclusions ................................................................................................ 60
Chapter 4 : Inﬂuence of Polymer Molecular Weight on the Dewetting of Thin
Polymer-Nanoparticle Blend Films ......................................................................... 62
4.1 Introduction ............................................................................................... 62

4.2 Results and Discussion .............................................................................. 64

4.2.1 Entropic Push and Dewetting ............................................................. 64

4.2.2 Diﬂusion Time and Dewetting Time ................................................... 69
Chapter 5 : Self-Assembled Multilayers of Nanocomponents .............................. 76
5.1 Supporting Online Material ....................................................................... 89

5.1 .1 Experimental ....................................................................................... 89

5.1.2 Neutron Reﬂectivity Data Analysis ..................................................... 90

5. I .3 Sample preparation for TEM .............................................................. 93
Chapter 6 : Conclusions and Summary .................................................................. 94
Appendix A: Dispersion Forces ............................................................................... 96

viii

Appendix B: Neutron Reﬂectivity — Data Sensitivity Analysis ............................. 99
References and Notes .............................................................................................. 106

List of Publications ................................................................................................. 1 l4

List of Tables

Table 1.1. The molecular weight (MW), Polydispersity Index (PDI) and Radius of
gyration (Rg) or radius (R) of the polymer standards and nanoparticles respectively .. 8

Table 2.1. Values of the prefactor, A, as determined for the linear-linear blends of
polystyrene shown in Figure 2.3. Note that for pure PS 19 kDa the value of A is 5
times greater than that for pure PS 75 kDa. However, addition of the low molecular
weight polymer actually reduces the value of A with respect to the higher molecular
weight .......................................................................................................................... 21

Table 2.2. Parameters used in Parratt32 software (from HMI Berlin) to ﬁt the data in
Figure 2.5. An air interface is assumed at the top of the ﬁrst layer. The roughness of
layer 2 used in the simulation of the after annealing ﬁlm is extreme and represents an
empiricism to generate a scattering length density proﬁle. Other modeling with many
layers near the wafer surface generates a similar scattering length density (SLD)
proﬁle. The system consisted of 10 wt% 25.3 kDa protonated polystyrene
nanoparticles blended with 63 kDa deuterated linear polystyrene and was annealed at
140 °C for 2 hrs ........................................................................................................... 24

Table 3.1 The molecular weight (MW), Polydispersity Index (PDI) and Radius of
gyration (R8) of the polymer standards. ...................................................................... 34

Table 3.2 The molecular weight (MW), Polydispersity Index (PDI) and Radius of the
nanoparticles. .............................................................................................................. 34

Table 3.3 Values of the fractional coverage (0) based on the nanoparticle
concentration,(|>, by assuming perfect segregation (and using Eq. 3.1) and those
determined from neutron reﬂectivity data (93w) using the scattering length density
(SLD) of the layer next to the substrate (see Eq. 5.2), for blends of nanoparticles 78
kDa with dPS 63 kDa . There is good agreement between the values obtained from
neutron reﬂectivity data and those expected for ideal segregation ............................. 44

Table 3.4. The surface energy, dielectric constant and reﬂective index of the
materials. ..................................................................................................................... 51

Table 4.1. Parameters used in Parratt32 software (from HMI Berlin) to ﬁt the data in
Figure 4.5. An air interface is assumed at the top of the ﬁrst layer. ........................... 74

Table 5.1. The molecular weight (MW), Polydispersity Index (PDI) and Radius of
gyration (Rg) or radius (R) of the polymer standards and nanoparticles, respectively 90

Table 5.2. A) Parameters used in Parratt32 software (from HMI Berlin) to ﬁt the data
in Figure 5.1A. An air interface is assumed at the top of the ﬁrst layer. The before-

annealing ﬁlm is modeled as a single layer with a homogenous distribution of
nanoparticles corresponding to an average scattering length density (SLD) of
5.92><104’A'2. The after annealed ﬁlm was modeled as two layers with the layer next
to the substrate representing the nanoparticle rich phase. B) Parameters used to ﬁt the
data in Figure 5.1D (solid line). An air interface is assumed at the top of the ﬁrst
layer. The roughnesses used in the simulation of the after annealing ﬁlm has no
physical meaning and represents an empiricism to generate a scattering length density
proﬁle. Modeling a homogenous distribution of 3 layers generates a reﬂectivity
proﬁle represented by the dotted line in Fig. 5.1D. .................................................... 92

Table B.1. Parameters used in Parratt32 software (from HMI Berlin) to ﬁt the data in
Figure 3.1 for three different models. An air interface is assumed at the top of the
ﬁrst layer. The system consisted of 20 wt% 78 kDa protonated polystyrene
nanoparticles blended with 63 kDa deuterated linear polystyrene and was annealed at
160 °C for 24 hrs under vacuum. .............................................................................. 10]

xi

List of Figures

Figure 1.1: Schematic of the intramolecularly crosslinked polystyrene nanoparticles
produced from linear chain precursors with ~20 mol% pendent crosslinking groups. 7

Figure 2.1. Pure 25.3 kDa nanoparticle ﬁlm before (a) and after (b) annealing at
160°C for 24 hrs. There are no signs of ﬁlm rupture even after such extended
annealing times. However, the linear polystyrene of comparable molecular weight
(PS 19 kDa) dewets in less than 10mins (c & d) on the silanized surface (of. Figure
3). Film thickness is ca. 40 nm and the length of the scale bar is 100 um ................. 15

Figure 2.2. Optical micrographs of blends of linear polystyrene 75 kDa with 25.3
kDa nanoparticle. All the ﬁlms were annealed for time >24 hrs in vacuum: (a) Pure
PS, (b) 1%, (c) 5%, (d) 10%, (e) 15% and (f) 20% nanoparticle concentration. Note at
0.15-0.20 nanoparticle weight fraction (w), the dewetting is completely eliminated
(aerial fraction, 0 ~ 1.5-2.0). Film thicknesses are ca. 40nm and the length of the scale
bar is 100 um. ............................................................................................................. 17

Figure 2.3 (a) Hole radius versus time during aging at 160°C for pure linear PS 75
kDa, PSl9 kDa and their blends along with power law ﬁts. All the systems have been
modeled with a power law model and the growth exponent was ﬁxed at 2/3
(represented by the solid lines). Each data point is an average of ca. 30 holes over 3
samples. The power law prefactor, A, is given in Table 2.1. (b) A 20% PS 19 kDa
blend behaves fundamentally different and an initial power law of ~ 0.5 is seen,
however, this value is not constant. All ﬁlms are ca. 40 nm thick. ............................ 20

Figure 2.4. Hole growth as a function of time for pure PS 75 kDa linear polymer and
its 1% blend with 19 kDa linear polystyrene and 25.3 kDa tightly crosslinked
nanoparticle at 160°C. Clearly with just 1 wt% (0 = 0.13) addition of the
nanoparticle, the rate of dewetting has drastically slowed, emphasizing the effect of
molecular architecture on dewetting. Also note that the growth exponent for the 1%
blend is initially 0.32. All ﬁlms are ca. 40nm thick .................................................... 23

Figure 2.5. Reﬂectivity multiplied by reﬂectance wave vector to the fourth power
(Rq‘) vs. q for a silanized silicon wafer spin coated with 63 kDa deutrated
polystyrene blended with 10wt% 25.3 kDa hydrogenated, tightly crosslinked,
polystyrene nanoparticles before and after annealing at 140°C for 2hrs under vacuum.
The dotted line represents the reﬂectivity proﬁle ﬁt if the nanoparticles phase
separated to the air interface. The solid lines represent the ﬁts for the before and after
annealed ﬁlms as described in the text. ...................................................................... 24

Figure 2.6. Nanoparticle concentration proﬁle determined from the parameters listed

in Table 2.2 for the ‘aﬁer annealed’ ﬁlm. A scaled representation of the nanoparticle
is placed in the lower right-hand comer. .................................................................... 26

xii

Figure 2.7. Optical micrographs of 10 wt% blends of 75 kDa linear polystyrene with
25.3 kDa polystyrene nanoparticles (tightly crosslinked) at various surface coverage.
The surface coverage (0) was varied by changing the ﬁlm thickness from 16, 33, 50,
65 and 78 nm as indicated in the micrographs. All the ﬁlms were annealed for > 24hrs
in air at 170°C and when 0 > 1.5 the dewetting is almost eliminated. The insets show
the ﬁlm before annealing. The length of the scale bar is 100 um. Note that for the 16
nm ﬁlm, an atomic force micrograph has been shown for the after annealed ﬁlm due
to difﬁculties utilizing optical microscopy with this sample. In this case the image
size is approximately 50 um X 50 um. Also note that for ﬁlms thicker than 50 nm
some complicated surface features (buckling instabilities) are present in both before
and after annealed ﬁlms, due to the spincoating process. ........................................... 28

Figure 3.1: A cartoon depicting nanoparticle migration to the substrate after high
temperature annealing. Neutron reﬂectivity measurements have shown that the
nanoparticles are uniformly distributed in the thin (ca. 50 nm) ﬁlm prior to high
temperature annealing (a), yet after annealing they are found to localize at the solid
substrate (b). The aerial coverage of the nanoparticles could be determined by
assuming a hexagonal close packing of the nanoparticles forming a layer 2a thick
with A being the average half gap between the nanoparticles .................................... 38

Figure 3.2 Reﬂectivity data for nanoparticle blends of deuterated linear polystyrene,
dPS 63 kDa, and polystyrene nanoparticles (Mw78 kDa). The ﬁlms were annealed in
air at 160°C. The data in (a) and (b) is offset by a factor of 10 for clarity. (a) Data for
nanoparticle concentrations 5, 10 and 20 wt% at ﬁxed ﬁlm thickness, approximately
45nm. The 10 and 20% ﬁlms were annealed for 24hrs, while the 5% ﬁlm was
annealed for 2hrs. b) Data for ﬁxed nanoparticle concentration, 10 wt%, for ﬁlm
thicknesses 27, 39 and 77 nm, after annealing for 6 hrs. c) Reﬂectivity multiplied by
reﬂectance wave vector to the fourth power (RQ‘) vs. Q using the data of Fig. 3.2b
(39 nm ﬁlm), and a comparison of ﬁts to three different models (see text for details).
..................................................................................................................................... 41

Figure 3.3: Optical micrographs of blends of linear polymer (PS 75 kDa) with PS
nanoparticles (Mw78 kDa) after annealing the ﬁlms for 24 hrs in vacuum at 160°C
on a Sigmacote substrate. In Fig. 3.3a the nanoparticle fractional coverage (9) is
varied by changing the bulk nanoparticle concentration (4)) at a constant thickness of
56 run, while in Fig. 3.3b the fractional coverage (0) is varied by changing the overall
ﬁlm thickness at ﬁxed nanoparticle concentration of 10 wt %. In both cases a
fractional coverage of a monolayer is needed to severely retard the dewetting of the
linear polymer. The length of the scale bar is 200 um. .............................................. 45

Figure 3.4: Hole radius versus time during annealing in air at 160°C for pure linear

PS 75 kDa, PS 19 kDa and PS 5 kDa along with power law ﬁts on two different
silanized substrates. Use of OTS, as compared to Sigmacote, as the silanizing agent

xiii

increases the rate of dewetting for the pure PS 75 kDa. The ﬁlm thickness in all cases
is approximately 45 nm ............................................................................................... 46

Figure 3.5: Optical micrographs of blends of linear polymer (PS 75 kDa) with PS
nanoparticles of 4 different molecular weights: 41 kDa, 78 kDa, 211 kDa and 1.5
MDa. All the ﬁlms have a constant thickness of 50i7.2 nm. The bulk concentration
((1)) of the nanoparticle in the ﬁlms correspond to a nanoparticle fractional coverage
(9) of about a monolayer. All the ﬁlms have been at annealed at 170°C under vacuum.
The top row represents the ﬁlms after annealing (AA) on a Sigmacote substrate and
the middle and bottom rows represent the ﬁlms after annealing on OTS functionalized
substrates. The annealing time for the top two rows is one day. The ﬁlms in the
bottom row are the same as in the middle row but have been annealed for 5 days.
Scaled representations of the nanoparticles are also shown in the micrographs, and
their sizes are also displayed. Increasing the size of the nanoparticle increases the
stability of the polymer ﬁlm against dewetting. The length of the scale bar is 200 um.
..................................................................................................................................... 48

Figure 3.6: Reﬁ'active index as a ﬁtnction of ﬁlm thickness for linear polystyrene, PS
75 kDa (2Rg~ 15 nm) and PS nanoparticles: NP 41 kDa (2a~ 5 nm), NP 78 kDa (2a~
6.2 nm) and NP 211 kDa (2a~ 8.6 nm) at 583 nm. The dotted line is a guide to the
eye. .............................................................................................................................. 55

Figure 3.7: TEM micrograph of the cross-section of PS 211 kDa ﬁlm (~75 nm)
containing 22 wt% quantum dots after annealing at 230°C for 30 mins under vacuum
(layer 1) on an SiOz substrate. Unlike the PS-PS system, the quantum dots migrate to
both the air interface and the solid substrate. Note that the segregation of the quantum
dots at the air interface is much stronger than that at the substrate. Air interface in the
TEM image is tagged with a 50 nm thick gold layer. The dark grey region is the PS
phase. .......................................................................................................................... 56

Figure 3.8: Optical micrographs of blends of linear polymer (PS 75 kDa) with CdSe
quantum dots at a fractional aerial coverage (0) of a) 0.3 after annealing for 1hr and
b) 0.9 after annealing for 24 hrs. The ﬁlm thickness is ca. 45 nm and annealing was
done at 180°C under vacuum on an OTS substrate. It is interesting to note the
irregular shape of the holes for sub-monolayer coverage (left ﬁgure). ...................... 58

Figure 3.9: Optical micrographs of thin ﬁlms after annealing at 180°C for 24 hrs
under vacuum on an OTS substrate. a) Pure PMMA 76 kDa ;b) PMMA 76 kDa
containing 34 wt% quantum dots (QDs); c) PMMA containing 14 wt% NP 78 kDa; (1)
PS 75 kDa containing 16 wt% NP 78 kDa. The length of the scale bar is 200 pm.

Film thickness and the fractional coverage (0) is given in each ﬁgure. .................... 59
Figure 4.1: Optical micrographs of blends of linear polymers (PS 5 kDa, PS 19 kDa

and PS 75 kDa) with PS nanoparticles of 3 different molecular weights: NP 41 kDa (a
~ 2.5 nm), NP 78 kDa (a ~ 3.1 nm) and NP 211 kDa (a ~ 4.3 nm) .All the ﬁlms have a

xiv

constant thickness of 45 nm. The bulk concentration (4)) of the nanoparticle in the
ﬁlm corresponds to a nanoparticle fractional aerial coverage (0) of about a monolayer
(cf. eq. 4.1). All the ﬁlms have been annealed at 160°C under vacuum on a Sigmacote
substrate for 24 hrs. Clearly increasing the size of the nanoparticle and/or the polymer
Rg increases the stability of the polymer ﬁlm against dewetting. The length of the
scale bar is 200 um. .................................................................................................... 65

Figure 4.2: Fitted reﬂectivity data for thin ﬁlms of (a) dPS blends with nanoparticle
NP 78 kDa as a function of dPS molecular weight: dPS 21 kDa and dPS 63 kDa. The
bulk nanoparticle concentration in both cases is 40 wt% and the ﬁlm thickness is
approximately 45 nm. The ﬁlms were annealed for 6 hrs under vacuum at 160°C.
Reﬂectivity data has been offset by a factor of 10 for clarity and the ﬁts in both cases
correspond to a three layer model (see text). b)Nanopartic1e concentration proﬁle for
the reﬂectivity ﬁts shown in Fig. 2a along with sigmoid ﬁts as a function of
normalized ﬁlm thickness. It is clear that in both cases, the nanoparticle concentration
increases substantially at the substrate. However, under identical annealing
conditions, nanoparticle molecular weight and ﬁlm thickness, ﬁlm containing dPS 21
kDa has 24 % nanoparticles in the bulk from an initial value of 40 %. 1n the case of
the ﬁlm containing dPS 63 kDa, the bulk nanoparticle concentration is only 12 %
suggesting an increased entropic push for the nanoparticle migration to the substrate
due to the higher molecular weight of the linear polymer. ......................................... 67

Figure 4.3: Hole radius versus time during annealing in air at 160°C for pure linear
PS 75 kDa, PS 19 kDa and PS 5 kDa along with power law ﬁts on two different
silanized substrates. Clearly changing the silanizing agent to OTS increases the rate
of dewetting for the pure PS 75 kDa. The ﬁlm thickness in all cases is approximately
45 nm. ......................................................................................................................... 70

Figure 4.4: Optical micrographs of blends of linear polymers (PS 5 kDa, PS 19 kDa
and PS 75 kDa) with PS nanoparticles of 3 different molecular weights: NP 41 kDa (a
~ 2.5 nm), NP 78 kDa (a ~ 2.5 nm) and NP 211 kDa (a ~ 2.5 nm). All the ﬁlms have a
constant thickness of 45 nm. The bulk concentration (4)) of the nanoparticle in the
ﬁlm corresponds to a nanoparticle fractional aerial coverage (0) of about a monolayer
(of. eq. 4.1). All the ﬁlms have been annealed at 160°C under vacuum on an OTS
substrate for 24 hrs. Changing the substrate from Sigmacote to OTS decreases the
dewetting time of the linear polymer. A polymer ﬁlm would remain stable, provided
there is sufﬁcient entropic push for the nanoparticles to migrate to the substrate (see
text). The length of the scale bar is 200 um ................................................................ 71

Figure 4.5: a) Fitted reﬂectivity data for 30 nm thick ﬁlms of dPS 63 kDa containing
20 wt% nanoparticle NP 78 kDa as a function annealing time: 0 sec (before
annealing), 10 sec, 120 sec and 600 sec. Thermal annealing of the ﬁlms was done in
air 160°C. Reﬂectivity data has been offset by a factor of 10 for clarity and the ﬁts in
all cases correspond to a two layer model (see text) except for the case of the “before
annealing” ﬁlm which was modeled as a single layer with homogenously distributed

XV

nanoparticles. b) Nanoparticle concentration in the top and the bottom layer as
determined from the scattering length density of the two layers. Just aﬂer 120 sec of
annealing time, almost all the nanoparticles have migrated to the substrate with very
few remaining in the bulk. The error in estimating the scattering length density is
about 20-30%. ............................................................................................................. 73

Fig. 5.1 (A) Reﬂectivity (R) multiplied by reﬂectance wave vector (Q) to the fourth
power (RQ‘) versus Q for a silicon wafer spincoated with 63 kD deuterated
polystyrene blended with 10 wt% 211 kD protonated polystyrene nanoparticles before
and after annealing at 160°C for 2 hrs under vacuum. The dotted line represents the
reﬂectivity proﬁle ﬁt if the nanoparticles migrated to the air interface. The solid lines
represent the ﬁts for the before and after annealed ﬁlms as described in the text. Film
thickness is approximately 40 nm. (B) Nanoparticle concentration proﬁle determined
from the scattering density proﬁle for the ‘after annealing’ ﬁlm shown in A. A scaled
representation of the nanoparticle is placed in the lower right-hand comer. (C)
Spincoating process to make the multilayered ﬁlms. A nanoparticle — polymer
mixture is spincoated onto a substrate and thermally annealed to allow nanoparticle
segregation and polymer crosslinking. This process is repeated to form multiple
layers as demonstrated by the numbers: 1, 2, . (D) RQ4 versus Q for a silicon wafer
spin coated with three layers of 211 kD protonated linear polystyrene containing 20
mol% crosslinker, blended with 15 wt% 78 kD deuterated polystyrene nanoparticle
after annealing each layer at 230°C for 24 hrs under vacuum. The ﬁt (solid line)
corresponds to six alternating layers of protonated polymer and deuterated
nanoparticle while the dotted line is the prediction if the nanoparticles were
homogeneously distributed. The thickness of each polymer -nanoparticle layer is
approximately 44 nm. ................................................................................................. 80

Fig. 5.2 (A) Transmission electron micrograph (TEM) of an assembly of sixteen
layers: eight CdSe quantum dots alternating with eight crosslinked polystyrene layers,
assembled on a silicon wafer. Each bilayer is numbered on the micrograph from 1 to
8. The quantum dot layers are approximately a monolayer thick (~3-4 nm) whereas
the polystyrene layers are about 75 nm in thickness. In all the micrographs a gold
layer was sputtered on the ﬁlm after fabrication to mark the air interface and masks
the uppermost quantum dot layer. The silicon wafer was treated with hydroﬂuoric
acid and removed prior to microtoming as described in the Supporting Online
Material. There are some quantum dots next to the substrate as discussed in the text.
(B) Assembly of six layers formed by aging four layers to form three layers of CdSe
quantum dots and three layers of crosslinked polystyrene. This sample was made by
aging a polystyrene — quantum dot layer then Spincoating a pure polystyrene layer on
top and crosslinking it then repeating the process. The quantum dot layers are about a
monolayer thick. The inset shows a TEM micrograph of the ﬁrst layer normal to the
substrate surface demonstrating a reasonably uniform ﬁlm. (C) Assembly of eight
layers: four CdSe quantum dots and four crosslinked polystyrene layers, assembled
on a silicon wafer. In this case, the quantum dot layers are approximately 13 nm thick
while the polystyrene layers are 15 nm thick to demonstrate control over the thickness

xvi

of each layer is dictated by the initial conditions developed through the Spincoating
operation. .................................................................................................................... 84

Fig. 5.3 (A) Optical micrographs of a 75 kD polystyrene ﬁlm (~58 nm thick) ﬂoated
onto a 76 kD ﬁlm PMMA ﬁlm (~56 nm thick) and thermally aged at 180°C for 24
hrs. under vacuum. Isolated polystyrene drops can be seen on the surface of PMMA
after annealing the bilayer assembly on a silicon (Si) wafer with the native oxide
(SiOz) layer. (B) PMMA ﬁlm ﬂoated on polystyrene subject to the same annealing
procedure given in A to show a similarly unstable ﬁlm. The instabilities shown in A
and B disappear in C and D, respectively, when the top layer is replaced by a
composite ﬁlm composed of both the quantum dots (ODS) and the polymer. The ﬁlm
ordering is given in the ﬁgure with the abbreviations listed above, the length of the
scale bar is 200 um. .................................................................................................... 87

Figure B]: RQ4 vs. Q for the data and ﬁts as explained in the text. ....................... 100
Figure B.2: x2 as a function of interfacial roughness for three two layer models at
constant layer thicknesses and SLDs. The solid lines in the ﬁgure correspond to

polynomial ﬁts (see Table 3.1) ................................................................................. 102

Figure B.3: Scattering length density proﬁle as a ﬁtnction of normalized ﬁlm
thickness .................................................................................................................... 103

Figure B.4: Concentration proﬁle of the nanoparticles in the ﬁlm after annealing. 105

xvii

Chapter 1

Introduction

1.1 Motivation

The self assembly of layer-by-layer nanostructured materials has gained
attention due to their relative ease of preparation and robust nature which renders
them useful for many applications. For example, self assembled ultrathin ﬁlms can
function as membranes for gases and may also serve as building blocks for the
construction of nano-devices used in molecular recognition and sensors. However,
polymer thin ﬁlms made by Spincoating or dipcoating are frequently unstable and
may dewet upon thermal annealing or solvent exposure. Thus, when nanodevices
such as functional nanoparticles are incorporated within the ﬁlm they will not be in

equilibrium or desired assembly.

The thickness of the polymer ﬁlm plays a very important role in governing its
stability. For example, a ﬁlm whose thickness is on the order of a few micrometers
could be stable by gravity whereas a thin ﬁlm (~100 nm) is either metastable or
unstable on nonwettable substrates. As the ﬁlm thickness approaches molecular
dimensions, intermolecular forces, of course become very important. With the size
and dimension of various devices employing thin ﬁlms becoming increasingly
smaller, their stability is of primary importance requiring the interfacial forces acting

across a thin ﬁlm to be understood.

We ﬁnd that blends of polystyrene nanoparticles with linear polystyrene chain

either do not dewet or dewet aﬁer considerable annealing times. These observations

are very similar to ﬁillerenes or dendrimer ﬁlled polystyrene thin ﬁlms. Neutron
reﬂectivity on annealed ﬁlms clearly shows enrichment of the nanoparticles at the
substrate even though the nanoparticles are homogeneously distributed before
annealing. The nanoparticle stabilized thin ﬁlm may allow self or guided assembly of
nanodevices since the now stable ﬁlm can be heated and the assembled structure

manufactured then upon cooling trapped in the desired state.

The migration of the nanoparticles to the substrate results from an overall gain
in entropy of the system that also changes the overall system energy. At the substrate
these nanoparticles form a robust layer that can’t move due to the inherent yield stress
in these particles. This effect becomes more pronounced at a monolayer concentration
due to the jamming between the particles. The jammed state results in an effective
solid coating where the particles are held to the substrate by entropy and not allowed
to move past one another along the substrate. This yields a more diffuse interface
generating a stable system by the nanoparticles shielding the adverse van der Waals
forces due to the relative dielectric differences between the nanoparticle and the

polymer.

Since entropy plays a very important role in pushing the nanoparticles down
to the substrate, it is obvious that the three most important length scales viz. size of
the particle, linear polymer radius of gyration and thickness of the ﬁlm would also be
crucial in determining the ultimate conﬁguration of the system which we have

explored to some extent in the present work.

Finally, the nanoparticle induced wetting of a polymer ﬁlm due to the self-

assembly of nanoparticles at the interface can be used to construct the layered

assembly of polymer-nanoparticle sandwich ﬁlms and we call this the Self Assembled

Multilayers of Nanocomponents or SAMON process.

1.2 Background

Thin polymeric ﬁlms have numerous applications; from microelectronics and
sensors to adhesives and bio-medical devices to name a few. Yet, polymer ﬁlms made
by Spincoating or dipcoating are frequently unstable and may dewet upon thermal
annealing or solvent exposure. The thickness of the polymer ﬁlm plays a very
important role in governing its stability (1-4). For example, a ﬁlm whose thickness is
on the order of a few micrometers could be stable by gravity whereas a thin ﬁlm
(~100 nm) is either metastable or unstable on nonwettable substrates. As the ﬁlm
thickness approaches molecular dimensions, intermolecular forces, of course become
very important (3). With the size and dimension of various devices employing thin
ﬁlms becoming increasingly smaller, their stability is of primary importance requiring
the interfacial forces acting across a thin ﬁlm to be understood.

The short range interactions, corresponding to forces operating between
individual molecules, determine to a large extent the macroscopic equilibrium contact
angle and hence the wettability between a liquid drop and substrate. The long range
dispersion forces acting across a thin ﬁlm are due to induced dipole-induced dipole
interactions and depend on the dielectric constant of the intervening medium for the
systems studied here. The strength and magnitude of these interactions depend on the
ﬁlm thickness and so the effective interface potential, which is the sum of the long
range and short range interactions acting across a ﬁlm, ultimately govern the ﬁlm

stability on any given substrate (5-11).

In the past decade, much work has been done in studying the dynamics of thin

ﬁlm dewetting. It has been reported in the literature that thick ﬁlms (>100 nm) may

dewet inorganic substrates by nucleation and hole growth (2, 4) whereas thin ﬁlms
(<100 nm) dewet by spinodal decomposition (12, 13) and/or by ﬁlm imperfection.
(14, 15) Recently, Seemann et al. (8) reported the dewetting of thin polystyrene ﬁlms
on silicon oxide substrate and classiﬁed them as being unstable due to heterogeneous
nucleation or a spinodal process or both. Although these studies have helped to better
understand the process of dewetting, a simple method is still desired to inhibit this

undesirable phenomenon.

Numerous studies have been reported in the literature to stabilize thin ﬁlms
(16-23). Most of these are based on two approaches: a) Surface treatment of the
substrate thereby changing the polymer-surface thermodynamic interaction and/or b)
Modifying the polymer. For example, Henn et al. (21) reported the use of end
functional groups with more afﬁnity for the polar substrate, as a means for inhibiting
the dewetting of thin polystyrene ﬁlms. A similar study involving the modiﬁcation of
the homopolymer by the introduction of ionic functionalities was reported by Karim
et al. (19) Yerushalmi and coworkers (22) discussed stabilization of low molecular
weight polystyrene by the addition of long polystyrene chains. In their work the
improved wetting of the styrene oligomers was attributed to the formation of an
entangled network between the free chains and surface tethered molecules. In another
study by the same authors, ﬁlm stability of a polymer melt on a cross-linked network
of itself was investigated. It was observed that the melt wets at both low and high

cross-link densities but dewets at intermediate densities (autophobicity) (20).

All these approaches have their own advantages and disadvantages depending

on the desired application. One disadvantage is that, polymer modiﬁcation to control

the wetting behavior of a polymer ﬁlm leads to a change in the ﬁlm properties which

might be undesirable, for example, in sensor applications (24).

Recently, Barnes et al. (25) discovered that fullerene nanoparticles inhibit the
dewetting of thin polymer ﬁlms on silicon substrates. A similar study by Sharma et al.
(26), reported the improved wetting of thin polystyrene ﬁlms in the presence of
carbon black and colloidal silica particles. This is contrary to the conventional
thought of enhanced dewetting by the presence of impurities and other ﬁlm
heterogeneities (14). Also, Mackay et al. (27) showed that poly(benzyl ether)

dendrimers (28) can inhibit dewetting of thin polystyrene ﬁlms.

The inhibition of dewetting by the addition of nanoparticles, therefore, seems
to be a general phenomenon, however, the underlying physics is still a matter of
debate. In this work we present the effect of polystyrene nanoparticles and
macromolecular architecture (29) on the dewetting of polystyrene ﬁlms. The
polystyrene nanoparticles were made from linear polystyrene macromolecules
containing up to ~ 20% pendant cross-linking groups by an intramolecular collapse
strategy where the size and dispersity of the nanoparticle is dictated by the initial
linear precursor polymer. A schematic of this molecule’s molecular architecture is

shown in Figure 1.1.

The polystyrene nanoparticle-linear polystyrene system is unique since there
are simple enthalpic interactions between the linear polymer’s and the nanoparticle’s
chemical constituents and both the linear polymer and the nanoparticle will have
similar interactions with the substrate. Table 1.1 lists the nanoparticles and polymers

used in this study. The polystyrene nanoparticles have already been shown to

O a O 0 0 Linear Chain Precursor

 

 

 

. 3 BC B
T=25°°C ' i W
BCB Ultradilute
Conditions l
T== 250°C ‘
Nanoparticle

Figure 1.1: Schematic of the intramolecularly crosslinked polystyrene nanoparticles
produced from linear chain precursors with ~20 mol% pendent crosslinking groups.

dramatically affect the bulk properties of linear polymers to induce a non-Einstein-

like decrease in viscosity (30, 31).

In this work, an attempt has been made to explain the mechanism of dewetting
inhibition in the presence of nanoparticles. We have used a combination of neutron
reﬂectivity and transmission electron microscopy to determine the stabilization
mechanism induced by nanoparticles when amorphous polymer ﬁlms are heated
above their glass transition temperature. We examine the results in light of the long
range dispersion forces responsible for destabilizing the ﬁlm. We also supplement our

results by using inorganic nanoparticles (cadmium selenide quantum dots) and other

Table 1.1. The molecular weight (MW), Polydispersity Index (PDI) and Radius of
gyration (Rg) or radius (R) of the polymer standards and nanoparticles respectively.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Polymer (PS or
PMMA)/Nanoparticle (NP) MW (kDa) PDI Polymer Rs or NP R (nm)

PS 5 kDa 5.1 1.07 2.0
PS 19 kDa 19.3 1.07 3.8
PS 75 kDa 75.7 1.17 7.6
PS 211 kDa (~ 20% cl) 193.0 1.28 12.1
dPS 63 kDa* 63.5 1.10 6.9
PMMA 76 kDa 76.7 1.06 7.6
dNP 78 kDa* 78 1.14 3.1
NP 41 kDa 41.0 1.04 2.5
NP78 kDa 78.0 1.14 3.1
NP 211 kDa 211.0 1.32 4.3
NP 1.5 MDa 1500.0 1.40 9.0
CdSe quantum dot 34.3 1.01 2.4

 

linear polymers such as PMMA. We conclude by postulating a general mechanism

for dewetting inhibition of polymer ﬁlms in the presence of the nanoparticles, based

on the relative difference in the dielectric properties of the nanoparticle and the linear

polymer involved.

Finally, we use a combination of neutron reﬂectivity and transmission

electron microscopy to show that the nanoparticle induced wetting of a polymer ﬁlm

due to the self-assembly of nanoparticles at the interface can be used to construct the

layered assembly of polymer-nanoparticle sandwich ﬁlms which we call the Self

Assembled Multilayers of Nanocomponents or SAMON process.

 

Chapter 2

Inﬂuence of Molecular Architecture on the Dewetting of Thin

Polystyrene Films

2.1 Introduction

Polymer ﬁlms have numerous technological and biological applications such as
sensors, biomedical devices, insulating dielectric layers, paints, adhesives, etc. Most
of these applications require the presence of a continuous and homogeneous ﬁlm.
However, depending on their thickness, polymer ﬁlms, are either metastable or
unstable on nonwettable substrates (2, 4). The rupture and break up of these ﬁlms is
therefore a problem of technological importance, especially for thin ﬁlms, which are

increasingly required in many applications.

The ﬁlm dewetting rate depends on the surface energy of the substrate,
hydrodynamic boundary condition and ﬁlm thickness (13, 32). Much work has been
done to understand the dynamics of the dewetting process by considering two primary
cases: thick ﬁlms (>100nm) that dewet by nucleation and hole grth and thin ﬁlms
(<100nm) which dewet by spinodal decomposition (12, 13) and/or by ﬁlm
imperfection (14, 33). Films thinner than the polymer coil size are denoted ultra-thin
ﬁlms and they have altogether different dynamics (34). Although these studies have
helped to ﬁirther understand the fundamental mechanism of thin ﬁlm dewetting, a

simple method is still desired to inhibit this undesirable phenomenon.

Various approaches have been adopted to stabilize thin ﬁlms. Film stabilization
by modifying the polymer such as introducing a specialized end group onto the
polymer with a high afﬁnity for the substrate or changing the substrate surface energy
by irradiation or other mechanisms have been common approaches (16, 18-23, 35).
Recently, Barnes et al. (25) discovered that ﬁlllerene nanoparticles inhibit the
dewetting of thin polymer ﬁlms on silicon substrates. This is contrary to the normal
effect of enhanced dewetting by the presence of impurities and other ﬁlm
heterogeneities. Also, Mackay et al. (27) showed that polyether dendrimers (28) can
inhibit dewetting of thin polystyrene ﬁlms. The inhibition of dewetting by the
addition of nanoparticles, therefore, seems to be a general phenomenon, however, the
underlying physics is still a matter of debate. In this paper we present the effect of
polystyrene nanoparticles and macromolecular architecture (36) on the dewetting of
polystyrene ﬁlms. The nanoparticles were made from linear polystyrene
macromolecules containing up to ~ 20% pendent crosslinking groups by an
intramolecular collapse strategy where the size and dispersity of the nanoparticle is

dictated by the initial linear precursor polymer.

Our system is unique in the sense that there are minimum enthalpic interactions
between the linear polymer and the nanoparticle and the polymer-nanoparticle
interactions would be purely a result of entropic interactions. The nanoparticle used in
this study is ca. 4.2nm in diameter (number average molecular mass (Mn) = 25.3 kDa)
and is 20% crosslinked i.e. every ﬁfth monomer unit of the precursor linear
polystyrene chain was crosslinked during the reaction. These nanoparticles have

already been shown to dramatically affect the bulk properties of linear polymers (30).

10

Mackay et al. (30) showed that these nanoparticles, when blended with their linear
analogs, induce a non-Einstein like decrease in viscosity. Here, in this work, we ﬁnd
that these nanoparticles also behave similarly to dendrimers and C60 in inhibiting the

dewetting of polystyrene ﬁlms.

11

2.2 Materials and Methods

The syntheses and characterization of polystyrene nanoparticles used in this
study is reported elsewhere (37). We use a 25.3kDa molecular weight nanoparticle
where 20% of the monomer units are crosslinked (denoted as tightly crosslinked).
Polystyrene standards were obtained from Scientiﬁc polymers (deutrated polystyrene,
dPS, Mw=63.lkDa, PDI=1.1 and hydrogenated polystyrene, PS75kDa, Mw=75.0 kDa,
PDI=1.12; PSl9kDa, Mw=19.3 kDa, PDI=1.07; MW is the weight average molecular
mass and PD] is ratio of weight- to number-average molecular mass; both MW and
PDI were supplied by the manufacturer). The solvent, benzene, was procured from

Sigma Aldrich Co.

Both nanoparticle and polystyrene were puriﬁed to remove silicon containing
compounds (38). This was achieved by digesting any silicon containing contaminants
within the nanoparticle/polymer in hydroﬂuoric acid (HF) for 24 hrs, the acid does
not affect the organic materials. The polymer or nanoparticle was then dissolved in
benzene while still in contact with HF. The two solutions were allowed to settle and
the organic layer decanted. The solution was then dripped in methanol to precipitate
the cleaned polymer/nanoparticle. Separate XPS measurements conﬁrmed that any
silicon containing compounds were removed by the HF treatment. This procedure
was found necessary to ensure reproducibility of the results, especially when the melt

surface tension was measured (39).

Blends were prepared by mixing appropriate volumes of stock solutions
(concentration = 5 mg/mL) to obtain solutions containing 1-20% of nanoparticles by

weight with respect to the linear polymer. Linear-linear blends were also prepared in

12

the same way as were the blends for the neutron reﬂectivity experiments. All the
solutions were ﬁltered through a 0.2-um ﬁlter before Spincoating. The silicon wafer
substrates were used as received from Wafer World Inc. For dewetting experiments,
all the substrates were silanized (40) using Sigmacote (Sigma-Aldrich Co.). This was
done by Spincoating a thin layer of Sigmacote onto the wafer surface followed by

rinsing the surface with water and subsequently Spincoating with the solvent.

The ﬁlms were prepared by Spincoating, at 5000rpm for 40 see, from a benzene
solution onto freshly cleaved mica sheets (41) (Ted Pella 1nc.). Under these
conditions the ﬁlms produced were approx. 40nm thick as determined by
ellipsometry. They were then ﬂoated onto a clean, deionized water surface and picked
up by the silanized substrate. The roughnesses of silanized surfaces and the
polystyrene ﬁlms, as determined by atomic force microscopy (Paciﬁc
Nanotechnology NanoR AFM), were below lnm. The samples for neutron reﬂectivity
measurements were made by direct Spincoating of the polymer solution onto the
silanized surface. The surfaces were, however, silanized using Siliclad (Gelest Inc.)
which has a higher critical surface energy (yc=32mJ/m2) which enabled direct
Spincoating of the benzene solution. All the ﬁlms were dried under vacuum for at
least 12 hrs before annealing. This was done to ensure complete removal of the
entrapped solvent and other contaminants. The ﬁlms were mostly annealed in air but

in some cases the annealing was performed under vacuum.

The ﬁlm morphologies, after annealing, were captured using optical microscopy
in the reﬂection mode. The real time measurements of the samples were performed by

heating them in air under a standard bright ﬁeld light microscope. The neutron

l3

reﬂectivity measurements were performed at POSY2 Neutron Reﬂectometer
(resolution in q-space, Aq/q =0.05) at Argonne National Laboratory, on polymer
blend ﬁlms that were previously annealed under vacuum at 140°C for 2hrs. The
reduced neutron reﬂectivity data was analyzed using Paratt 32 software from HMI

Berlin.

14

2.3 Results and Discussion

Figure 2.1 shows the optical micrographs of the pure nanoparticle (25.3kDa) and
linear polystyrene (PS) l9kDa before and after annealing. Whereas the nanoparticle
by itself does not dewet at all, the familiar pattern of Voronoi polygons (I3) is

observed for the dewetting of the equivalent molecular weight linear polymer.

 

 

 

 

 

 

Figure 2.1. Pure 25.3 kDa nanoparticle ﬁlm before (a) and after (b) annealing at
160°C for 24 hrs. There are no signs of ﬁlm rupture even after such extended
annealing times. However, the linear polystyrene of comparable molecular weight
(PS 19 kDa) dewets in less than 10mins (c & d) on the silanized surface (cf. Figure
3). Film thickness is ca. 40 nm and the length of the scale bar is 100 pm.

This points to the effect molecular architecture has on surface properties. The
spreading coefﬁcient S for the wetting/dewetting of a liquid on a solid substrate is

deﬁned as:
S= "YLA + YSA "YLS (2'1)
where y is the surface energy (tension) for the liquid-air (LA), substrate-air (SA) and

liquid-substrate (LS) interfaces. For wetting, S>0 and for dewetting, S<0. The

15

dispersive surface energy of the silanized surface used in this study is ca. 25mJ/m2 at
25°C as determined by the Fowkes method (42). The surface tension (39) of PSl9kDa
at 160°C is approximately 27mJ/m2. Assuming that the polystyrene chains only
interact with the substrate through dispersive forces, it can be concluded that the
dewetting of the linear polymer occurs at that temperature because the surface tension
of the polymer is higher than the surface energy of the substrate causing the spreading

coefficient, S, to be negative.

The fact that the nanoparticle by itself does not dewet has an important
implication. By simply changing the molecular architecture of the polymer it is
possible to tune its surface properties. We have tried to measure the surface tension of
the nanoparticle at high temperatures but the data does not suggest wetting as
anomalously high forces were measured with the Wilhelmy micro-ﬁber technique
(43) and a melt surface tension of order 150 mN/m is recorded at 240°C. This
suggests an alternative explanation for the unusual wetting by the nanoparticle at
temperatures much higher than its glass transition (Tg=97°C). For example, a yield
stress could stabilize the ﬁlm, yet, rheological measurement indicate a terminal
viscosity for the bulk sample of this nanoparticle (30). We believe that contemporary
rheological characterization (Rheometrics ARES rheometer) of the nanoparticle
system destroys the structure and the micro-ﬁber technique is quite sensitive to this
type of ﬂow property.

To better understand the surface properties of these crosslinked nanoparticles, we
blended them with linear polystyrene of higher molecular weight (Mw=75kDa).

Previously, Barnes et al. (25) and Mackay et al. (27) found that fullerenes and

16

dendrimers inhibit or delay the dewetting of linear polystyrene. In both cases, it was
speculated that the nanoparticles phase separate to the substrate and form a diffuse
layer, which in turn pinned the contact lines of the growing holes thereby arresting
dewetting or changed the surface energy by creating a fractal-like nanoroughness. In

some cases, use of higher concentrations of these nanoparticles completely eliminated

 

H . 'Jsao.1s..»:* .. _ . 7‘ B's "easygei-‘I j“? :,:
"-0 FPO-01. , I, .Jﬁ=.0.95‘:1 . .. ,. -, ,r

      

 

 

 

 

8=1.12 a $1.56 .‘7 e=2.os f
w=0.‘| NJ 5 w=0.2

 

 

 

 

 

Figure 2.2. Optical micrographs of blends of linear polystyrene 75 kDa with 25.3
kDa nanoparticle. All the ﬁlms were annealed for time >24 hrs in vacuum: (a) Pure
PS, (b) 1%, (c) 5%, (d) 10%, (e) 15% and (f) 20% nanoparticle concentration. Note at
0.15-0.20 nanoparticle weight fraction (w), the dewetting is completely eliminated
(aerial fraction, 0 ~ 1.5-2.0). Film thicknesses are ca. 40nm and the length of the scale
bar is 100 um.

dewetting. In both of these studies the molecular weight of the polystyrene used was
of the order of lSkDa, which has a low viscosity and dewets very quickly depending
on the surface energy of the substrate (40). In this work we silanized our substrates to
ensure the reproducibility of the results (44) enabling us to quantify the effect of

nanoparticle induced stabilization of thin ﬁlms on this low energy surface.

17

We blended the 25.3kDa nanoparticle, concentration ranging from 1-20% by
weight, with PS75kDa and Figure 2.2 shows the after annealed optical micrographs of
these blend ﬁlms. The thickness of these ﬁlms was ca. 40nm as determined by
ellipsometry and were annealed for more than 24 hrs at 160°C under vacuum. Clearly
at around 15-20% weight fraction, the dewetting is completely eliminated. To rule out
any molecular weight effects, we blended PSl9kDa with PS75kDa in the same
concentration range as above. The goal of this study was to serve as a control for the
dewetting kinetics of the nanoparticle-linear polymer blends. All the linear-linear
polymer blend ﬁlms dewetted, yet, some discussion on these systems is warranted

prior to discussion of the nanoparticle-linear polymer blends.

Figure 2.3 shows the rate of hole grth of the blends with time demonstrating
that increasing the PS19kDa concentration in PS75kDa actually decreases the rate of
hole growth which is counterintuitive as one would expect the dewetting rate to be
faster since the blend has a lower viscosity than the pure polymer (PS75kDa). The
viscosity varies as MW“, and so the viscosity of the blend will be lower than that of
the pure polymer (MW is the weight average molecular mass). For example, for the 10
wt% blend, the reduction in viscosity is almost 7% (45, 46) with respect to the pure

higher molecular weight linear polymer;

Theoretical description of the dynamics of hole growth has been developed in
previous studies (32, 47, 48). In general the hole radius (R) in a polymer ﬁlm grows
with time (t) as R~t“, where n is the hole growth parameter whose value depends on
the hydrodynamic boundary condition at the polymer substrate interface. In these

studies the rate of hole growth was divided into two regimes: if the polymer was

18

allowed to slip on the substrate the hole radius grows with time as R ~ tz’3 (n=2/3). If
instead there is no slip, then the radius scales as R ~ t (n=1). The polymer is expected
to slip on an extremely smooth surface or an “ideal” surface (48). A silanized surface
is close to an ideal surface, albeit chemically dissimilar, as it is a thin, smooth

compact molecular surface of aliphatic chains.

As shown in Figure 2.3 the rate of dewetting slows with increasing concentration
of the lower molecular weight polymer. Table 2.1 lists the values of the prefactor A
for each of the blends and should be inversely proportional to the viscosity. Even
though there is a substantial decrease in viscosity, the apparent decrease in the value
of A suggests otherwise. It is also interesting to note that the 20 wt % blend does not
follow the usual growth trend as the rest of the blends. The growth exponent (n) for
this blend was found to be ~0.5 unlike 2/3 for other blends. This is clearly shown in
Figure 2.3 when compared to results for pure PSl9kDa and PS75kDa. So, the
dewetting behavior for this blend is outside the perfect slipping regime (14, 15).

The work by Hariharan et al. (49) and Schaub et al. (50) suggest that there is an
entropy driven segregation of the lower molecular weight polymer to the substrate.
The phase separation of the lower molecular weight polymer to the substrate may
alter the spreading coefﬁcient by changing the surface roughness and/or energy at the
interface. Should the lower molecular weight polymer (PSl9kDa) phase separate to
the substrate, the fractional aerial coverage (0) can be estimated through a simple

mass balance as

e = [mag] x 4) (2.2)

19

 

  

 

 

 

 

 

 

 

 

 

 

.Ii r raisini : srstiiti r r
84- D 75kDa
7" o 1%19kDa
°" I 5%19kDa
5“ A 10%19kDa l
A 44- 0 19kDa “g
E
a 3..
(I)
.2
'O 2-
E
.92
o
I 1
10 1‘ "
.i§ :
74- q
6... .1
5“ d
r? I r IlLlllI 1 1111111 1 I r
2 4 e a 2 4 e a 2 4
.1 0 1
10 1O 10
Time (min.)
I r IIIIIII I iiiiiiii 3‘31“
El 75kDa I!-
V 20%19kDa L-
o 19kDa ill .I
I . l
Powerlaw I-I illlll
’e‘ 2’3 - «It'll!
a. . I'IIIILI' i
'o f
9 n
2 ’ '
o
I
" Powerlaw
. 0.471001
II
b
1 11111111 I 11111111 4 1r
2 4 6 a 2 4 a a 2 4
-1 o 1
10 1O 10
Time (min)

Figure 2.3 (a) Hole radius versus time during aging at 160°C for pure linear PS 75
kDa, P819 kDa and their blends along with power law ﬁts. All the systems have been
modeled with a power law model and the growth exponent was ﬁxed at 2/3
(represented by the solid lines). Each data point is an average of ca. 30 holes over 3
samples. The power law prefactor, A, is given in Table 2.1. (b) A 20% PS 19 kDa
blend behaves fundamentally different and an initial power law of ~ 0.5 is seen,
however, this value is not constant. All ﬁlms are ca. 40 nm thick.

20

Table 2.1. Values of the prefactor, A, as determined for the linear-linear blends of
polystyrene shown in Figure 2.3. Note that for pure PS 19 kDa the value of A is 5
times greater than that for pure PS 75 kDa. However, addition of the low molecular
weight polymer actually reduces the value of A with respect to the higher molecular
weight.

 

 

 

 

 

 

 

% PS 1 9kDa Frag/‘21)"
o 8.2
1 8.8
5 3.7
10 5.3
202! 4.1
100 40.4

 

 

 

 

aThe value of A for the 20% blend is based on n = 0.4, all other cases have n = 2/3.

where A is the ﬁlm thickness, Rg, the radius of gyration of the phase separating
polymer and (b, the bulk polymer volume fraction. The radius of gyration (Rg) of
linear polystyrene is given by (51)

RE (nm) = 0.87 x [Mw(kDa)] 0'5 (2.3)
Therefore, for PSl9kDa, Rg = 3.8nm, and a 20% blend in a 40 nm thick ﬁlm has a
fractional coverage (0) of 1.05. This means that at 0.2 volume ﬁ'action there is a
monolayer of the lower molecular weight polymer at the solid substrate. Of course, at
smaller concentrations of the low molecular weight polymer, the fractional coverage
is sub-monolayer.

To understand how this phase separation affects the dewetting kinetics we review
the considerable amount of work done on the wetting behavior of polymer melts on
brushes of identical molecules (autophobicity) (21, 22, 52, 53). It has been
determined that the polymer brush density is an important parameter and that slipping

of a polymer melt on top of brushes of identical molecules depends on the interaction

21

between the ﬁlm and the brush (54-56) with a weaker interaction implying stronger
slippage. Liu et al. (5 7) suggest that an energy penalty in stretching a brush affects the
wetting properties and when the ratio of the polymer to brush molecular weight is in

excess of ~ 5 then dewetting will occur.

Our system is clearly different to a polymer brush; low molecular weight
polymer (probably) phase separates to the solid substrate and both components slip
along it. Note the molecular weight ratio is ~ 4 and dewetting may be predicted. Yet,
when the phase separated polymer approaches a monolayer drastic changes in the
dewetting kinetics appears and so an exact parallel between autophobic dewetting
from a brush and our system can not probably be made. Conjecture to a probable
mechanism is not given here, it is important to note that at all low molecular weight
linear polymer concentrations dewetting is present and a change in the dewetting

kinetics is apparent when a monolayer is present.

As shown in Figure 2.2 the blends of nanoparticle-linear polymer don’t dewet or
there is a marked reduction in the dewetting velocity. This is graphically illustrated
in Figure 2.4, where the rate of hole growth for pure PS75kDa is compared with that
of a 1% blend with the nanoparticle (25.3kDa) and also with the low molecular
weight linear polymer (PS19kDa) at the same concentration. As discussed above for
the 20 % linear-linear blend, the 1% nanoparticle blend does not show perfect
slippage or a complicated dewetting mechanism is present. The rate of dewetting has
drastically slowed even at 1% nanoparticle addition. Indeed, Mackay et a1. (30) found
addition of 1% of the nanoparticle caused a substantial decrease in the viscosity, of

course, implying a faster rate of dewetting.

22

Onset of hole
impingement §

.145. " ii“

nlnLl ail [ll
I V l

Hole radius (um)

 

D 75 kDa Linear PS
Power law 0 1%25 kDa nanoparticle

C
11111.

 

 

: 0.32 A 1%19kDaLinearPS
11111111....141111111....1 11111111.
2 4 68 2 4 63 2 4 68
1 2 3
10 10 10
Time(min)

Figure 2.4. Hole growth as a function of time for pure PS 75 kDa linear polymer and
its 1% blend with 19 kDa linear polystyrene and 25.3 kDa tightly crosslinked
nanoparticle at 160°C. Clearly with just 1 wt% (0 = 0.13) addition of the
nanoparticle, the rate of dewetting has drastically slowed, emphasizing the effect of
molecular architecture on dewetting. Also note that the grth exponent for the 1%
blend is initially 0.32. A11 ﬁlms are ca. 40nm thick

Since bulk transport properties do not appear to explain our results, surface
effects could certainly explain this observation as postulated in the control
experiments performed with the linear 19.3 kDa polymer. We performed neutron
reﬂectivity measurements on a 10 wt% representative sample before and after
annealing. The ﬁlm was ca 17nm thick and was annealed under vacuum for 2 hrs at
140°C prior to the measurement. Figure 2.5 shows the neutron reﬂectivity data for a
10wt% blend of 25.3kDa hydrogenated nanoparticle blended with 63kDa deuterated

linear polystyrene ﬁlm (R is the reﬂectivity). The ﬁtting parameters used for the data

23

 

35x10 ‘iiiiiiiiiiiiiiiiiiiii‘rijtiiiiiiiiiim
.3, A Before Annealing ::

30 52‘. 0 After Annealing 1:
51’. um Nanoparticles at air interface ::

25 5'. 3: -:i

v? ‘t‘. . . . . :’

3 ‘ is Homogeneous Distnbutron of Nanoparticlesﬂ

20 3' ' '3 l

.2 ‘ l

' l

Rq41A‘)

 

1111111111441111111]

wﬁvwff

’ -.

0.12 0.16

 

 

MA")

Figure 2.5. Reﬂectivity multiplied by reﬂectance wave vector to the fourth power
(Rq‘) vs. q for a silanized silicon wafer spin coated with 63 kDa deutrated
polystyrene blended with 10wt% 25.3 kDa hydrogenated, tightly crosslinked,
polystyrene nanoparticles before and after annealing at 140°C for 2hrs under vacuum.
The dotted line represents the reﬂectivity proﬁle ﬁt if the nanoparticles phase
separated to the air interface. The solid lines represent the ﬁts for the before and after
annealed ﬁlms as described in the text.

Table 2.2. Parameters used in Parratt32 software (from HMI Berlin) to ﬁt the data in

Figure 2.5. An air interface is assumed at the top of the ﬁrst layer. The roughness of
layer 2 used in the simulation of the after annealing ﬁlm is extreme and represents an
empiricism to generate a scattering length density proﬁle. Other modeling with many
layers near the wafer surface generates a similar scattering length density (SLD)
proﬁle. The system consisted of 10 wt% 25.3 kDa protonated polystyrene
nanoparticles blended with 63 kDa deuterated linear polystyrene and was annealed at
140 °C for 2 hrs.

 

 

 

 

 

 

 

 

 

 

System treatment Layer Thickness (A) SLD (1045 XA'Z) Roughness (A)
. l 169 5.80 1
Before annealing wafer _ 2.07 1 5
l 142 6.42 1
After annealing 2 21 4.59 21
wafer - 2.07 11

 

 

 

24

are shown in Table 2.2 and the before annealed ﬁlm was modeled as a single layer
with a homogeneous distribution of the nanoparticles. The after annealed ﬁlm was
modeled in many ways including a gradual gradient proﬁle, nanoparticles at the air-
polymer interface (solid line in Figure 2.5) and nanoparticles at the polymer-wafer

interface. Only the latter model predicted the reﬂectivity proﬁle correctly.

To rule out phase separation of the nanoparticles due to any deuteration effects
(58), we did neutron reﬂectivity on similar ﬁlms (data not shown) with different
deuteration contrast (i.e. partially deuterated nanoparticles in hydrogenated linear
polymer and partially deutrated nanoparticles in deuterated linear polymer). Changing
the deuteration did not affect the phase separation of the nanoparticles to the solid
substrate. Also, since both the nanoparticle and the linear polymer have identical
repeat units (styrene monomer), the enthalpic interactions between the nanoparticles
and the linear polymer is minimum (49) and the phase separation of the nanoparticles

to the solid substrate is primarily an entropic effect.

When ﬁtting aged nanoparticle ﬁlms’ data, the roughness at the nanoparticle-
wafer interface was extremely large (cf. Table 2.2). However, this is an empiricism of
the model and the roughness has no physical meaning, equivalent ﬁts were found
upon assuming multiple layers with an equivalent scattering length density proﬁle.
The scattering length density proﬁle generated from this model changed more
gradually than that for a sharp interface, yet, the plateau for Rq4 vs. q is a signature
for a fairly sharp interface. As seen in Figure 2.5 it seems that the nanoparticles phase
separate to the solid substrate surface since other scattering length density proﬁles do

not ﬁt the data at all.

25

1°
0)

 

 

 
 
   

:t—V’ + a T _,_. l T _I .,
s: q
l? i
m 'e f‘
2 0.5 L— :1
.2 E 0::
o-I . —1
I: :1
a .,
g 0.4 :_ 9
z t O 5
a) f "
5 I“ O :1
0 3 — 4'
'5 : 0 El
C C 3
.9 ~- 0 a
‘8: : o 3
1.. L L
LL 02 C .. Z
0 Z . 2
E F '3
2 Z T.
g 0.1 ;- .

 

   

0.0 “ - ‘ I
0 20 40 60 80 100 120 140 160
Distance from Air lnterface(A)

Figure 2.6. Nanoparticle concentration proﬁle determined from the parameters listed

in Table 2.2 for the ‘aﬁer annealed’ ﬁlm. A scaled representation of the nanoparticle
is placed in the lower right-hand comer.

One can convert this proﬁle to a concentration and this was done as shown in
Figure 2.6 with a scaled representation of the nanoparticle in the lower right-hand
comer of the graph. It is clear that the concentration proﬁle changes substantially near
the substrate surface and that the concentration proﬁle correlates almost exactly with
the nanoparticle size. Based on experiments with other deuteration levels (i.e.
partially deuterated nanoparticle in either protonated or deuterated polystyrene) and
substantial modeling we believe that the polystyrene nanoparticles phase separate to
the substrate surface forming a reasonably coherent layer. Note, according to eq 2.1

the nanoparticle areal coverage for this sample should be ~ 0.4 and so a coherent

26

layer is not present at the substrate. Instead a layer of protonated nanoparticles mixed

with deuterated linear polymer is at the substrate.

To ascertain a possible mechanism for this observation, the nanoparticle
fractional coverage at which the dewetting is completely eliminated in Figure 2.2 was
determined. According to eq 2.1, taking the particle diameter (really 2 R3) to be 4.2
nm, which we have determined separately through small angle neutron scattering, we

ﬁnd dewetting is eliminated when a ﬁ'actional coverage of order 1 — 1.5 is present.

To clarify if the areal coverage is the determining factor, a ﬁnal series of
experiments was performed. Changing the ﬁlm thickness while keeping the bulk
concentration constant is equivalent to varying the bulk concentration while keeping
the thickness constant, see eq 1. This was done by keeping the bulk concentration at
10% and changing the ﬁlm thickness from 16 — 78 nm with results shown in Figure
2.7. All ﬁlms were annealed for time greater than 24 hrs at 170°C in air. Again it is
found that an areal coverage of 1 — 1.5 is necessary to eliminate dewetting in

agreement with Figure 2.2.

The physics of the phase separated layer will clearly be different to those for a
brush layer and whether arguments given by others (56, 57) can be used with our
system is unclear. For example, nanoroughness affects surface wetting behavior (59)
and so a nanoparticle rich layer may alter the spreading coefﬁcient by changing the
roughness of that interface. Further, recent simulations (60) show the mechanism of
dewetting inhibition is affected by nanoparticle mobility on a substrate, the
interaction between the nanoparticle and the polymer and also the size of the

nanoparticle. These simulations were based on the work of Barnes et al. (25) where

27

 

 

33nm
0=0.78 8:1.18

 

 

 

50nm

 

 

 

 

 

 

 

65nm 78nm
13-154 6 =1.88

 

 

 

 

 

 

 

 

 

Figure 2.7. Optical micrographs of 10 wt% blends of 75 kDa linear polystyrene with
25.3 kDa polystyrene nanoparticles (tightly crosslinked) at various surface coverage.
The surface coverage (0) was varied by changing the ﬁlm thickness from 16, 33, 50,
65 and 78 nm as indicated in the micrographs. All the ﬁlms were annealed for > 24hrs
in air at 170°C and when 0 > 1.5 the dewetting is almost eliminated. The insets show
the ﬁlm before annealing. The length of the scale bar is 100 um. Note that for the 16
nm ﬁlm, an atomic force micrograph has been shown for the after annealed ﬁlm due
to difﬁculties utilizing optical microscopy with this sample. In this case the image
size is approximately 50 um X 50 um. Also note that for ﬁlms thicker than 50 nm
some complicated surface features (buckling instabilities) are present in both before
and after annealed ﬁlms, due to the Spincoating process.

the ratio of the size of the nanoparticle to that of the polymer Rg was of the order of
0.4. According to ref. 46 a weak interaction between the polymer and the nanoﬁllers
would tend to agglomerate the nanoparticles and enhance the dewetting. In our
system the enthalpic interaction between the nanoparticle and the polymer is almost
negligible yet dewetting can be eliminated.

Finally, even though the nanoparticles phase separate to the solid substrate the
mechanism for dewetting inhibition may be different than that for fullerenes.
According to the Douglas Hypothesis (25, 61) the nanoparticles phase separate to the

28

 

solid substrate and create fractal-like trees that pin contact lines and modify the
conformational energy of the linear polymer. This hypothesis may not apply to our
system due to the nanoparticle size and layer thickness exempliﬁed in Figure 2.6. An
alternate hypothesis is that the phase separated layer merely changes the global
surface energy equivalent to silanizing, yet with polystyrene nanoparticles. Why this
is more effective with a nanoparticle architecture than a linear polymer is not clear

and may be related to the associated dynamics along the substrate surface.

29

2.4 Conclusion

It has been shown that addition of polystyrene nanoparticles to thin polystyrene
ﬁlms inhibits and in some cases eliminates dewetting as long as an enriched layer of
nanoparticles is present at the substrate surface. Since the nanoparticles are
compositionally very similar to the bulk linear polymer, it is apparent that molecular
architecture plays an important role in eliminating dewetting. Addition of low
molecular weight, linear polymer to longer, linear chains does not eliminate
dewetting but the dewetting rate is slower. It is clear that the phase separated
component’s dynamics along the substrate is important to inﬂuencing the dewetting

kinetics and in promoting wetting.

30

Chapter 3

N anoparticle Induced Wetting of Polymer Thin Films

3.1 Introduction

Thin polymer ﬁlms have numerous applications ranging from
microelectronics and sensors to adhesives and bio-medical devices. Yet, polymer
ﬁlms made by Spincoating or dipcoating are frequently unstable and may dewet upon
thermal annealing or solvent exposure. The thickness of a polymer ﬁlm plays a very
important role in governing its stability (1-4). For example, a ﬁlm whose thickness is
on the order of a few micrometers may be stable by gravity whereas a thin ﬁlm (~100
nm) of the same material is either metastable or unstable on nonwettable substrates.
As the ﬁlm thickness approaches molecular dimensions, intermolecular forces of

course become more important (3).

The macroscopic equilibrium contact angle is related to the surface energies in
the system and is frequently used to assess dewetting of various ﬁlm/substrate
systems. However, long range dispersion forces due to induced dipole-induced dipole
interactions also contribute to the surface energy. The strength and magnitude of
dispersion forces depends on the ﬁlm thickness and so the effective interface
potential, which is the sum of the long range and short range interactions acting
across a ﬁlm, ultimately yields the effective surfaces and hence governs ﬁlm stability

(5-11).

In the past decade, much work has been done in studying the dynamics of thin

ﬁlm dewetting. It has been reported in the literature that thick ﬁlms (>100 nm) may

31

dewet inorganic substrates by nucleation and hole growth (2, 4) whereas thin ﬁlms
(<100 nm) dewet by spinodal decomposition (12, 13) and/or by ﬁlm imperfection
(14, 15). Recently, Seemann et al. (8) reported the dewetting of thin polystyrene ﬁlms
on silicon oxide substrates and classiﬁed them as being unstable due to heterogeneous
nucleation or a spinodal process or both. Although these studies have helped to better
understand the process of dewetting, a simple method is still desired to inhibit this

undesirable phenomenon.

Numerous methods to stabilize thin ﬁlms have been reported in the literature
(16-23). Most of these are based on two approaches: a) Surface treatment of the
substrate thereby changing the polymer-surface thermodynamic interaction and/or b)
Modifying the polymer. For example, Henn et al. (21) reported the use of end
ﬁinctional groups with more affinity for the polar substrate, as a means for inhibiting
the dewetting of thin polystyrene ﬁlms. A similar study involving the modiﬁcation of
the homopolymer by the introduction of ionic functionalities was reported by Karim
et al. (19) Yerushalmi and coworkers (22) discussed stabilization of low molecular
weight polystyrene by the addition of long polystyrene chains. In their work the
improved wetting of the styrene oligomers was attributed to the formation of an
entangled network between the free chains and surface tethered molecules. In another
study by the same authors, ﬁlm stability of a polymer melt on a cross-linked network
of itself was investigated. It was observed that the melt wets at both low and high
cross-link densities but dewets at intermediate densities (autophobicity) (20). All
these approaches have their own advantages and disadvantages depending on the

desired application. One disadvantage is that, polymer modiﬁcation to control the

32

wetting behavior of a polymer ﬁlm leads to a change in the ﬁlm properties which

might be undesirable, for example, in sensor applications (24).

Recently, Barnes et al. (25) discovered that fullerene nanoparticles inhibit the
dewetting of thin polymer ﬁlms on native oxide coated silicon substrates. A similar
study by Sharma et al. (62), reported the improved wetting of thin polystyrene ﬁlms
in the presence of carbon black and colloidal silica particles. This appears contrary to
the conventional observation of enhanced dewetting in the presence of impurities and
other ﬁlm heterogeneities (14). Also, Mackay et al. (27) showed that poly(benzyl
ether) dendrimers (28) can inhibit dewetting of thin polystyrene ﬁlms. In an earlier
study (63) we demonstrated the stabilizing effect of polystyrene nanoparticles (37) on
thin polystyrene ﬁlms, highlighting the effect of molecular architecture on this
process. From neutron reﬂectivity measurements we showed that the nanoparticles
were uniformly distributed in the thin ﬁlms (of thickness about 40nm) prior to
annealing, yet after annealing above the glass transition temperature they were found
to separate to the solid substrate, a silanized silicon wafer. Dewetting was eliminated
when the nanoparticles formed a segregated monolayer or above while below this

surface coverage the dewetting dynamics were severely retarded.

In this report we explore the effect of polystyrene nanoparticles of different
molecular weights on the stability of thin polystyrene ﬁlms of various thicknesses.
This extends our previous work (63), where we considered only one nanoparticle size,
to a broad range of nanoparticle sizes. We ﬁnd that larger nanoparticles are more
effective at inhibiting dewetting and provide a entropic argument to support this

observation. We also show that even though CdSe quantum dots segregate primarily

33

to the air surface, they also strongly inhibit dewetting. We argue that segregation of

the quantum dots to the polymer surface is due to the low surface energy of the Oleic

acid brush which is on the surfaces of the organically functionalized CdSe quantum

dots, and that once they have segregated, they form a stable wetting layer at the

air/polymer interface. Tables 3.1 and 3.2 summarize the polymers and nanoparticles

used in this study.

Table 3.1 The molecular weight (MW), Polydispersity Index (PDI) and Radius of
gyration (R8) of the polymer standards.

 

 

 

 

 

 

 

 

 

 

 

Polymer (PS or PMMA) Mw (kDa) PDI Polymer R 3 (nm)
PS 5 kDa 5.1 1.07 2.0
PS 19 kDa 19.3 1.07 3.8
PS 75 kDa 75.7 1.17 7.6
PS 211 kDa (~ 20% clt) 193.0 1.28 12.1
dPS 63 kDa* 63.5 1.10 6.9
PMMA 76 kDa 76.7 1.06 7.6

 

* deutrated ; +cl : cross-link units

Table 3.2 The molecular weight (MW), Polydispersity Index (PDI) and Radius of the

 

 

 

 

 

 

nanoparticles.

Nanoparticle (NP) Mw (kDa) PDI NP Radius (nm)
NF 41 kDa 41.0 1.04 2.5
NP78 kDa 78.0 1.14 3.1
NP211kDa 211.0 1.32 4.3
NP 1.5 Mda 1500.0 1.40 9.0

CdSe Quantum Dot 34.3 1.01 2.4“

 

 

 

 

 

# core radius

34

 

 

3.2 Experimental

The synthesis and characterization of polystyrene nanoparticles and cadmium
selenide quantum dots used in this study are reported elsewhere (3 7, 64, 65) and their
molecular weights along with their polydispersities are listed in Table 3.2. Linear
polymer standards were obtained from Scientiﬁc Polymers and are listed in Table 3.1.

The solvents, benzene and toluene, were procured from Sigma Aldrich Co.

All the polystyrene nanoparticles and polymer standards were checked for any
impurities using XPS and any contamination was removed by appropriate treatment
with hydroﬂuoric acid (63). Nanoparticle blend solutions were made by mixing
appropriate volumes of stock solutions to obtain solutions containing 1-20 wt% of
nanoparticles with respect to the polymer, at an overall concentration ranging from 4-

12 mg/ml. All the solutions were ﬁltered using a 0.2-um ﬁlter before spin coating.

The silicon wafer substrates were used as received from Wafer World Inc. For
both dewetting and neutron reﬂectivity experiments, the wafers were silanized using
Sigmacote ((SiC12C4H9)2O) (Sigma Aldrich Co.). In some dewetting experiments,
OTS or Octadecyltrichlorsilane (C13H37C13Si) (Sigma Aldrich Co.) was also used as
the silanizing agent. The silanization was performed by cleaning the wafers in a
Piranaha bath (70% H2SO4 and 30% H202) for 30 minutes. The cleaned wafers were
then rinsed with copious amounts of millipore water followed by drying with N2. The
dried wafers were then immersed in a 2% silanizing solution in heptane for 2 hrs. The
silanized wafers were then rinsed with chloroform and methanol to remove any
unreacted silanizing agents (66). This treatment of silicon wafers resulted in

substrates with uniform surface energy. The rms roughnesses of the silanized silicon

35

wafers was also below 1 nm as checked with atomic force microscopy (Paciﬁc
Nantechnology NanoR AF M). For some neutron reﬂectivity experiments, the wafers

were used as received and without any silanization.

Thin ﬁlms were spun cast, at 5000 rpm for 40 s, from a benzene or a toluene
solution onto freshly cleaved mica sheets (Asheville-Schoonmaker Mica Co.).
Depending on the concentration of the solution, under these conditions, the ﬁlms
produced were approximately 25-80 nm thick as checked With ellipsometry (J.A.
Wollam ellipsometer). The refractive index of the nanoparticle and polymer ﬁlms as a
ﬁrnction ﬁlm thickness was also measured using the same ellipsometer. For both the
dewetting and neutron reﬂection experiments, the ﬁlms were ﬂoated onto a clean,
deionized water surface and then picked up by the silanized substrate. All the ﬁlms
were dried at 40°C under vacuum for at least 12 hrs to ensure complete removal of the
entrapped solvent and other potential contaminants. The size of the wafers used for
neutron reﬂectivity experiments was 2 inches in diameter whereas those used for the

dewetting experiments was around 1 cm x 1 cm.

All the ﬁlms were annealed in either air or under vacuum and the ﬁlm
morphologies after annealing were captured using optical microscopy in the reﬂection
mode. The real-time measurements of the samples were performed by heating them in
air under a standard bright ﬁeld light microscope. The neutron reﬂection
measurements were performed at POSY2 Neutron Reﬂectometer (resolution in Q-
space, AQ/Q ~ 0.05) at Argonne National Laboratory, on polymer blend ﬁlms that

were previously annealed at 160—170°C for times ranging from 2-24 hrs. The reduced

36

neutron reﬂectivity data was analyzed using Paratt 32 reﬂectivity software from HMI

Berlin.

TEM samples were prepared by transferring the polymer ﬁlm onto an epoxy
resin (Polybed-812, Polysciences, Warrington, PA) which was then cured at 60°C for
24 hrs. The samples were then ultramicrotomed (Power Tome XL, RMC, Tucson,

Arizona) and imaged using a JEOL lOOCX transmission electron microscope.

37

3.3 Results and Discussion

Figure 3.1 shows a cartoon depicting the localization of nanoparticles after
annealing a polymer blend ﬁlm at a temperature greater than the glass transition
temperature (Tg) of the linear polymer. Neutron reﬂectivity measurements have

shown that the nanoparticles are uniformly distributed in the ﬁlm before annealing yet

Polymer Film 0 a) b)
O O o 0 High Temperature

00 000

0 09—5032:

Figure 3.1: A cartoon depicting nanoparticle migration to the substrate after high
temperature annealing. Neutron reﬂectivity measurements have shown that the
nanoparticles are uniformly distributed in the thin (ca. 50 nm) ﬁlm prior to high
temperature annealing (a), yet after annealing they are found to localize at the solid
substrate (b). The aerial coverage of the nanoparticles could be determined by
assuming a hexagonal close packing of the nanoparticles forming a layer 2a thick
with A being the average half gap between the nanoparticles

after annealing they are found to separate to the solid substrate (63) which is
responsible for the retarded dewetting kinetics of the linear polymer. Assuming that
the nanoparticles segregate to a a layer of thickness h at the substrate, the volume

fraction of nanoparticles in this layer, (1)5, is given by,
4% =(h/20) x 4) (3-1)
where h is the ﬁlm thickness, a, the nanoparticle radius and (15, the bulk nanoparticle

volume fraction. As discussed later, we extract it), from analysis of neutron reﬂectivity

data. Assuming that the nanoparticles are spheres regular (triangular), dim, and

38

random, (1),“, dense packing of nanoparticles at the substrate occur at volume

fractions,
11),“ = (2/3)XO.91 (triangular) tbs“ = (2/3)XO.82 (random). (3.2)
It was shown in previous work (63), and is further emphasized here, that
dewetting kinetics is severely retarded for nanoparticle volume ﬁ'actions
corresponding to a segregated nanoparticle monolayer or more, which corresponds to
¢S~0.65 for dense packing. Using this value, we present results in terms of the

normalized number of dense packed nanoparticle monolayers at the surface,
0 = tbs/0.65 ~ 3 tbs/2. (3.3)

The value 0=l corresponds to a dense packed monolayer of nanoparticles at the

substrate.

It is evident from Eq. (3.1) that if the nanoparticles segregate to the substrate,
then it, can be controlled by either varying the thickness of the ﬁlm or the bulk
nanoparticle volume ﬁ'action.’ This control is illustrated by analysis of neutron
reﬂection measurements from thin polymer ﬁlms (thickness, 27-77 nm) containing 5-
20 wt% protonated nanoparticles (Ms/~78 kDa) with deuterated linear polystyrene
(dPS, Mw~63 kDa). Neutron reﬂectivity proﬁles for nanoparticle blend ﬁlms
containing 5-20 wt% nanoparticles and thickness approximately 45nm are given in
Fig. 3.2a, where the proﬁles are offset by a factor of 10 for clarity. All the ﬁlms in
this ﬁgure were annealed in air for 24 hrs at a temperature of 160°C, except for the

5% blend for which annealing was restricted to 2 hrs to limit dewetting. The

39

R (arbitrary units)

R (arbiuaray units)
0

_L
O

.3
O

—l
on

I
N

a
U

 

A

L

 

 

 

0.12

 

-|
.' Cl 27nm
I
I
I‘ |‘
1 . < ------- —H
2 ‘, I
' A I .
0 I _'
- .
:t‘ - .--.
i" .ll ‘1'
. r 51.... .5 ........... .—
. ' I
. r 1 ‘
1° ‘. kt: 1
4‘ I _,.' l- P. .
4"“. A.- " '5‘ | i
1 'Q ’A - :
1 1m -- . e- r. .. .,. ...... _
1.9‘ ‘ _7 -, -.‘ .
1 p..\ " II If 1
J 1‘ . .
,. Ilrllll"
\JO“ ’ ‘
I - l
r ‘\‘ .. . l .—
' 9 _
° C. ; IA -
' ', .ID'IIIA
' O. : L-
“"‘- ’ i d
...... . .-- _. . . .. .
b l ‘5 '0, '
4 ‘ O
\‘ ‘
) “".'-‘-
m‘l‘lx‘t‘.
.

40

0.16

 

6‘- ,_ . ,
i " 1‘; 0 After AnnealingE
3 _ g'", . — NanOparticles at substrate
3, - - - Nanoparticles at air interf. :
2E . « , é - - Homogeneous distribution
b , ‘ |1~ ' I
_ '1 it P) ‘ ' '
10.8 r :5 h‘ . :\ 3‘. s,“ a -(,. 4
YA - r" 4 . p U 0'“, \ A, .' '\ I ' if!“ .; W.
.< 6: .5 ‘ ‘ V, ,\ i a: 1‘); é. \. .0.
V ' ‘ l o
v 5- p ' ~ |
O 41' 1 t 5 ‘6. "
a: 3L 1 . 0‘ "
. .. 0. Q '
2- a ' 0 0 , e o
I U . .. U P O O ' : g p
- c 50 ' o s . o o
.- 0 O . . . "
10" - ’~ ‘ - **** -i.* .
I ‘ ‘ o o g.
6- '0' ' '
' I 0-
5F C) 'c Q.
4..

0.16

o
8
.0
2
o
o
on
o
A
N

Figure 3.2 Reﬂectivity data for nanoparticle blends of deuterated linear polystyrene,
dPS 63 kDa, and polystyrene nanoparticles (Mw~78 kDa). The ﬁlms were annealed in
air at 160°C. The data in (a) and (b) is offset by a factor of 10 for clarity. (a) Data for
nanoparticle concentrations 5, 10 and 20 wt% at ﬁxed ﬁlm thickness, approximately
45nm. The 10 and 20% ﬁlms were annealed for 24hrs, while the 5% ﬁlm was
annealed for 2hrs. b) Data for ﬁxed nanoparticle concentration, 10 wt%, for ﬁlm
thicknesses 27, 39 and 77 nm, after annealing for 6 hrs. c) Reﬂectivity multiplied by
reﬂectance wave vector to the fourth power (RQ’) vs. Q using the data of Fig. 3.2b
(39 nm ﬁlm), and a comparison of ﬁts to three different models (see text for details).

41

reﬂectivity proﬁles in Fig. 3.2b correspond to nanoparticle blend ﬁlms with one
concentration, approximately 10 wt%, of nanoparticles, however the ﬁlm thickness

was varied from 27 nm to 77 nm. These ﬁlms were annealed in air for 6 hrs at 160°C.

Fits to the neutron reﬂectivity data of Figs. 3.2a and 3.2b were based on a two
layer model, where the top layer is composed of pure (deuterated) linear polymer
(dPS 63 kDa) and the bottom layer consists of both the linear polymer and
nanoparticles. For example, for the 39 nm thick ﬁlm of a 10 wt% blend shown in Fig.
3.2b, one ﬁnds the scattering length density (SLD) of the layer next to the substrate to
be approximately 4.72>< 1045 A'2 whereas the top layer has an SLD of 6.42X10’° A'z.
In ﬁtting this reﬂectivity proﬁle, the polymer-nanoparticle interface roughness was
found to be 4 nm and the nanoparticle layer thickness was 6.2 nm. The resolution of
the Q-vector (AQ/Q) for the instrument was also employed as a ﬁtting parameter and

was found to be approximately 0.03.

A test of alternative models is presented in Fig. 3.2c for the case of the 39 nm
thick ﬁlm containing 10 wt% nanoparticles, where a plot of RQ4 versus Q for the data
and various modeling results are presented. Three models are compared to the data in
this ﬁgure, ﬁrstly a model where complete phase segregation to the substrate occurs,
secondly a case of homogeneous distribution of nanoparticles in the ﬁlm and thirdly a
case where the nanoparticles segregate to the air interface. Clearly the latter model
does not predict the observed reﬂectivity proﬁle. The case of homogeneous
distribution shows little difference to the observed reﬂectivity at low Q but does not
represent the data at high Q. Only the case of nanoparticles separating to the hard

substrate, gives a satisfactory agreement to the observed reﬂectivity at both low and

42

high values of the wave vector. In all cases in Fig. 3.2, a critical wave vector (QC)
which is approximately 0.017zt0.001 A"'; denotes the wave vector where the
reﬂectivity falls rapidly from one. If the nanoparticles migrated to the air interface the
critical wave vector would have moved to lower values since the nanoparticles have

lower SLD than the bulk polymer.

These results conclusively show that nanoparticles migrate to the substrate
after annealing and that by changing both the thickness of the ﬁlm and/or the bulk
nanoparticle volume fraction, the ﬁ'actional aerial coverage of the nanoparticles can
be varied. Though segregation to the substrate clearly does occur, some nanoparticles
may remain dispersed with the polymer in the bulk of the ﬁlm. However, all ﬁts
shown in Fig. 3.2 are modeled with the top layer composed of the pure polymer with
an SLD of 6.42X 10'6 A’2 corresponding to complete segregation. For the case of the
39 nm thick ﬁlm, changing the top layer SLD from 6.42><10'6 A'2 to a lower value
(say 5.92><106 A'z) does not drastically change the ﬁt (not shown), though reducing
the SLD below this value does produce a signiﬁcantly poorer ﬁt.

In previous studies (31, 65), we have shown that the nanoparticles have
density similar to that of bulk polystyrene, making the nanoparticle volume and
weight fractions equivalent. For submonolayer coverages, the nanoparticle layer
thickness next to the substrate is approximately given by its diameter (~6.2 nm). By
knowing the polymer and nanoparticle SLD, 6.42X10’° A'2 and 1.41><10'° A’z,
respectively, one is then able to determine the nanoparticle volume fraction (05w) in
the nanoparticle layer. This is related to the monolayer coverage via Eq. (3.3), so that

using 0ng = 3051,90 the fractional monolayer coverage after segregation can be

43

Table 3.3 Values of the fractional coverage (0) based on the nanoparticle
concentration,¢, by assuming perfect segregation (and using Eq. 3.1) and those
determined from neutron reﬂectivity data (05w) using the scattering length density
(SLD) of the layer next to the substrate (see Eq. 5.2), for blends of nanoparticles 78
kDa with dPS 63 kDa . There is good agreement between the values obtained from
neutron reﬂectivity data and those expected for ideal segregation

 

 

 

 

 

 

 

 

Nanoparticle wt% (Film

thickness) SLD (10-6xA-2) 05w lama/21>SLD 0
5 (45 nm) 5.66 0.15 _ 0.23 0.36
Fig. 2a 10 (45 nm) 4.21 0.44 0.66 0.73
20 (45 nm) 1.41 1.00 1.50 1.45
I 10 (27 nm) 5.16 0.25 0.38 0.44
Fig. 2b 10 (39 nm) 4.72 0.34 0.51 0.63
10 (77 nm) 1.88 0.91 1.37 1.24

 

 

 

 

 

 

 

 

directly determined from the neutron data. Using this procedure to extract 95w from
the neutron scattering data yields good agreement with that found by assuming that
all of the nanoparticles segregate to the surface (see Table 3.3). Error estimates on
these values are difficult to quantify and are primarily due to uncertainties in the

parameters extracted from ﬁtting the neutron data.

Figure 3.3 illustrates the effect of varying the fractional surface coverage on
dewetting of thin ﬁlms. It is observed that at fractional coverages 0, less than a
monolayer dewetting occurs, however at coverages greater than a monolayer
dewetting is eliminated in the these experiments. The effect of substrate surface
energy on dewetting behavior of pure polystyrene ﬁlms is illustrated in Fig. 4, by a
comparison of dewetting behavior of linear polystyrene on Sigmacote surfaces, which
have surface energy 28.5i4.9 mJ/m2 at room temperature, as compared to
octadecyltrichlorosilane (OTS) surfaces, which have lower surface energy (z24

44

 

 

Figure 3.3: Optical micrographs of blends of linear polymer (PS 75 kDa) with PS
nanoparticles (Mu/~78 kDa) after annealing the ﬁlms for 24 hrs in vacuum at 160°C
on a Sigmacote substrate. In Fig. 3.33 the nanoparticle fractional coverage (0) is
varied by changing the bulk nanoparticle concentration (0) at a constant thickness of
56 nm, while in Fig. 3.3b the fractional coverage (0) is varied by changing the overall
ﬁlm thickness at ﬁxed nanoparticle concentration of 10 wt %. In both cases a
fractional coverage of a monolayer is needed to severely retard the dewetting of the
linear polymer. The length of the scale bar is 200 um.

mJ/mz). The ﬁlms in these ﬁgures were spuncast on a mica surface, ﬂoated onto a
silanized (Sigmacote or OTS) silicon substrate and then annealed in vacuum. In the
case of OTS, the alkane chain is 18 carbon atoms long (~ 4 nm) while the alkane
chains of a Sigmacote surface contain only 4 carbon atoms (~1 nm) (66). More rapid
dewetting occurs for OTS substrates, as is illustrated in Fig. 3.4 where we present the
rate of hole grth for linear polystyrene ﬁlms (ca. 45 nm thick) at 160°C. In all
cases the hole radius grows with time as R ~ tz’3 implying slip (32, 47, 48). The
dewetting velocity for polymers is inversely related to their viscosity which increases

with molecular weight, as can be seen in the data for PS on Sigmacote. The rate of

45

 

P's 7513155811 'o"T's
PS 75 kDa on Sigmacote
PS 19 kDa on Sigmacote
PS 5 kDa Sigmacote

r FI‘IIIII""I

    
  

 
  
 

DDOOE

  
    

 

 

A 3
E
3 3 -
m .
.2
g .
Dd 2 1
2 :
o :
3: :
10 —'
7 PS75k on OTS -
6 IL I I l lllll 1 llll l l I I III. 1 I all“
2 3 4 5 6 2 3 4 56
0.1 _1 _ 10
Time (mm)

Figure 3.4: Hole radius versus time during annealing in air at 160°C for pure linear
PS 75 kDa, PS 19 kDa and PS 5 kDa along with power law ﬁts on two different
silanized substrates. Use of OTS, as compared to Sigmacote, as the silanizing agent
increases the rate of dewetting for the pure PS 75 kDa. The ﬁlm thickness in all cases
is approximately 45 nm.

hole growth for PS75 kDa on an OTS functionalized substrate is signiﬁcantly faster
than for the same linear polymer on Sigmacote substrates, due to the lower surface
energy of OTS surfaces (32).

The optical microscopy images of Fig. 3.5 illustrate the combined effects of
substrate surface energy and nanoparticle size on dewetting behavior for linear
polystyrene/polystyrene nanoparticle ﬁlms with a constant thickness of around 50nm.

As discussed above the pure ﬁlms dewet at times on the order of minutes while the

46

ﬁlms in this ﬁgure were annealed for a day or more demonstrating the effect
nanoparticles have on this phenomenon. The nanoparticle surface coverage is just
above a monolayer in all cases and the ﬁlms were annealed at 170°C under vacuum.
For the smallest nanoparticles, dewetting occurs on both Sigmacote and OTS
substrates (Figs. 5a, e, i). Nanoparticles of size 6.2 nm (Figs 5b, f, j) are stable after
one day on Sigmacote, but begin to dewet in one day and completely dewet from
OTS substrates after 5 days (Fig. 3.5 f&j). The larger nanoparticles are stable after 5
days OTS substrate (Figs. 5k, 1) and most likely similarly stable on Sigmacote
substrate due to its higher surface energy. Moreover, we noticed that for each of the
nanoparticle blend ﬁlms in Fig. 3.5, increasing the nanoparticle concentration further
above a monolayer, imparts more stability to the ﬁlm against dewetting. This data
supports and extends the results of reference 31 where 5.0 nm nanoparticle (MW ~ 41
kDa) blend ﬁlms were stable on the Sigmacote substrate at concentrations above a

monolayer.

In this system segregation of the nanoparticles to the substrate is driven by an
entropy gain for the entire system, by a mechanism similar to that arising when low
molecular weight polymers are mixed with high molecular weight polymers of the
same composition (67). However in the case of nanoparticles the entropic gain due to
segregation is larger as they have lower intrinsic conformational entropy. Jammed
nanoparticles at the substrate interface lose approximately 3/2 kBT in translation
entropy per nanoparticle, where by is Boltzmann’s constant and T, temperature. We
also found in our previous work (68) that each nanoparticle gains approximately

[a/o]2 x 3 worth of enthalpic contact energy between the nanoparticle and polymer

47

when it is dispersed in the polymer. Here 8: is the components’ monomeric interaction
energy and o is the monomer size, so the nanoparticle loses enthalpic contacts with

the polymer chains as well as translational entropy. Even small enthalpic terms can be

important as is evident in the case of mixtures of deuterated polymers with

 

 

 

Figure 3.5: Optical micrographs of blends of linear polymer (PS 75 kDa) with PS
nanoparticles of 4 different molecular weights: 41 kDa, 78 kDa, 211 kDa and 1.5
MDa. All the ﬁlms have a constant thickness of 503:7.2 nm. The bulk concentration
(4)) of the nanoparticle in the ﬁlms correspond to a nanoparticle fractional coverage
(0) of about a monolayer. All the ﬁlms have been at annealed at 170°C under vacuum.
The top row represents the ﬁlms after annealing (AA) on a Sigmacote substrate and
the middle and bottom rows represent the ﬁlms after annealing on OTS functionalized
substrates. The annealing time for the top two rows is one day. The ﬁlms in the
bottom row are the same as in the middle row but have been annealed for 5 days.
Scaled representations of the nanoparticles are also shown in the micrographs, and
their sizes are also displayed. Increasing the size of the nanoparticle increases the
stability of the polymer ﬁlm against dewetting. The length of the scale bar is 200 um.

48

hydrogenated polymers where phase segregation of the deuterated component to the
air surface occurs due to small differences in enthalpy (69). This loss is countered by
the conformational entropy gain of moving the linear chains away from the substrate.
An estimate is akBTX[a/o]3, with (1 representing the degrees of freedom gained by a

monomer. In order for segregation to the substrate to occur, we must have,
3 3 2
a [a/o] > /2 + [a/o] X E (3.4)

where a has been made dimensionless with kBT and so is of order 0.1 - 1 for
dispersion forces. Since a and o are of order 1 - 10 nm and 0.1 nm, respectively, then
a must be greater than order 0.1 to allow this segregation. We believe this is possible
since a monomer unit will gain at least one degree of freedom and the dimensionless

interstice volume gained at the substrate is also of order one.

The wetting stability of thin polymer ﬁlms as a function of thickness in the
nanometer range depends on long-range van der Waals forces acting across the ﬁlm
(9). Though a complete assessment of these forces is based on sophisticated
electromagnetic theories (3, 70), the qualitative behavior can often be captured using
the pairwise additive theory introduced by Hamaker. Consider the case of polystyrene
(PS) ﬁlm on a Silicon substrate: in case of the Si/PS/Air system, the PS/air interface
is repelled by the Si/PS interface such that the Hamaker constant (AAi,-ps-s,) for this
system is negative and therefore the ﬁlm is stable against dewetting (see Appendix A)
(9). For the multilayer Si/SiOz/PS/Air, the PS/air interface is attracted to the SiOz/PS
interface but repelled by the Si/S102 interface. The long range part of the net potential
(Vd) experienced by the PS/Air interface, assuming additivity of forces and ignoring

any retardation is (3, 8-11):

49

45102 -' Asr _ 45102

= 3.5
d 127481502”)2 127822 ( )

where 515,02 is the silicon oxide layer thickness and h is the polymer ﬁlm thickness.
Here A902 and A5,- are the Hamaker constants for the tri-layer systems Air/PS/SiO;
and Air/PS/Si respectively. The values of A5102 and As, can be estimated from the
optical properties of the three media involved (3) (see Appendix A) and are
approximately equal to 1.56XI0'ZOJ and -2.01>< 10'19 J, respectively. The fact that the
Hamaker constant for the Si/PS/Air trilayer is much larger in magnitude than that for
the SiOZ/PS/Air system is due to the much larger refractive index differences in the
Si/PS/Air case (3). The net Hamaker constant for this system is positive resulting in
the observed dewetting of PS on Silicon oxide substrates (9). Also note the
dependence of the magnitude of the interface potential on the thickness of the
interlayer ((1902). A systematic study of PS dewetting on silicon oxide substrates was
carried out by Seeman et al. (9), where it was reported that the stability of the

polymer ﬁlm could be tuned by the varying the thickness of the $102 interlayer.

A simple rule for deciding whether a (non-polar) trilayer system is stable is by
consideration of the refractive indexes of the three components (see Eq. (5) of the
Appendix). If we assign refractive indexes n1, n3, n2 to a trilayer, then the middle
slab forms a stable wetting layer, provided the refractive indexes, in the order n1, n3,
n2, are either monotonically increasing or monotonically decreasing. Table 3.4
provides the values of the refractive indexes for the materials used in this study. For
example the system Si/PS/Air has values 3.8/1.59/1 and so is stable, while the system

OTS/PS/Air has values 1.4/1 .59/1 and is not stable.

50

F unctionalization of the SiO; surface by an alkane brush (e.g. Sigmacote or
OTS) changes the system to Si/SiOz/brush/PS/Air and modiﬁes the effective potential

seen by the polymer/air interface to:

A5102 " As, ABrush " A3102 _ Aerh

12746180, +81%, +11)2 "122w“, +12)2 12711112

 

Vd=

(3.6)

Here dbmh is the thickness of the brush and Abmh is the Hamaker constant for the
system Air/PS/Brush. The value of the Hamaker constant of the brush can be

estimated from its surface energy by the relation (3):

Table 3.4. The surface energy, dielectric constant and refractive index of the

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

materials.

Material Surface Energy (mJ/mz) 23:12:: Eatiaive

Linear PS 45 2.5 1.59

Linear PMMA 42 3.1 1.49

NP PS 45 2.5 1.59

NP CdSe 36 2.4 1.54

NP Silica 1500 2.3 1.45

NP Polyethylene 36 2.3 1.51

Silicon 1060 11.8 3.80

8102 1500 2.3 1.45

Sigmacote (C-4 alkane) 28 2 1.40

OTS (C-l8 alkane) 24 2 1.40

Abrush-air-brush= 24711357 (3.7)

where Do is the interatomic cut-off separation (~0.165 nm) and y is the surface

energy. The surface energy for alkanes would depend on its thickness (66) while for

51

the case of Sigmacote (dbmsh ~1 nm), the surface energy, as determined by Fowkes
method (42), is of the order of ~29 mJ/m2 whereas that for OTS (dbmsp1 ~4 nm), it is 24
mJ/mz. From eq. 3.5, the Hamaker constants for the pure Sigmacote and OTS are
6.1 ><10'20 J and 5.0><10'20 .1, respectively. The value of the Hamaker constant (Abmsh)

for the trilayer Air/PS/Brush could then be determined using the relation (3),
Alzz Al' (AIA2)”2 (3-8)

with A] being the Hamaker constant of the polymer (PS) and A2 that of the substrate
(brush). Using the value of 6.6><10'20 for the Hamaker constant of PS, one ﬁnds the
Hamaker constant, Abrush, to be 0.3><10'20 and 0.9><10'20 J for the systems
Air/PS/Sigmacote and Air/PS/OTS respectively. Note the difference in the values of
Asm and Abmsh; the latter is almost 3 times lower in magnitude (71 ). Since dewetting
occurs more readily on a slilanized substrate than on a silicon oxide substrate for
similar thicknesses of polymer ﬁlms, it is apparent that in case of the silanized
system, the attraction of the PS/air interface towards the PS/Brush interface is much
stronger than that towards the PS/SiOz interface as also indicated by the values of the
Hamaker constants calculated above. This attraction in the case of the silanized
system should also increase with the thickness of brush and together with the
unfavorable Hamaker constant, explains why the PS ﬁlm dewets more rapidly on the

OTS covered substrate.

As discussed above, insertion of a thicker brush layer (viz. changing from
Sigmacote to OTS) increases the destabilizing force on a polymer ﬁlm. The addition
of polystyrene nanoparticles to thin polystyrene ﬁlms counters this force by forming

an enriched nanoparticle layer at the substrate, effectively “silanizing” yet with

52

polystyrene nanoparticles. Since the nanoparticles are chemically very similar to the
bulk linear polymer, it is apparent that molecular architecture plays a very important
role in retarding dewetting. The mechanism for segregation of the nanoparticles to the
substrate is entropic and this entropic mechanism also favors a uniform nanoparticle
layer at the substrate. Once this layer is in place, the assessment of dispersion forces

on the melt/air interface is altered.

A qualitative discussion of the effect of a layer of polystyrene nanoparticles
(NP) at the substrate can be carried out by again consider a Hamaker-type analysis,
but now using the six layer system Si/SiOlerush/NP/PS/Air with an interface

potential of:

= 45102 ‘As: _ ASi/ane ‘Asx'oz _ ANP 'Asztane _ ANP (3 9)
d 12n(d5,~02 my,“ +de +h)2 127r(d5,,a,,e +de M)2 1274de +h2) 122m2 ‘

 

where de is the thickness of the nanoparticle layer which for monolayer coverage
would correspond to the size (diameter) of the particle. The quantity ANp is the
Hamaker constant for the system Air/PS/NP. Increasing either the SiOz or the brush
layer thickness promotes dewetting. However, addition of a stable layer of
nanoparticles leads to stabilization of the ﬁlm for two reasons: dewetting of the linear
PS from the PS nanoparticles is not favored by dispersion forces, provided the
dielectric properties of the nanoparticles are similar to those of linear polystyrene; and

breakup of the PS nanoparticle layer is entropically unfavorable.

However, we also need to check that the polystyrene nanoparticles used in this
study may have a slightly different refractive index than linear polystyrene, due to
their nanoparticle morphology, crosslinking within the nanoparticle or perhaps due to

interface effects (72). The intramolecular crosslinking strategy used for making these

53

nanoparticles induces a particle like nature to the linear precursor chains (30).
Analyzing the small angle neutron scattering (SANS) data on these particles
demonstrates a peak in the Kratky plot indicative of particle like nature that is more
predominant for higher molecular weight linear precursors. Crosslinking of the
individual chains imposes constraint and changes the morphology from that of a
Gaussian chain to a particle-like structure. To test whether these changes have
signiﬁcant effects on the optical properties, we have measured the refractive index of
linear PS and nanoparticle ﬁlms in the bulk as a function of ﬁlm thickness as shown
in Fig. 3.6. Both the nanoparticle and linear polystyrene have almost the same
refractive index for all ﬁlm thicknesses and the difference in refractive indices
between these two materials is not apparent in this measurement. Note the reﬁ'active
index differences are independent of the ﬁlm thickness and that deviations from bulk
behavior set in only at ﬁlm thicknesses less than about 20nm, validating the
assumption that the polystyrene nanoparticles and linear polystyrene have similar

dielectric properties.

A blend composed of linear polystyrene and cadmium-selenide quantum dots
(diameter ~ 4.7 nm) exempliﬁes a system where dispersion forces and surface energy
compete with entropy in determining nanoparticle segregation. Figure 3.7 shows a
TEM micrograph of the cross-section of four layer ﬁlm of PS 211k. This multilayer

ﬁlm was fabricated using crosslinkable linear polystyrene of molecular weight 21 1k.

54

 

1.6; .
q— _”'9
: U 0? ”~43"
1: 9 ﬂ.—”
1.53:- 91>”
E 35 0
m .._ 1
ﬂ 1: I
.. I
@143:— f
5 SE 0’
E SE .A
"" I
”1.335- .
é :: .-
3: i:- . o PS75kDa
g ?E .' ONP4lkDa
1'23? 4 [:1 NP78kDa
E ' A NP211kD.
1.1j '
o 10 20 30 4o 50 so 70

Thickness (nm)

Figure 3.6: Reﬁactive index as a function of ﬁlm thickness for linear polystyrene, PS
75 kDa (2Rg~ 15 nm) and PS nanoparticles: NP 41 kDa (Za~ 5 nm), NP 78 kDa (2a~
6.2 nm) and NP 211 kDa (2a~ 8.6 nm) at 583 nm. The dotted line is a guide to the
eye.

The ﬁrst layer of the ﬁlm contained a blend of the crosslinkable PS and 22 wt% (4) ~
0.046, 0 ~ 0.75) quantum dots and after spincasting it was annealed for 30 mins at
230°C under vacuum. This annealing enables both segregation of the nanoparticles

and crosslinking of the linear PS providing a solid substrate for the second layer. The

second layer was spin cast using pure PS and then crosslinked in a similar manner.

55

 

 

Figure 3.7: TEM micrograph of the cross-section of PS 211 kDa ﬁlm (~75 nm)
containing 22 wt% quantum dots aﬁer annealing at 230°C for 30 mins under vacuum
(layer 1) on an SiOz substrate. Unlike the PS-PS system, the quantum dots migrate to
both the air interface and the solid substrate. Note that the segregation of the quantum
dots at the air interface is much stronger than that at the substrate. Air interface in the
TEM image is tagged with a 50 nm thick gold layer. The dark grey region is the PS
phase.

The third and fourth layers were fabricated using a similar procedure to that used for
the ﬁrst layer. It is evident from this multilayer experiment that the quantum dots
segregate to both the substrate and to the air interface in the ﬁrst layer, but segregate
almost exclusively to the air interface in the third layer. The hydrocarbon coating
(oleic acid) capping the quantum dots has a lower surface energy relative to the PS
matrix and this provides a driving force for their segregation to the air interface" Of

course, due to the presence of a substrate another entropic force is present which

56

pushes the nanoparticles to the substrate; hence competition occurs. We ﬁnd that
CdSe nanoparticles form a stable wetting layer at the ﬁlm surface and stabilize the

whole ﬁlm.

The ability of quantum dots to promote wetting of polymer thin ﬁlms on OTS
substrates is demonstrated in Fig. 3.8, where it is seen that even a coverage of 0.9
strongly enhances wetting of PS on OTS. A further demonstration of the ability
quantum dots to stabilize thin ﬁlms is presented in Figure 3.9 where stabilization
PMMA thin ﬁlms is demonstrated. In this ﬁgure, we blended PS nanoparticles or
quantum dots with Poly(methyl-methacrylate) (PMMA). PMMA is a well
characterized polymer with a refractive index of 1.49, dielectric constant of 3.1 and
surface energy, YPMMA ~ 41mN/m at 25°C.(73) The optical micrograph of a 56 nm
thick ﬁlm of pure PMMA (Mw ~ 76 kDa) after annealing for 24 hrs at 180°C on an
OTS substrate is shown in Fig. 3.9a. The pure polymer ﬁlm dewets on the OTS
substrate as expected based on dispersion force or surface energy arguments. Optical
micrographs of PMMA ﬁlms containing 34 wt% quantum dots and 14 wt% NP 78
kDa respectively are presented in Figs.9b &c respectively. These ﬁlms were annealed
at 180°C for 24 hrs under vacuum and remain continuous for the duration of the
experiment. For comparison an optical micrograph of a PS 75 kDa ﬁlm containing
l4wt°/o NP 78 kDa is also shown (Fig. 3.9d). The stability of both the CdSe/PMMA
and PS/PMMA systems can be understood based on dispersion force arguments and

by assuming that the CdSe nanoparticles segregate to the air interface, while the PS

57

 

   

 

 

Figure 3.8: Optical micrographs of blends of linear polymer (PS 75 kDa) with CdSe
quantum dots at a fractional aerial coverage (0) of a) 0.3 aﬁer annealing for 1hr and
b) 0.9 aﬁer annealing for 24 hrs. The ﬁlm thickness is ca. 45 nm and annealing was
done at 180°C under vacuum on an OTS substrate. It is interesting to note the
irregular shape of the holes for sub-monolayer coverage (leﬁ ﬁgure).

nanoparticles segregate to the substrate. By comparing Figs. 3.9c,d it is evident that
PS nanoparticles are more effective in stabilizing PMMA than linear PS, which is
also consistent with dispersion force arguments.

We have also discovered two nanoparticles that do not inhibit dewetting of
thin polystyrene ﬁlms to a signiﬁcant extent, namely silica nanoparticles and
hyperbranched polyethylene nanoparticles. In the former case, where the
nanoparticles do not have a brush layer at their surface, dispersion of the
nanoparticles in a common solvent is problematic. In the case of polyethylene
nanoparticles, though segregation to interfaces may occur, the nanoparticles
themselves have a relatively low viscosity and dewet from a silicon substrate.
Nevertheless, these counter examples highlight the difficulty of theoretically

assessing the interplay between solubility and segregation entropy, along with surface

58

         

-. PMIVIA+NP 78 kDa/OTS

3* ..........

PS+NP 78 kDa/OTS

”“4

Figure 3.9: Optical micrographs of thin ﬁlms after annealing at 180°C for 24 hrs
under vacuum on an OTS substrate. a) Pure PMMA 76 kDa ;b) PMMA 76 kDa
containing 34 wt% quantum dots (QDs); c) PMMA containing 14 wt% NP 78 kDa; d)
PS 75 kDa containing 16 wt% NP 78 kDa. The length of the scale bar is 200 pm.
Film thickness and the fractional coverage (6) is given in each ﬁgure.

energy and long range dispersion terms in determining nanoparticle segregation and
their effect on wetting behavior. Moreover, it is clear from the CdSe example that a
brush layer at a hard nanoparticle surface plays an extremely important role in the

dispersion, segregation and wetting behavior of nanoparticle ﬁlled polymer ﬁlms.

59

3.4 Conclusions

We have presented a systematic study of the effect of nanoparticles on the
dewetting behavior of polymer thin ﬁlms. Two different types of nanoparticles were
used to inhibit dewetting, polystyrene nanoparticles and CdSe quantum dots. Neutron
reﬂectivity data provide convincing evidence that polystyrene nanoparticles phase
segregate to the substrate on annealing, while TEM images show that CdSe quantum
dots segregate to the air surface. Though entropic gain drives the nanoparticles to the
substrate, this is offset in the CdSe case by the low surface energy of the Oleic acid
layer on the quantum dot surfaces, leading to the observed surface segregation of

these nanoparticles.

Higher molecular weight polystyrene nanoparticles strongly enhance the
wetting behavior of polystyrene thin ﬁlms, moreover for all polystyrene nanoparticle
sizes a substrate coverage of a monolayer is essential to provide the strongest
inhibition of dewetting. Neutron reflectivity data demonstrate that the nanoparticle
substrate coverage can be changed by varying the thickness of the ﬁlm and/or the
bulk volume fraction of nanoparticles, providing two avenues for control of

dewetting.

Although CdSe nanoparticles preferentially segregate to the air surface of
polystyrene thin ﬁlms, these particles strongly inhibit dewetting of these ﬁlms.
Moreover both PS nanoparticles and CdSe quantum dots also stabilize wetting of

PMMA on OTS substrates.

The segregation of nanoparticles to interfaces is determined by the interplay

between entropic gain, which is dominant in the case of polystyrene nanoparticles,

60

and enthalpic terms which dominate in the case of CdSe nanoparticles. However,
wetting behavior is often most simply understood in terms of long range dispersion
forces. A uniﬁed description of both nanoparticle segregation and thin ﬁlm wetting is

however lacking and remains a major challenge to theorists in the ﬁeld.

61

Chapter 4

Inﬂuence of Polymer Molecular Weight on the Dewetting of Thin

Polymer-Nanoparticle Blend Films

4.1 Introduction

We reported in Chapter 2 that polystyrene nanoparticles made by an
intramolecular collapse strategy eliminate dewetting of linear polystyrene ﬁlms.
Comprehensive neutron reﬂectivity measurements were used to show that the
nanoparticles are uniformly distributed in the ﬁlm prior to annealing, yet, after
annealing they were found to segregate at the solid substrate. Since the chemical
constituents of both the linear polymer and the nanoparticles are similar, the
localization of the nanoparticles at the substrate is primarily an entropic effect. The
mechanism believed to contribute to the stability is that the nanoparticles are forced
to the solid substrate since they lose only a couple of translational modes of entropy
while the polymer would lose many more degrees of ﬁeedom creating an entropic
push to keep the nanoparticles at the interface. This yields a more diffuse interface
generating a stable system by the nanoparticles shielding the adverse van der Waals
forces due to the relative dielectric differences between the nanoparticle and the
polymer.

In this work we report the inﬂuence of polymer molecular weight on the
dewetting of polymer—nanoparticle blend ﬁlms. We analyze the results by ﬁrst

relating the diffusion time of the nanoparticles to the entropic push created by the

62

linear polymer chains as well as the nanoparticle size. We then compare the diffusion
time of the nanoparticles to the dewetting time of the pure polymer and conclude by
hypothesizing that a polymer-nanoparticle blend ﬁlm would remain stable as long as
the diffusion time of the nanoparticles is smaller than the polymer dewetting time

both of which are a function of the polymer molecular weight.

63

4.2 Results and Discussion:

4.2.1 Entropic Push and Dewetting:

The localization of the nanoparticles at the solid substrate after annealing is
responsible for the retarded dewetting kinetics of the polymer ﬁlm. A crude estimate
of the fractional aerial coverage (0) of the nanoparticles at the substrate, based on a

simple mass balance, is given by:
0=(h/2a)><d> (4.1)

where h is the ﬁlm thickness, a, the nanoparticle radius and ¢, the bulk nanoparticle
volume fraction. In formulating this equation, it’s assumed that all the nanoparticles
migrate to the substrate. It was shown in our previous work (63, 74) that the
dewetting kinetics is severely retarded for nanoparticle fractional coverage of a
monolayer or more. Since the nanoparticles are held at the substrate due to a gain in
entropy, the nanoparticle size and the linear polymer radius of gyration would play an

important role in determining the ultimate conﬁguration of the system (75, 76).

Figure 4.1 shows the optical micrographs of polystyrene (PS) ﬁlms of three
different molecular weights, containing PS nanoparticles of three different molecular
weights at a monolayer concentration. The ﬁlms have been annealed for 24 hrs at
160°C under vacuum on a Sigmacote surface. Clearly, increasing the linear PS
molecular weight or the PS nanoparticle size increases the stability of the ﬁlm against
dewetting. As indicated in the ﬁgure, going from left to right or top to bottom
increases the entropic push associated with the nanoparticle size and the linear

polymer radius of gyration respectively.

64

 

Effmmégm '- 1:.

    

           

5:125“; - -, rm; .

 
 

Increasing entropic push

 

    

  

 

 

Figure 4.1: Optical micrographs of blends of linear polymers (PS 5 kDa, PS 19 kDa
and PS 75 kDa) with PS nanoparticles of 3 different molecular weights: NP 41 kDa (a
~ 2.5 nm), NP 78 kDa (a ~ 3.1 nm) and NP 211 kDa (a ~ 4.3 nm) .All the ﬁlms have a
constant thickness of 45 nm. The bulk concentration (4)) of the nanoparticle in the
ﬁlm corresponds to a nanoparticle fractional aerial coverage (9) of about a monolayer
(cf. eq. 4.1). All the ﬁlms have been annealed at 160°C under vacuum on a Sigmacote
substrate for 24 hrs. Clearly increasing the size of the nanoparticle and/or the polymer
Rg increases the stability of the polymer ﬁlm against dewetting. The length of the
scale bar is 200 pm.

In the case of linear PS, increasing the PS molecular weight imparts more
conformational entropy to the chains. To maximize this entropy, the polymer in the
thin ﬁlms expels some of the nanoparticles to walls; in this way, the chains gain
conformational freedom as they do not have to stretch around the particles within the
ﬁlm (75, 76). Since the associated entropy loss would be smaller for smaller chains,

the fraction of nanoparticles expelled to the interface would also be smaller. This is

65

shown in Fig. 4.2a for neutron reﬂectivity data (R vs. Q; R is the reﬂectivity)
measurements on thin ﬁlms (~ 45 nm) of 40 wt% blends of protonated NP 78 kDa
(diameter ~ 6.2 nm) with deutrated polystyrene (dPS) as a function of dPS molecular
weight (dPS 21 kDa & dPS 63 kDa). The ﬁlms have been annealed for 6 hrs at 160°C
under vacuum on silicon oxide substrate. The reﬂectivity proﬁles and the

corresponding ﬁts (solid lines in the ﬁgure) are offset by a factor of 10 for clarity.

The model used to predict the ﬁts for the reﬂectivity proﬁles shown in Fig.
4.2a are based on a three layer model: in the both cases, the layer next to the substrate
is that of pure protonated nanoparticle with a thickness of 6.2 nm and scattering
length density (SLD) of 1.41X 106 A2. In ﬁtting the data, the parameters for this
layer were kept ﬁxed. The top and the middle layer consist of both the nanoparticle
and the polymer (SLD ~ 6.42x 10’6 A'z). The thicknesses as well as the SLDs, for
these layers were employed as the ﬁtting parameters. The interfacial roughness was
also allowed to vary, however, the roughnesses don’t have any physical meaning and
they just represent empiricism of the model since equivalent ﬁts were found upon

assuming multiple layers with an equivalent scattering length density proﬁle.

For the case of a uniform distribution of the nanoparticles in the ﬁlm, the SLD
of the ﬁlm in both cases would correspond to 4.42>< 1045 IV. In the case of the dPS 63
kDa ﬁlm, the top layer SLD was found to be 5.78X 10'6 A'2 while for the dPS 21 kDa
ﬁlm, it was 5.36X 106 A'z. Now, the SLD proﬁle from ﬁts shown in Fig. 4.2a are
converted into concentration proﬁle of the nanoparticles in the ﬁlm and are plotted as

a function of normalized ﬁlm thickness in Fig. 42b. The concentration proﬁles can be

66

A dPSZikDa
0 dPS 593°“-

R (Arbitrary units)

.1.
o
do

.A
O
L

 

 

 

0.00 0.04 10.08 0.12
Q ( A' )

A dPS 21k

dPS 63k
03 ......O ,, . .. ,.

.°
00

Volume Fraction

 

 

0.0 0.2 0.4 0.6 0.8 1 .0
Normalized Thickness

Figure 4.2: Fitted reﬂectivity data for thin ﬁlms of (a) dPS blends with nanoparticle
NP 78 kDa as a function of dPS molecular weight: dPS 21 kDa and dPS 63 kDa. The
bulk nanoparticle concentration in both cases is 40 wt% and the ﬁlm thickness is
approximately 45 nm. The ﬁlms were annealed for 6 hrs under vacuum at 160°C.
Reﬂectivity data has been offset by a factor of 10 for clarity and the ﬁts in both cases
correspond to a three layer model (see text). b) Nanoparticle concentration proﬁle for
the reﬂectivity ﬁts shown in Fig. 2a along with sigmoid ﬁts as a function of
normalized ﬁlm thickness. It is clear that in both cases, the nanoparticle concentration
increases substantially at the substrate. However, under identical annealing
conditions, nanoparticle molecular weight and ﬁlm thickness, ﬁlm containing dPS 21
kDa has 24 % nanoparticles in the bulk from an initial value of 40 %. In the case of
the ﬁlm containing dPS 63 kDa, the bulk nanoparticle concentration is only 12 %
suggesting an increased entropic push for the nanoparticle migration to the substrate
due to the higher molecular weight of the linear polymer.

67

ﬁtted with sigmoid type model where the base values represent the concentration of

the nanoparticles in the bulk of the ﬁlm away from the solid substrate.

From Fig. 4.2b, it is clear that the nanoparticle concentration proﬁle changes
substantially near the solid substrate. Also, in case of the dPS 21 kDa (Rg ~ 3.98 nm)
ﬁlm, the volume fraction of the nanoparticles in the bulk is 0.24 (top layer) from an
initial value of 0.4. However, in case of the dPS 63 kDa ﬁlm, the bulk nanoparticle
concentration is 0.12 from an initial value of 0.4. Therefore, under identical annealing
conditions, ﬁlm thickness and nanoparticle size, the ﬁlm containing dPS 63 kDa
expels more nanoparticles to the wall than the ﬁlm with dPS 21 kDa suggesting an
increased entropic push for the nanoparticle migration to the substrate for the dPS 63

kDa ﬁlm due to its larger molecular weight.

As shown in Figure 4.2, the entropy required to push the nanoparticles to the
substrate scales with the radius of gyration of the linear polymer. This entropic
driving force would also increase with the size of the nanoparticle. This is also shown
in Fig. 4.1 where the PS 5 kDa (Rg ~ 1.95 nm) ﬁlm containing NP 211 kDa (diameter
~ 8.6 nm) at a monolayer concentration is stable against dewetting. In the case of the
bigger particles, if the particles remain within the ﬁlm, the chains would have to
stretch more around the particles and hence the loss in conformational entropy would
be more than that would result from stretching around the smaller particles (75, 76).
Therefore, by increasing the size of the particle for the same chain length, the entropic
driving force for the particle migration to the substrate increases, and the ﬁlms remain

stable against dewetting.

68

4.2.2 Diffusion Time and Dewetting Time:

It is important to note that in the optical micrographs shown in Fig. 4.1, the
substrate (Sigmacote) used has a surface energy of 28.5i4.9 mJ/m2 at room
temperature. By employing OTS (octadecyltrichlorosilane) as the silanizing agent, the
surface energy of the substrate reduces to 24 mJ/mz. Now, consider the case of pure
polymer dewetting on these substrates where we show the rate of hole grth versus
time for three different molecular weight linear polystyrene ﬁlms (ca. 45 nm thick), at
160°C in air, on a Sigmacote functionalized substrate (Fig. 4.3). In all three cases the
hole radius grows with time as R ~ tz’3 implying slip (32, 47, 48). The dewetting
velocity for the polymers is inversely related to their viscosity which increases with
molecular weight, as can be seen in the ﬁgure. For comparison, the rate of hole
growth versus time for PS 75 kDa is also plotted for an OTS ﬁrnctionalized substrate.
Owing to OTS’s lower surface energy the polymer ﬁlm dewets faster. For example,
the dewetting time of PS 75 kDa ﬁlm on the Sigmacote substrate is of the order of 20
mins whereas on the OTS functionalized substrate, it is ~ 3 mins. The data at the

longest time represent the point where the holes begin to impinge.

Figure 4.4 shows the optical micrographs of the ﬁlms shown in Fig. 4.1 but
annealed on an OTS substrate rather than Sigmacote. Clearly, almost all the ﬁlms
have dewetted except for the ﬁlm containing PS 75 kDa in blend with NP 211 kDa.
As explained above, the entr0pic push necessary to expel the nanoparticles to the
substrate increases with both the polymer radius of gyration as well as the
nanoparticle size. For PS 75 kDa and NP 211 kDa system, both of these factors favor

faster entropic segregation of the nanoparticles to the substrate. If the diffusion time

69

 

    
  

,1 mo” P's 7'5'k'152'a'61i kit's ' H I ”11....
7’ 0 PS 75 kDa on Sigmacote

6‘ El PS 19 kDa on Sigmacote

5" A PS 5 kDa Sigmacote

  
  

A
I

 

 

 

A 3
E
3 _
m .
.2 2
'c: -
(U
m ‘:
2 .
O
m i
10: _‘
PS75k on OTS -
L I lllllllnnail 1 I IIIIIIIALLI—
2 3 4 5 6 2 3 4 5 8 2
0.1 _1 _ 10
T1me(m1n)

Figure 4.3: Hole radius versus time during annealing in air at 160°C for pure linear
PS 75 kDa, PS 19 kDa and PS 5 kDa along with power law ﬁts on two different
silanized substrates. Clearly changing the silanizing agent to OTS increases the rate
of dewetting for the pure PS 75 kDa. The ﬁlm thickness in all cases is approximately
45 nm.

of the nanoparticles which in turn depends on the RE of the linear polymer and the
nanoparticle size, is smaller than the polymer dewetting time, then the ﬁlm would
remain stable against dewetting.

Now consider the case of the ﬁlm containing PS 5 kDa and NP 211 kDa. This
ﬁlm is stable on the Sigmacote substrate but dewets on the OTS functionalized
substrate. One can estimate the dewetting time of pure PS 5 kDa ﬁlm on OTS from

the dewetting time of pure PS 75 kDa since the dewetting time scales inversely with

70

 

 

9k+NP41k '

 

 

Figure 4.4: Optical micrographs of blends of linear polymers (PS 5 kDa, PS 19 kDa
and PS 75 kDa) with PS nanoparticles of 3 different molecular weights: NP 41 kDa (a
~ 2.5 nm), NP 78 kDa (a ~ 3.1 nm) and NP 211 kDa (a ~ 4.3 nm). All the ﬁlms have a
constant thickness of 45 nm. The bulk concentration (4)) of the nanoparticle in the
ﬁlm corresponds to a nanoparticle fractional aerial coverage (0) of about a monolayer
(cf. eq. 4.1). All the ﬁlms have been annealed at 160°C under vacuum on an OTS
substrate for 24 hrs. Changing the substrate from Sigmacote to OTS decreases the
dewetting time of the linear polymer. A polymer ﬁlm would remain stable, provided
there is sufﬁcient entropic push for the nanoparticles to migrate to the substrate (see
text). The length of the scale bar is 200 pm.

the viscosity of the polymer. Therefore, one ﬁnds the dewetting of pure PS 5 kDa on
OTS to be on the order of ~0.18 sec. Now, if the diffusion time of the nanoparticles is
greater than the dewetting time of the polymer, the polymer ﬁlm would dewet before

it is stabilized by the particles. We can estimate the diffusion time from the Stokes-

Einstein relation

71

D= kBT/6m1a (4.2)

where a is the nanoparticle raidus and n is the polymer viscosity. Therefore, for a ﬁlm
of PS 5 kDa at 160°C, the polymer viscosity is approximately 20 Pa-s, which for a
nanoparticle of radius ~ 5nm yields a diffusion coefﬁcient of ~ 4000 an/sec. The
diffusion time I would approximately scale as ~ h2/D. Hence for a 45 nm thick ﬁlm, a

diffusion time of the order 0.5 see is estimated.

To get a more quantitative idea on the diffusion time of the nanoparticles in
thin ﬁlms, we performed a series of neutron reﬂectivity measurements on the 30 nm
thick ﬁlms containing 20 wt% NP 78 kDa in blend with dPS 63 kDa as a function of
annealing time. The ﬁlms were annealed for various times at 160°C in air. The
neutron reﬂectivity proﬁles and their corresponding ﬁts (solid lines) of such ﬁlms is
shown in Fig. 4.5a where a plot of reﬂectance, R, vs. the wave vector, Q, is shown as

a function of annealing time.

The proﬁles and the ﬁts have been offset by a factor of 10 for clarity. All the
ﬁts shown in Fig. 4.5a are based on a two layer model. The thickness of the layer next
to the substrate was ﬁxed at 6.2 nm, so were the interfacial roughnesses. The only
parameters that were allowed to vary were the SLDs of the top and the bottom layer.
In addition, the thickness of the top layer was also employed as a ﬂoating variable to
check for mass conservation. Table 4.1 lists the parameters used to ﬁt the data shown

in Fig. 4.5a.

72

 
    
 
   
  
 
 
    

10
0 Before Annealing
102 _ A 3OSec-AfterAnneaing -
n 120 sec-After Annealing
101- v 600 sec-After Annealing _l
3;; 5
g , ".
0 i "
3 10 F If.
5 3‘.
:5 7r
._ .1 . t
'2 1O _ 1': . "£1111”,qu
V .4“ It, 'rilulmlrr
m ,2 . A "o v I" '
10 - ° Ir. a """"'-ru-r.
.. r.
(. O‘ ’1, I/ In”
0 “
10.3 _ . ti“ ((0 ‘n‘u'u‘n -
a: '. u
'f' ““‘.
104 _ “ "r "
"“((("'
I(.‘.

 

 

0.00 0.02 0.04 0.06 0.08 0.10 0.12
Q(A“)

-0- Top layer
-0— Bottom layer

 

Nanoparticle Concentration

 

 

0 100 200 300 400 500 600
Annealing time (sec)

Figure 4.5: a) Fitted reﬂectivity data for 30 nm thick ﬁlms of dPS 63 kDa containing
20 wt% nanoparticle NP 78 kDa as a function annealing time: 0 sec (before
annealing), 10 sec, 120 sec and 600 sec. Thermal annealing of the ﬁlms was done in
air 160°C. Reﬂectivity data has been offset by a factor of 10 for clarity and the ﬁts in
all cases correspond to a two layer model (see text) except for the case of the “before
annealing” ﬁlm which was modeled as a single layer with homogenously distributed
nanoparticles. b) Nanoparticle concentration in the top and the bottom layer as
determined ﬁ'om the scattering length density of the two layers. Just after 120 sec of
annealing time, almost all the nanoparticles have migrated to the substrate with very
few remaining in the bulk. The error in estimating the scattering length density is
about 20-30%.

73

 

Table 4.1. Parameters used in Parratt32 software (from HMI Berlin) to ﬁt the data in
Figure 4.5. An air interface is assumed at the top of the ﬁrst layer.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

System Layer “$31685 SLD (106XA'2) Rou(gAh;1 ess
63.5 kDa deutrated PS blended with 78 kDa protonated l 289 5'42 5
NPs —Before Annealing (BA) wafer , 107 10
1 212 5.83 5
63.5 kDa deutrated PS blended with 78 kDa protonated 2 70 5 58
NPs —Aﬁer Annealing (AA)-30 sec '
wafer - 2.07 10
l 272 6.15 5
63.5 kDa deutrated PS blended with 78 kDa protonated 2 70 2 54 5
NPs -Aﬂer Annealing (AA)-120 sec '
wafer - 2.07 10
l 248 6.33 5
63.5 kDa deutrated PS blended with 78 kDa protonated 2 70 2 36 5
NPs -Aﬂer Annealing (AA)-600 sec '
wafer - 2.07 10

 

 

 

 

 

 

From the SLDs, one can derive the nanoparticle volume fraction in the layer.
For example, for the case of 20 wt% blend annealed for 120 sees, the top and bottom
layer SLDs are 6.15>< 104’ A'2 and 2.54X 10*S A'2 respectively. This corresponds to a
nanoparticle volume fraction in top and the bottom layers to be 0.05 and 0.77
respectively. A plot of nanoparticle volume fraction in the top and the bottom layer as
a function of annealing time is made in Fig. 4.5b. Hence, just after annealing for 2
mins at 160°C, almost all the nanoparticles have migrated to the substrate. Of course,
the data and ﬁts in Fig. 4.5 represent the diffusion of NP 78 kDa (diameter ~ 6.2 nm)

in blend with dPS 63 kDa (Rg ~ 6.9 nm). Nevertheless, it gives a crude estimate of the

74

diffusion time scale of the nanoparticles in thin ﬁlms irrespective of the polymer or

nanoparticle molecular weight.

Finally, for the case of nanoparticle blends of PS 5 kDa on OTS (Fig. 4.4), the
dewetting time of the pure polymer is much smaller than the diffusion time of the
nanoparticles and hence the ﬁlm dewets. A polymer ﬁlm could be made stable as
long as a monolayer of nanoparticles is present at the substrate. It is clear that the
entropic push associated with the polymer radius of gyration and nanoparticle size
together with the dewetting time due to substrate’s surface energy would ultimately

govern the stability of a polymer ﬁlm.

75

Chapter 5

Self-Assembled Multilayers of Nanocomponents

Self assembled, ultrathin ﬁlms function as gas membranes, sensors as well as
photovoltaic devices and capsule structural elements, exemplifying their ubiquitous
nature and application (77-83). Layered self-assembly of amphiphilic materials using
the Langmuir-Blodgett procedure (84) is well known and more recently
electrostatically driven Layer-by-Layer (LbL) assembly of polymeric
multicomposites (85, 86) has been demonstrated. In LbL self-assembly the fabrication
of polymeric multilayers is achieved by a consecutive adsorption of polyanions and
polycations and hence it is driven by electrostatic forces. Extension of the LbL
method to self-assembly of alternate layers of polymers and nanoparticles has also
been done and signiﬁcantly extends the scope of this approach (87). However, LbL
self-assembly cannot be used for non-polar or uncharged nanoparticles and non-polar
engineering polymers, which excludes a wide range of functional materials of
interest. Here we show that layered self assembly of nonpolar, uncharged linear
polymers and nanoparticles can be fully realized using relatively simple and robust

processing steps.

Nanoparticle self-assembly in non-polar, uncharged polymers and nanoparticles
is strongly inﬂuenced by entropic effects, however local enthalpic terms and long
range dispersion terms can also be signiﬁcant. Kinetic effects such as jamming and
self-assembly during drying are also important in some situations effectively trapping
the structures (88-90). We ﬁrst show that entropic effects due to architecture

differences can drive self-assembly of multilayers by using unique polystyrene

76

nanoparticle - linear polystyrene mixtures where the difference in monomer -
monomer enthalpic effects are minimized. Here the nanoparticles assemble at the
solid substrate to maximize the system entropy. We then show that multilayers
formed from Cadmium Selenide quantum dots and linear polystyrene are controlled
by the interplay between surface energy, dispersion forces and entropy. In this
system, the nanoparticles primarily segregate to the air interface yet multilayer
fabrication remains facile. A third system consisting of a multilayer of two
incompatible polymers, namely linear polystyrene and linear polymethylmethacrylate
(PMMA), where Cadmium Selenide quantum dots are used to stabilize the multilayer,
demonstrates the capability of our processing technique to incorporate a wide range

of polymer and nanoparticle combinations.

We have recently shown (63), using neutron reﬂectivity experiments, that
polystyrene nanoparticles made by an intramolecular collapse strategy (37, 65)
blended with linear polystyrene, are uniformly distributed in a spuncast thin ﬁlm (ca.
40 nm thick). Yet, after annealing the ﬁlm above the glass transition temperature of
the linear polymer, they were found to segregate to the solid substrate. Separate
experiments with different deuteration contrast ruled out migration of nanoparticles
due to any isotopic effect (58, 67). Also, since the nanoparticles and linear polymer
have identical repeat units (styrene monomer), adverse enthalpic interactions between
the linear polymer and the nanoparticles are minimal (91), and the migration of the
nanoparticles to the solid substrate is primarily an entropic effect (92). Nanoparticle

localization to an interface (75, 76) has great utility since it changes a range of

77

physical and mechanical properties of thin ﬁlms and it inhibits dewetting of polymer

ﬁlms (25, 61, 74), a phenomenon we use in the present work.

The key observation (93) is that a thin ﬁlm initially composed of a uniform
mixture of nanoparticles and polystyrene may be annealed to form a bilayer
consisting of a nanoparticle rich phase at the solid substrate and a polymer rich phase
at the air interface. Below we show that this process may be repeated to form
multilayers and that similar processing may be used for a wide range of nanoparticle
and polymer combinations. We call this the Self Assembled Multilayers of

Nanocomponents or SAMON process.

The process of entropy driven enrichment of nanoparticles at the silicon wafer
substrate is demonstrated in Fig. 5.1A where neutron reﬂectivity measurements (RQ4
vs. Q, R is the reﬂectance and Q, the wave vector) on a polymer ﬁlm containing 10
wt% protonated polystyrene nanoparticle (211 kD) blended with deuterated linear
polystyrene (63 kD) show a distinct change before and after annealing (see
Supporting Online Material for experimental details). Before annealing, the ca. 40 nm
thick ﬁlm was accurately modeled as a single layer with a homogeneous nanoparticle
distribution corresponding to an average scattering length density (SLD) of 5.92><10'6
A'Z, the solid line in the ﬁgure demonstrates the goodness of the ﬁt to the data. Here
the SLD of the pure deuterated polymer and that of the nanoparticle is 6.42><1045 A'2
and 1.41 ><10*S A'Z, respectively. The reﬂectivity proﬁle undergoes a profound change
after annealing for 2 hrs. at 160°C as demonstrated by the data presented in Fig. 5.1A
along with the results of using a two layer model with a nanoparticle rich layer at the

solid substrate. Note the nanoparticle surface coverage is approximately one-half a

78

10‘1

 

R0‘(A‘)
3
I IIIIIII"‘7 ¥lllll

0‘“.

A!
I

 

1 1 .....1

1441

 

 

Nanoparticle Volume Fraction

 

((u«((m«(((«((«11111111110
0.2 0.4 0.6

Normalized Thickness

 

0.12

    

 

 

 

 

 

 

18x10"PIIIIIIII|II|II|IIIIlllllIlllI'lIII|IIIIIIIII|IIIu
16 ; D O 0 Six layeraeeembly .:
E r - - - Homogenous trilayer E
12 :— -:
f“ : :
it, 10 :- -_
Q
O I 2
It 8 .- -:
6 Z- -I
4 I— -Z
l- .1
2 L —:
0 IIII’II‘I:I-ll.'l'l:l1l-llll1 ' I ' ' ' ' ' 'Irririllnl:
0.00 0.02 0.04 0.06 0.08 0.10

79

a 1A")

Fig. 5.1 (A) Reﬂectivity (R) multiplied by reﬂectance wave vector (Q) to the fourth
power (RQ‘) versus Q for a silicon wafer spincoated with 63 kD deuterated
polystyrene blended with 10 wt% 211 kD protonated polystyrene nanoparticles before
and aﬁer annealing at 160°C for 2 hrs under vacuum. The dotted line represents the
reﬂectivity proﬁle ﬁt if the nanoparticles migrated to the air interface. The solid lines
represent the ﬁts for the before and after annealed ﬁlms as described in the text. Film
thickness is approximately 40 nm. (B) Nanoparticle concentration proﬁle determined
from the scattering density proﬁle for the ‘after annealing’ ﬁlm shown in A. A scaled
representation of the nanoparticle is placed in the lower right-hand comer. (C)
Spincoating process to make the multilayered ﬁlms. A nanoparticle — polymer
mixture is spincoated onto a substrate and thermally annealed to allow nanoparticle
segregation and polymer crosslinking. This process is repeated to form multiple
layers as demonstrated by the numbers: 1, 2, . (D) RQ4 versus Q for a silicon wafer
spin coated with three layers of 211 kD protonated linear polystyrene containing 20
mol% crosslinker, blended with 15 wt% 78 kD deuterated polystyrene nanoparticle
aﬁer annealing each layer at 230°C for 24 hrs under vacuum. The ﬁt (solid line)
corresponds to six alternating layers of protonated polymer and deuterated
nanoparticle while the dotted line is the prediction if the nanoparticles were
homogeneously distributed. The thickness of each polymer -nanoparticle layer is
approximately 44 nm.

80

monolayer in this example, as determined by a simple mass balance assuming that all
the nanoparticles are located at the substrate (63), is thus conﬁrmed by the reﬂectivity

measurement.

The solid line in Fig. 5.1A corresponds to a model where the top layer consists of
the pure deuterated linear polymer and the bottom layer contains a combination of the
deuterated linear polymer and the protonated nanoparticles with an interface
roughness of 5 nm comparable to the nanoparticle diameter (20) of approximately 8.8
nm. The results of using an alternative model where the nanoparticles segregate to the
air interface yields the dotted line in Fig. 5.1A. This data and further analysis, using a
range of alternative models, clearly indicates that the nanoparticles migrate to the
solid substrate after high temperature annealing. This is further illustrated in Fig.
5.13, where the concentration proﬁle of the annealed ﬁlm has been extracted from
the reﬂectivity data. A scaled representation of the nanoparticle is also shown in the

lower right-hand comer of this ﬁgure.

To make multiple polymer-nanoparticle layers stacked on top of each other, we
spincoated an 85 wt% polymer — 15 wt% nanoparticle blend on top of a previously
aged ﬁlm via the procedure shown in Fig. 5.1C. The polymer was made from a ca.
200 kD random copolymer of 80% styrene and 20% benzylcyclobutane (BCB)
stabilized from dissolution during the subsequent Spincoating operation by heating to
230°C for 24 hrs. to activate the crosslinking process between adjacent BCB groups.
Subsequent experiments demonstrated a signiﬁcantly decreased aging time is actually
required. The 78 kD partially deuterated, crosslinked polystyrene nanoparticles (3 7)

are found to segregate to the substrate prior to completion of the crosslinking process

81

and allows repetition of this procedure two more times to give a six layer system with
each bilayer being about 44 nm in thickness. This was conﬁrmed by neutron
reﬂectivity measurements (Fig. 5.1D), where modeling conﬁrmed six layers,
demonstrated by the inset, with the nanoparticles at the solid substrate in each bilayer
(see Supporting Online Material). Modeling the nanoparticle distribution as if they
were homogeneously distributed, as shown by the dotted line, or at the air interface,
did not represent the data and so it is believed the reﬂectivity data does indeed show

the segregation given in the inset.

In this system segregation of the nanoparticles is driven by an entropy gain for
the entire system which has been shown to be important when cracks form in
nanoparticle ﬁlled polymers (93). Yet, one expects the translational entropy loss of
jammed nanoparticles at the substrate interface is approximately 3/2 kBT per
nanoparticle, where [(3 is Boltzmann’s constant and T, temperature, which does not
explain the segregation. We also found in our previous work (91) that each
nanoparticle gains approximately [a/o]2 x 5 worth of enthalpic contact energy
between the nanoparticle and polymer when it is dispersed in the polymer. Here [1 is
the components’ monomeric interaction energy and o is the monomer size, so the
nanoparticle loses enthalpic contacts with the polymer chains as well as translational
entropy. This loss is countered by the conformational entropy gain of moving the
linear chains away from the substrate. An estimate is ak3T><[a/o]3, with 0 representing
the degrees of freedom gained by a monomer unit when it is released from constraints
or more precisely the roughness (interstice) volume induced by the nanoparticle layer

made dimensionless with a nanoparticle volume. The conformation entropy gain of

82

the polymer should be greater than the translation entropy and mixing enthalpy loss

of the nanoparticle or
a [0/0'13 > 3/2 + [a/a]2 x s (5.1)

where a has been made dimensionless with kg?“ and so is of order 0.1 - 1 for
dispersion forces. Since a and o are of order 1 - 10 nm and 0.1 nm, respectively, then
a must be greater than order 0.01 - 0.1 to allow this segregation. We believe this is
possible since a monomer unit will gain at least one degree of freedom and the

dimensionless interstice volume gained at the substrate is also of order one.

The versatility of this process is further demonstrated by the ability to replace the
polystyrene nanoparticles with inorganic-based materials. Though the above entropic
and enthalpic terms are always important in nanoparticle segregation, other enthalpic
terms play an important role for these systems. Dispersion of CdSe quantum dots in
non-polar polystyrene is made possible by attachment of Oleic acid chains to the
quantum dot surfaces to yield a sterically stabilized system that is soluble in Toluene.
Phase segregation of the quantum dots from linear polystyrene, in thin ﬁlms, is
clearly evident in transmission electron microscopy (TEM) images shown in Fig. 5.2.
We note that these quantum dots are completely soluble in bulk polystyrene, as
occurs for others systems where nanoparticle architecture enables bulk miscibility,
with a particularly notable case being dendritic polyethylene (94) in polystyrene (91).

The TEM image in Fig. 5.2A shows eight bilayers self-assembled with the
SAMON process (Fig. 5.1C) using the same linear polystyrene as above, having 20
mol% BCB groups. Each quantum dot layer is close to a monolayer coverage

(approximately 34 nm thick) and the thickness of each polymer layer is about 75 nm.

83

 

 

Fig. 5.2 (A) Transmission electron micrograph (TEM) of an assembly of sixteen
layers: eight CdSe quantum dots alternating with eight crosslinked polystyrene layers,
assembled on a silicon wafer. Each bilayer is numbered on the micrograph from 1 to
8. The quantum dot layers are approximately a monolayer thick (~3-4 nm) whereas
the polystyrene layers are about 75 nm in thickness. In all the micrographs a gold
layer was sputtered on the ﬁlm after fabrication to mark the air interface and masks
the uppermost quantum dot layer. The silicon wafer was treated with hydroﬂuoric
acid and removed prior to microtoming as described in the Supporting Online
Material. There are some quantum dots next to the substrate as discussed in the text.
(B) Assembly of six layers formed by aging four layers to form three layers of CdSe
quantum dots and three layers of crosslinked polystyrene. This sample was made by
aging a polystyrene - quantum dot layer then Spincoating a pure polystyrene layer on
top and crosslinking it then repeating the process. The quantum dot layers are about a
monolayer thick. The inset shows a TEM micrograph of the ﬁrst layer normal to the
substrate surface demonstrating a reasonably uniform ﬁlm. (C) Assembly of eight
layers: four CdSe quantum dots and four crosslinked polystyrene layers, assembled
on a silicon wafer. In this case, the quantum dot layers are approximately 13 nm thick
while the polystyrene layers are 15 nm thick to demonstrate control over the thickness
of each layer is dictated by the initial conditions developed through the Spincoating
operation.

84

We believe the quantum dots assemble at the air interface in this system with the
exception of the ﬁrst layer, layer 1 in the ﬁgure, where they are at both interfaces.
This is made clear by viewing Fig. 5.28 which has the following layer deposition
scheme: layer 1, polymer + quantum dots; layer 2, pure polymer; layer 3, polymer +
quantum dots; layer 4, pure polymer; with each layer being processed by thermal
aging after Spincoating to activate the crosslinking process before the subsequent
layer is deposited. Some quantum dots have assembled at the substrate interface in
layer 1, yet, most have segregated to the air interface. This is more evident by
viewing the interface between layers 2 and 3 and 3 and 4. Here it is clear that the
quantum dots in layer 3 have mostly gone to the air interface which is subsequently

covered by a pure, crosslinked polymer layer.

The assembly is easily described by careful consideration of the Hamaker
constant for trilayers making-up a multilayer assembly. If the constant is negative
then that trilayer is stable by and the effective interface potential positive to ensure
stability (10). If we consider a trilayer of air (component 1) — quantum dots (3) —
polystyrene (2) then one can determine the sign of the Hamaker constant (A132) using
(3)

A132 ~ [n12 — n32] >< [n22 — n32], (5.2)
which is a good heuristic for nonconducting materials. Here n,- is the refractive index
of component i with the following approximate values: 1.0 (air), 1.54 (quantum dots)
and 1.59 (polystyrene). The value for the quantum dots’ reﬁ'active index was arrived
at by computing a volume average of a CdSe inner core with a 2.2 nm radius

(refractive index of 2.8) surrounded by an Oleic acid layer which is 2.5 nm thick

85

(refractive index of 1.4). The Oleic acid layer thickness was determined by dynamic
light scattering of a dilute Toluene solution and is a reasonable value based on the
chemical structure. With these values, the ordering of air — quantum dots -

polystyrene is stable while others are not.

Of course, this type of assembly requires similar forces as described by Gupta et
al.(93). However, since the nanoparticles are presumably homogeneously dispersed
after the initial Spincoating step, they must rapidly diffuse to form the stable
conﬁguration before dewetting occurs. Using the Stokes-Einstein relation and the
viscosity for the polystyrene melt (95) a diffusion coefﬁcient of ca. 50 an/s is
calculated. Since the layer thickness is of order 50 nm then approximately one minute
is required for the nanoparticles to diffuse to either interface. This time scale is so
small we believe the dewetting behavior is stabilized throughout the diffusion process
as nanoparticles accumulate to their stable conﬁguration thereby prohibiting
nucleation and growth of holes. Some nanoparticles are trapped at the unstable
position, for example near the substrate, either due to the entropic stabilization
described by Eq. 5.1 or by kinetic means where local scale crosslinking conﬁnes them
at the given position. This later hypothesis seems unlikely since we have not observed

quantum dots trapped in the middle of the ﬁlm.

Much thicker quantum dot layers and thinner polymer layers can also be formed
as demonstrated in Fig. 5.2C where ca. 13 nm thick quantum dot layers have been
assembled with 15 nm thick crosslinked polystyrene. Again, the ﬁrst layer shows a
thin quantum dot layer at the substrate with most of them located at the upper part of

this ﬁlm. Subsequent ﬁlms show alternating layers of the two components which are

86

        
   

   
  

‘ 3.5-‘11 #2.]; - .
.1 “gi‘j‘iljf'i‘ >

MMA/PS/SiOZ/Si

 
   

1" ism-4* .a'
PS/PMMA/SlOZ/Sl

: PS+QDs/PMMA/Si02/Si

 

 

Fig. 5.3 (A) Optical micrographs of a 75 kD polystyrene ﬁlm (~58 nm thick) ﬂoated
onto a 76 kD ﬁlm PMMA ﬁlm (~56 nm thick) and thermally aged at 180°C for 24
hrs. under vacuum. Isolated polystyrene drops can be seen on the surface of PMMA
after annealing the bilayer assembly on a silicon (Si) wafer with the native oxide
(SiOz) layer. (B) PMMA ﬁlm ﬂoated on polystyrene subject to the same annealing
procedure given in A to show a similarly unstable ﬁlm. The instabilities shown in A
and B disappear in C and D, respectively, when the top layer is replaced by a
composite ﬁlm composed of both the quantum dots (ODS) and the polymer. The ﬁlm
ordering is given in the ﬁgure with the abbreviations listed above, the length of the
scale bar is 200 um.

87

not as coherent as the layers formed with a lesser amount of quantum dots, Figs. 5.2
A&B, although they are certainly distinct. We believe the layers can be further

reﬁned through optimization of the processing conditions.

Generalization of the SAMON technique to incompatible, uncrosslinked polymers
and nanoparticles is demonstrated in Fig. 5.3 where optical micrographs of PMMA
and polystyrene polymers are considered. The ﬁrst layer, either PMMA (Fig. 5.3A) or
polystyrene (Fig. 5.38), was spincoated onto the silicon wafer, that has its native
oxide layer, followed by ﬂoating the other polymer on top and aging the composite
for 24 hrs. at 180°C.. Both systems were found to dewet as expected, however, when
the top layer contained quantum dots the dewetting was eliminated. Previous work
has demonstrated that other nanoparticles will slow the dewetting dynamics (96),
however, our work shows complete elimination of dewetting. So, the SAMON
process applies to a wide range of nanoparticles, to incompatible polymers, and
stabilization may be carried out both with or without crosslinking, yielding a robust
procedure for self-assembly of functional multilayers from non-polar nanoparticles

and polymers.

88

5.1 Supporting Online Material
5.1.1 Experimental

The synthesis and characterization of polystyrene nanoparticles, their linear
analogues and cadmium-selenide quantum dots used in this study are reported
elsewhere (3 7, 64, 65). Linear polystyrene and poly(methyl methacrylate) (PMMA)
standards were obtained from Scientiﬁc Polymers and the solvents, benzene and
toluene were procured from Sigma Aldrich Co. Nanoparticle blend solutions were
made by mixing appropriate volumes of stock solutions to obtain solutions containing
10-40 wt% of nanoparticles with respect to the polymer, at an overall concentration
ranging from 5-12 mg/ml. The silicon wafer substrates were cleaned and treated as
described below and the layered assemblies were prepared by Spincoating the
appropriate solution onto the substrate followed by thermal annealing. All the
solutions were ﬁltered using a 0.2-um ﬁlter before spin coating. The rrns roughnesses
of the silicon wafers was also below 1 nm as checked with atomic force microscopy
(Paciﬁc Nantechnology NanoR AFM) and were used as received after degreasing

with solvent (benzene/toluene).

The molecular weights along with the polydispersities of the nanoparticles and
linear polymers used in this study are listed in Table 5.1. The silanization was
performed by cleaning the wafers in a Piranaha bath (70% H2804 and 30% H202) for
30 minutes. The cleaned wafers were then rinsed with copious amounts of millipore
water followed by drying with N2. The dried wafers were then immersed in a 2%
silane solution in heptane for 2 hrs. The silanized wafers were then rinsed with

chloroform and methanol to remove any unreacted silane (66). This treatment of

89

silicon wafers resulted in substrates with uniform surface energy and was only

performed with the silicon wafer used in Fig. l A&B.

Table 5.1. The molecular weight (MW), Polydispersity Index (PDI) and Radius of
gyration (Rg) or radius (R) of the polymer standards and nanoparticles, respectively

 

 

 

 

 

 

 

 

 

Polymer (PS)/Nanoparticle (NP) Mw (kDa) PDI PS Rgor NP R (nm)

PS 75 kDa 75.7 1.17 7.6

PS 211 kDa (20% cl) 193 1.28 12.1

dPS 63 kDa‘ 63.5 1.10 6.9

dNP 78 kDa‘ 78 1.14 3.1
NP211kDa 211 1.32 4.3

PMMA 76 kDa 76.7 1.06 7.6
CdSe Quantum Dots 34.3 1.01 2.4 ”

 

 

 

 

 

* deutrated; # core radius

Thin ﬁlms were spun cast, at 5000 rpm for 40 s, from a benzene or a toluene
solution and depending on the concentration of the solution, under these conditions,
the ﬁlms produced were approximately 20-80 nm thick as checked with ellipsometry
(J .A. Wollam ellipsometer). All the ﬁlms were annealed under vacuum and the ﬁlm
morphologies after annealing were captured using optical microscopy in the reﬂection
mode.

5.1.2 Neutron Reﬂectivity Data Analysis

The neutron reﬂection measurements were performed at POSY2 Neutron
Reﬂectometer (resolution in Q-space, AQ/Q ~ 0.05) at Argonne National Laboratory,

on polymer blend ﬁlms that were previously annealed at 160-230°C for times ranging

90

from 2-24 hrs. The reduced neutron reﬂectivity data was analyzed using Paratt 32

reﬂectivity

91

Table 5.2. A) Parameters used in Parratt32 software (from HMI Berlin) to ﬁt the data
in Figure 5.1A. An air interface is assumed at the top of the ﬁrst layer. The before-
annealing ﬁlm is modeled as a single layer with a homogenous distribution of
nanoparticles corresponding to an average scattering length density (SLD) of
5.92X lO'bA'z. The aﬁer annealed ﬁlm was modeled as two layers with the layer next
to the substrate representing the nanoparticle rich phase. B) Parameters used to ﬁt the
data in Figure 5.1D (solid line). An air interface is assumed at the top of the ﬁrst
layer. The roughnesses used in the simulation of the after annealing ﬁlm has no
physical meaning and represents an empiricism to generate a scattering length density
proﬁle. Modeling a homogenous distribution of 3 layers generates a reﬂectivity
proﬁle represented by the dotted line in Fig. 5.1D.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

A s stem La er T“°""°” SLD mod-1 “"3“”
63.5 kDa deutrated PS blended with 211 kDa protonated l 375 5'92 ‘7
NPs —Before Annealing (BA) wafer _ 2.07 30
63.5 kDa deutrated PS blended with 211 kDa protonated 1 323 6'42 2
NPs -Aﬁer Annealing (AA)-Nanoparticles at the 2 80 4.8 50
substrate wafer - 2.07 26
63.5 kDa deutrated PS blended with 211 kDa protonated 2 80 4'8 50
NPs —Aﬁer Annealing (AA)-Nanoparticles at the air 1 323 6.42 2
‘me'face wafer - 2.07 26
Thickness Roughness
S stem La er SLD 10‘8A‘1
l 440 1.87 1
211 kDa PS with 20 mol% crosslinker blended with 78 2 440 1.87 20
kDa deutrated NPs -Three layers after annealing (AA)-
Homogeneous distribution 3 440 1'87 1
wafer - 2.07 20
1 375.8 1.50 0.7
2 64.6 4.51 1.4
3 313.5 1.66 1.4
211 kDa PS with 20 mol% crosslinker blended with 78 4 121 9 3 29 29
kDa deutrated NPs —Three layers after annealing (AA) ' '
5 372.7 1.48 1.4
6 61.8 4.46 38
wafer - 2.07 23.5

 

 

 

 

 

92

 

software from HMI Berlin and also using the Motoﬁt reﬂectometry analysis program,
developed by Andrew Nelson at the Australian Nuclear Science and Technology

Organisation (ANSTO).

The models used to ﬁt the data in Fig. 1A and ID are listed in Table 5.2A & B.
In ﬁtting the data, the value of the roughness at the interface only serves the purpose
of smearing the scattering length density (SLD) proﬁle. Without any roughness at the
interface, the SLD proﬁle would be that for a sharp interface and would correspond to
a step ﬁinction instead of the gradual smearing. Of course, the roughness has no
physical meaning and one could obtain equivalent SLD proﬁles by assuming multiple
layers with appropriate SLDs.

5.1.3 Sample preparation for TEM:

The layered assembly of the polymer-nanoparticle sandwich ﬁlms were prepared
on a silicon oxide substrate. The silicon oxide layer was then dissolved in 48%
hydroﬂuoric acid. After this a thin layer of gold (~ 20-30 nm) was sputtered on the
surface of the ﬁlm. This was followed by the deposition of a thin layer of epoxy resin
(Polybed-812, Polysciences, Warrington, PA) on the gold surface. The entire sample
was then cured at 60°C for 24 hrs. The sample was then dropped in liquid nitrogen
and the epoxy layer together with the layered assembly was peeled from the silicon
substrate. This transfers the entire assembly as well as the gold layer to the epoxy
substrate. The sample was then ultramicrotomed (Power Tome XL, RMC, Tucson,

Arizona) and then imaged using a JEOL 100CX transmission electron microscope.

93

Chapter 6

Conclusions and Summary

In this work, we show that the addition of polystyrene nanoparticles to thin
polystyrene ﬁlms inhibits and in some cases eliminates dewetting as long as an
enriched layer of nanoparticles is present at the substrate surface. Since the
nanoparticles are compositionally very similar to the bulk linear polymer, it is
apparent that molecular architecture plays an important role in eliminating dewetting.
Addition of low molecular weight, linear polymer to longer, linear chains does not
eliminate dewetting but the dewetting rate is slower. It is clear that the phase
separated component’s dynamics along the substrate is important to inﬂuencing the

dewetting kinetics and in promoting wetting.

Higher molecular weight polystyrene nanoparticles strongly enhance the wetting
behavior of polystyrene thin ﬁlms, moreover for all polystyrene nanoparticle sizes a
substrate coverage of a monolayer is essential to provide the strongest inhibition of
dewetting. Neutron reﬂectivity data demonstrate that the nanoparticle substrate
coverage can be changed by varying the thickness of the ﬁlm and/or the bulk volume

fraction of nanoparticles, providing two avenues for control of dewetting.

Although CdSe nanoparticles preferentially segregate to the air surface of
polystyrene thin ﬁlms, these particles strongly inhibit dewetting of these ﬁlms.
Moreover both PS nanoparticles and CdSe quantum dots also stabilize wetting of

PMMA on OTS substrates.

94

Finally, we have been able to create layered assembly of polymer-nanoparticle
sandwich ﬁlms (SAMON process, Chapter 5) by utilizing the entropic and enthalpic
expulsion of nanoparticles to the interfaces in thin ﬁlms. This is achieved by
successive Spincoating and aging of the polymer-nanoparticle composite ﬁlm. In case
of the polystyrene-polystyrene nanoparticle system, an assembly of 6 layers was
demonstrated by neutron reﬂectivity measurements. Transmission Electron
Microscopy was used to observe the layered assembly of PS-Quantum Dots
composite ﬁlms. We were able to exhibit an unparalleled control on the ﬁnal structure
of the composite ﬁlm by being able to accurately control both the thickness of the

organic polymer layer as well as that of the inorganic nanostructures.

95

Appendix A: Dispersion Forces

Dispersion forces are due to induced dipole- induced dipole interactions which
depend on the dielectric constant of the intervening medium. In addition to the
dispersion forces, there are also short distance repulsive forces due to steric effects.
For a layered system, these short range repulsion terms are often approximated by the
form,

C
Vr (h) =h—8 (A.l)

which comes from integrating the l/rl2 repulsive term appearing in the Lennard-Jones
interaction. Here Vr is the potential due to the short range forces, h is the ﬁlm

thickness and C is a constant.

For a thin layer of thickness h (component 3), wedged between two very thick
layers (components 1 & 2), the potential due to dispersion forces is approximately
given by,(3)

_ A132

V1 W {5.17

(A2)

where A132 is the Hamaker constant and depends on the dielectric properties of the
three layers involved as explained below. The expression is true in the regime where
the interaction is non-retarded. Retardation effects are prevalent when h> 20, where 20
is the typical wavelength of light dominating the dispersion forces. Nevertheless, this
expression is a reasonable approximation to the dispersion forces dominating any tri-
layer system as the Hamaker constant A, contains information about the relevant

material properties.

96

Calculation of A is difﬁcult and based on Lifshitz theory(3) an approximate

expression for the Hamaker constant for a tri-layer system is given by:

Al32=3kT 81—83 82 _83 +2” £1(1.0")_£3(ivﬂ)£2(ivn)_£3(iun)

4 51+53 62 +83 47r U,£l(ivn)+£3(iun) 82(ivn)+£3(ivn)

 

 

dv (A.3)
Here an is the static dielectric constant of the layer n and en(iv) are the values of an at
imaginary frequencies. The ﬁrst term in the above expression is the classical van der
Waals term and constitutes the zero-frequency contribution to the Hamaker constant.
The second term is the London dispersion term and for most non-polar polymers
contributes of order 90 percent to A. From equation (3), it is clear that the sign of A
could be either positive or negative and depends strongly on the dielectric properties
of the three media involved. For example, for the case of an air gap (or vacuum)
between two thick slabs, the dielectric constants have the properties 81>e3 and c2>es
for all frequencies; so A is positive and the potential is always attractive.

The above expression could be further simpliﬁed, if the dielectric properties

of the medium are approximated by a Drude-like behavior(3) which takes into

account characteristic rotational and electronic relaxation processes,

 

g—n2 112—]
+
U

a(iv)= 1+ 2 (A.4)
UFO! U

c

where 5= 3 (0), and n is the refractive index which is assumed to be constant in the
IR-UV regime. The typical frequencies are vrot < 1012 Hz; v, z 3 XIO'S Hz (UV).(3)
Using this approximation, the integral in equation in (3) can be carried out to yield a

rather simple expression for the Hamaker constant,

97

3 81
A132: —kT 1/2
4 8, +83 82 +83 8(2 ) (S1S2(S1 +sz))

2 2 2 2
‘53 52 '33 3hue (n1 -n3 )(n2 "n3)

 

(A.5)

1/2
)

where s, = (n,2 + n,2 . The relative importance of the van der Waals as compared to

London terms contributing to A is roughly hot /(8" 2 )kT so that for a dominant UV

electronic process and room temperature, this ratio is approximately 1ev/3(0.025ev)z

10 indicating that the London term is approximately 10 times more important. This
effect may be mitigated by tuning the ratios aI/nl and 5.3/n2 , so there is

considerable scope for controlling the interactions using these variables. Dispersion
forces are electro-magnetic in origin so that changing the dielectric properties of the
ﬁlm and/or the nanoparticles in the ﬁlm can have a marked effect on the sign of
Hamaker constant which may lead to either attraction (positive A, dewetting) or

repulsion (negative A, wetting).

98

Appendix B: Neutron Reﬂectivity — Data Sensitivity Analysis

Neutron reﬂection measures the reﬂectivity R which is deﬁned as the ratio of
the intensity of the neutrons reﬂected from a surface to that of the incident intensity at
low angles. The specular reﬂectivity from a sample is measured as a function of

momentum transfer perpendicular to the surface of (the sample, Qz=41tSin0/}t. By

either varying the incident angle 0 or the wavelength of the neutrons X, the

reﬂectivity as a function of Q2 can be determined.

Neutron reﬂection is a versatile tool for investigating polymer blend ﬁlms, as
one can achieve great contrast between the different components by making
appropriate isotopic substitutions of hydrogen with deuterium and vice versa.
However, the information gained about the scattering density proﬁle of a sample
through neutron reﬂection measurements is sometimes difﬁcult to translate into a
concentration proﬁle. This appendix deals with the sensitivity analysis of some of the

parameters used to ﬁt typical neutron reﬂection data.

The neutron reﬂectivity measurements were performed at POSY2 Neutron
Reﬂectometer (resolution in Q-space, AQ/Q = 0.05) at Argonne National Laboratory,
on polymer blend ﬁlms that were previously aged for 24 hrs at 160°C under vacuum.
The reduced neutron reﬂection data was analyzed using Paratt 32 software from HMI
Berlin. The Paratt 32 software is a program to calculate the reﬂectivity of neutrons
from ﬂat surfaces. The calculation is based on Paratt’s recursion scheme for stratiﬁed
media. The goodness of a ﬁt to a set of neutron reﬂection data is shown by the value
of the parameter )8. The minimization of this parameter is done by a least squares

ﬁtting as given below:

99

Rcalc meas

2 M -—Q
I=Z

,-=1 weighting (3")

 

where weighting is either 1 (no weighting) or R5?” (statistical weighting) or 8R5“
(error weighting); M is the number of data points.

Figure 3.1 shows the neutron reﬂectivity data plotted as RQ4 vs. Q for a
polymer blend ﬁlm of dPS 63 kDa containing 20% by volume of NP 78 kDa after
annealing the ﬁlm for 24 hrs at 160°C under vacuum. The ﬁlm thickness as measured

by ellipsometry is approximately 45 nm. The ﬁts for the data shown in the ﬁgure

correspond to three two layer models as listed in Table 3.1.

 

 

 

 

7M.IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIITII_
egg 0 Data ‘
5 3 ---Modell ‘
3 E —Mode12 f
3 : --- Model3 _
. I
’. I
A 2 o ‘2
‘1 ,‘.
'5 ' :3
v '0‘ O.
3 -8 o '1 5“". _
a -_ 0‘ 5\'; w. ‘ ‘ -
" s
7L 3. o v.- " o o,“" " ” "
e- ~ 1 - - .X «we.
I . 9 ‘0. .‘%¢ 'u’ro‘\ 119‘ 0
5'- '. '. 3. .(‘~, “ ‘ ’ \
4'- t .. 5" (f. 0'
.0 '0
co 5-
.o
3. o.
r o
2 o
rfrﬁlrigrflﬁrnrrlr“1111111111111rrrlrrgrlru_
0.02 0.04 0.06 0.08
QiA“)

Figure B.l: RQ4 vs. Q for the data and ﬁts as explained in the text.

100

Table B.l. Parameters used in Parratt32 soﬁware (from HMI Berlin) to ﬁt the data in
Figure 8.1 for three different models. An air interface is assumed at the top of the
ﬁrst layer. The system consisted of 20 wt% 78 kDa protonated polystyrene
nanoparticles blended with 63 kDa deuterated linear polystyrene and was annealed at
160 °C for 24 hrs under vacuum.

System Thickness (A) SLD (IOWA-1) Roughness (A)

392 5.82 1
Model 1 62 2.86 23

2.07 10

6.02
Model 2 1.60

2.07

6.02
Model 3 3.07

2.07

 

The differences in the three models are the value of the thickness for each
layer, the interfacial roughness and relative scattering length densities for the layers.
The SLD for pure dPS and pure nanoparticle is 6.42><10'6 and 1.41 ><10’6 respectively.
Using a mass balance, one can estimate the fractional coverage of the nanoparticles at
the substrate aﬁer annealing which for the above blend is ca. 1.5. Of course, mass

balance also ﬁxes the values of SLD and the thickness for each layer.

Figure 8.2 shows the variation of the parameter x2 as a function of the
polymer-nanoparticle interfacial roughness. It’s interesting to note that for all the
three models the value of x2 goes through a minimum at a ﬁxed value of interfacial
roughness. The minimums for these x2 values is the lowest for model 2 which
corresponds to a thickness of 39 nm and 6.2 nm for the top and bottom layer

respectively. It’s also important to mention that any other combination of SLD and

101

thickness for the two layers produced 12 values, an order of magnitude higher than
that those shown in Fig. 8.2 and also didn’t show any signiﬁcant dependence on the

values of the interfacial roughness.

 

 

 

 

 

800 :nrrlntllnﬁllrnllrlrlnrllrlr1|HHIHHIHIT:
E. A Modell E
700:; C1 Mode12 .2
i O Mode13 E
600%- ‘2
5003 5
~.. E a
400:— E
300;— ‘2
200;— —E
100 511144111111111111'1llllllllllllLLlllllllllLLllllE
0 20 40 60 80 100
Roughness(A)

Figure B.2: x2 as a function of interfacial roughness for three two layer models at
constant layer thicknesses and SLDs. The solid lines in the ﬁgure correspond to
polynomial ﬁts (see Table 3.1).

For model 2 the minimum in x2 corresponds to an interfacial roughness of about 4 nm

which is reasonable considering the thickness of the nanoparticle layer to be 6.2 nm.
Now, one can also estimate the volume fraction of the nanoparticles in the two

layers from their SLDs. For example, for model 2 this corresponds to a volume

fraction of 0.08 and 0.96 in the top and the bottom layer respectively. Converting this

102

volume fraction in the bottom layer into an aerial fraction gives a value of 1.4 which

is in good agreement with that estimated above.

In ﬁtting the data above, the value of the roughness at the interface only
serves the purpose of smearing the scattering length density proﬁle for the sample.
For example, for the ﬁts above the scattering density proﬁle as a function of
normalized ﬁlm thickness is plotted in Figure 33. Without any roughness at the

interface, the SLD proﬁle would be

 

 

.5
7X10 Jrrrlnrqmﬁinllnn[nulnnnntllirurinlllnnitiltnnntn
I I
L'— - - Model] I
I- .1
: — Mode12 :
6 : --- Mode13€
1. - - - -0-.-. a-
b E .1
“a :' 7-
g h- Cl:
I Z
5 5 .T' .2
on ' .4
t: Z 2
0 _. .—
._1 _ ..
no ‘ Z
.5 C ‘
g; 4 .— —_
1'3 - _
a - -
o _ -
m r .:
3 :- j
- d
= 5
I :

 

 

 

2
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Normalized Thickness

Figure B.3: Scattering length density proﬁle as a function of normalized ﬁlm
thickness.

that for a sharp interface and would correspond to a step function instead of the

gradual smearing seen in Fig. 8.3. Of course, the roughness has no physical meaning

103

and one could obtain equivalent SLD proﬁles by assuming multiple layers with

appropriate SLDs.

The SLD proﬁle from the ﬁts can be converted into a concentration proﬁle as
shown in Fig. 3.4. The proﬁle shown in Fig. B.4 also takes into account the effects
related to the scattering length density of the substrate. This is also evident from the
maximum in the concentration proﬁle for the nanoparticles at the substrate which
then tapers off to a constant value. However, the concentration proﬁle of the
nanoparticles when plotted only as a function of the ﬁlm thickness (normalized
thickness = 1.0) (without any substrate effects) would lead to a monotonic increase in
the concentration. It is, however, important to understand that this monotonic increase
in the concentration proﬁle is only an artifact associated with representing the data. If
one takes into account the substrate effects then the concentration will hit a maximum

and then taper off to a constant value as explained above.

104

 

1-() _IITl]IiIIlIIII]IiililiililllllillllillillllliillliilliiiliilliiiiiilA

- - Model I
— Mode12

0'8 -- - Model 3

0.6

0.4

Nanoparticle Volume Fraction

0.2

.llllllllllllTlTllllllll'mlllll111111111

1:11:iiiilliliiiiliiiilll1111L141111111111111

 

 

r
l

 

 

p—
() () 11111111iiiill11111111111111111111155111111111111111111111111111111T

'00 0.2 0.4 0.6 0.8 1.0 12 1.4

Normalized Thickness

 

Figure B.4: Concentration proﬁle of the nanoparticles in the ﬁlm after annealing.

105

10.

ll.

12.

13.

14.

References and Notes

C. Redon, F. Brochardwyart, F. Rondelez, Phys. Rev. Lett. 66, 715 (Feb 11,
1991)

o. Reiter, Phys. Rev. Lett. 68, 75 (Jan 6, 1992).

J. N. Israelachvili, Intermolecular and Surface Forces (Academic Press,
London, 1992), pp.

A. Sharma, G. Reiter, J. Colloid Interface Sci. 178, 383 (Mar 25, 1996).

I. M. Tidswell, T. A. Rabedeau, P. S. Pershan, S. D. Kosowsky, Phys. Rev.
Lett. 66, 2108 (Apr 22, 1991).

G. Reiter et al., Langmuir 15, 2551 (Mar 30, 1999).
A. Sharma, R. Khanna, J.Chem.Phys. 110, 4929 (Mar 8, 1999).

R. Seemann, S. Herminghaus, K. Jacobs, J. Phys. Cond. Matt. 13, 4925 (May
28,2001)

R. Seemann, S. Herminghaus, K. Jacobs, Phys. Rev. Lett. 86, 5534 (Jun 11,
2001).

R. Seemann et al., J. Phys. Cond. Matt. 17, S267 (Mar 9, 2005).

H. P. Zhao, Y. J. Wang, 0. K. C. Tsui, Langmuir 21, 5817 (Jun 21, 2005).
Vrij, Discuss. Faraday. Soc. 42, 23 (1966).

G. Reiter, Langmuir 9, 1344 (May, 1993).

T. G. Stange, D. F. Evans, W. A. Hendrickson, Langmuir 13, 4459 (Aug 6,
1997).

106

15.

16.

17.

l8.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

K. Jacobs, S. Herminghaus, K. R. Mecke, Langmuir 14, 965 (Feb 17, 1998).

J. A. Orlicki, J. S. Moore, 1. Sendijarevic, A. J. McHugh, Langmuir 18, 9985
(Dec 10, 2002).

O. Wunnicke, C. Lorenz-Haas, P. Muller-Buschbaum, V. Leiner, M. Stamm,
Appl. Phys. Mat. Sci. Proc. 74, S445 (Dec, 2002).

C. G. Yuan, O. Y. Meng, J. T. Koberstein, Macromolecules 32, 2329 (Apr 6,
1999)

Y. Feng, A. Karim, R. A. Weiss, J. F. Douglas, C. C. Han, Macromolecules
31, 484 (Jan 27, 1998).

T. Kerle, R. YerushalmiRozen, J. Klein, Europhys. Lett. 38, 207 (Apr 20,
1997).

G. Henn, D. G. Bucknall, M. Stamm, P. Vanhoome, R. Jerome,
Macromolecules 29, 4305 (Jun 3, 1996).

R. Yerushalmirozen, J. Klein, Langmuir 11, 2806 (Jul, 1995).

R. Yerushalmirozen, J. Klein, L. J. Fetters, Science 263, 793 (Feb 11, 1994).

J. W. Grate, R. A. McGill, Anal. Chem. 67, 4015 (Nov 1, 1995).

K. A. Barnes et al., Macromolecules 33, 4177 (May 30, 2000).

S. Sharma, M. H. Rafailovich, D. Peiffer, J. Sokolov, Nano Lett. 1, 511 (Oct,
2001)

M. E. Mackay et al., Langmuir 18, 1877 (Mar 5, 2002).

C. J. Hawker, J. M. J. Frechet, J.Am.Chem.Soc. 112, 7638 (Oct 10, 1990).

107

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

C. J. Hawker, E. E. Malmstrom, C. W. Frank, J. P. Kampf, J. Am. Chem. Soc.
119, 9903 (1997).

M. E. Mackay et al., Nat. Mater. 2, 762 (Nov, 2003).

A. Tuteja, M. E. Mackay, C. J. Hawker, B. Van Horn, Macromolecules 38,
8000 (Sep 20, 2005).

F. Brochardwyart, P. G. Degennes, H. Hervert, C. Redon, Langmuir 10, 1566
(May, 1994).

K. Jacobs, R. Seemann, G. Schatz, S. Herminghaus, Langmuir 14, 4961 (Sep
1,1998)

G. Reiter, P. G. de Gennes, Euro. Phys. J. E 6, 25 (Sep, 2001).

O. Wunnicke et al., Langmuir 19, 8511 (Sep 30, 2003).

C. J. Hawker, E. E. Malmstrom, C. W. Frank, J. P. Kampf, J.Am.Chem.Soc.
119, 9903 (Oct 15, 1997).

E. Harth et al., J.Am.Chem.Soc. 124, 8653 (Jul 24, 2002).

It was essential to remove these silicon containing compounds as these have

higher surface energy and might interfere with the dewetting experiments. For
example, surface tension measurements of some of the samples using the
micro- Wilhelmyﬁber technique showed variable results for samples with and

39.

40.

41.

without the silicon containing compounds.

G. T. Dee, B. B. Sauer, J. Colloid Interface Sci. 152, 85 (Aug, 1992).

P. Muller-Buschbaum, Euro. Phys. J. E 12, 443 (Nov, 2003).

It was not possible to spin coat the benzene solution directly onto the silanized
surface as the contact angle of the solution with the surface was extremely
high and the solution just rolled oﬂ when the spincoater was switched on.
Direct Spincoating of the solution resulted in highly non-uniformﬁlms.

108

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

F. M. F owkes, D. C. McCarthy, M. A. Mostafa, J. Colloid Interface Sci. 78,
200 (1980).

B. B. Sauer, N. V. Dipaolo, J. Colloid Interface Sci. 144, 527 (Jul, 1991).

Acid cleaning of the silicon wafers resulted in highly hydrophilic substrates.
Air borne hydrocarbons deposit on these high energy clean surfaces creating
a thin layer of contamination. The presence of this layer is very important for
dewetting. The uncontrollable thickness and the non-uniform nature of this
contamination layer resulted in highly inconsistent wetting/dewetting
behavior of the polymer. To ensure the dewetting of the pure polymer and to
be able to reproduce the same surface again, it was essential to silanize our
substrates. This resulted in a substrate of very low and uniform surface
energy.

H. K. Shigeharu Onogi, Shiro Ueki, Toshio Ibaragi, J. Poly. Sci. Part C 15,
481(1966)

K. K. Toshiro Masuda, Toshio Inoue, and Shigeharu Onogi, Macromolecules
3,116(1969).

P. G. de Gennes, C. R. Acad. Sci. Paris Serie B, 219 (1979).

C. Redon, J. B. Brzoska, F. Brochardwyart, Macromolecules 27, 468 (Jan 17,
1994)

A. Hariharan, S. K. Kumar, T. P. Russell, Macromolecules 23, 3584 (Jul 23,
1990)

T. F. Schaub et al., Macromolecules 29, 3982 (May 20, 1996).
J. P. Cotton et al., Macromolecules 7, 863 (1974).
G. Reiter, P. Auroy, L. Auvray, Macromolecules 29, 2150 (Mar 11, 1996).

C. Gay, Macromolecules 30, 5939 (Sep 22, 1997).

109

54.

55.

56.

57.

58.

59.

60.

61.

62.

63.

64.

65.

66.

67.

68.

F. BrochardWyart, C. Gay, P. G. deGennes, Macromolecules 29, 377 (Jan 1,
1996)

E. Durliat, H. Hervet, L. Leger, Europhys. Lett. 38, 383 (May 10, 1997).
G. Reiter, R. Khanna, Phys. Rev. Lett. 85, 2753 (Sep 25, 2000).
Y. Liu et al., Phys. Rev. Lett. 73, 440 (Jul 18, 1994).

R. A. L. Jones, E. J. Kramer, M. H. Rafailovich, J. Sokolov, S. A. Schwarz,
Phys. Rev. Lett. 62, 280 (Jan 16, 1989).

S. M. M. Ramos, E. Charlaix, A. Benyagoub, M. Toulemonde, Phys. Rev. E.
67 (Mar, 2003).

H. Luo, D. Gersappe, Macromolecules 37, 5792 (2004).

K. A. Barnes, J. F. Douglas, D. W. Liu, A. Karim, Adv. Colloid Interface Sci.
94, 83 (Nov 15, 2001).

S. Sharma, M. H. Rafailovich, D. Peiffer, J. Sokolov, Nano Letters 1, 511
(Oct, 2001).

R. S. Krishnan, M. E. Mackay, C. J. Hawker, B. Van Horn, Langmuir 21,
5770 (Jun 21, 2005).

S. Asokan let al., Nanotechnology 16, 2000 (Oct, 2005).
A. Tuteja, M. E. Mackay, C. J. Hawker, B. Van Horn, in press (2006).
J. B. Brzoska, I. Benazouz, F. Rondelez, Langmuir 10, 4367 (Nov, 1994).

A. Hariharan, S. K. Kumar, T. P. Russell, J.Chem.Phys. 98, 4163 (Mar 1,
1993)

M. E. Mackay et al., Science 311, 1740 (2006).

110

 

69.

70.

71.

72.

73.

74.

75.

76.

77.

78.

79.

80.

F. S. Bates, G. D. Wignall, Phys. Rev. Lett. 57, 1429 (Sep 22, 1986).

V. A. Parsegian, van der Waals Forces (Cambridge University Press, 2005),
PP-

In reference 8, A 3m}, is calculated from the refractive index of the silane taken
from ref 29 there in. However, determination of the thickness and the
reﬁactive index of silane simultaneously is very difficult. For eg. use of 1.42
as the reﬁactive index of silane layer leads to a Hamaker constant
(A ir/PS/Silane) of1.9e-20 as reported in ref 8. But this value of A 3m}. does not
explain the observed faster rates of dewetting for the pure polymer on the
silanized substrate compared to that of bare SiOz substrates. The use of eq.
3.5 and 3.6 is justified to calculate the Hamaker constant for a non-polar
material like an alkane brush.

M. D. Millan, J. Locklin, T. Fulghum, A. Baba, R. C. Advincula, Polymer 46,
5556 (Jul 11, 2005).

J. Brandrup, E. H. Immergut, Polymer Handbook 3rd edn (John Wiley &
Sons, New York) (1998).

R. S. Krishnan et al., Submitted (2006).

J. Y. Lee, G. A. Buxton, A. C. Balazs, J.Chem.Phys. 121, 5531 (Sep 15,
2004).

S. Tyagi, J. Y. Lee, G. A. Buxton, A. C. Balazs, Macromolecules 37, 9160
(Nov 30, 2004).

M.Ulbricht, Polymer 47, 2217 (2006).

N. Bagkar et al., Thin Solid Films 497, 259 (Feb 21, 2006).

C. H. Huang, N. D. McClenaghan, A. Kuhn, G. Bravic, D. M. Bassani,
Tetrahedron 62, 2050 (Feb 27, 2006).

Y. Lin et al., Nature 434, 55 (Mar 3, 2005).

111

 

81.

82.

83.

84.

85.

86.

87.

88.

89.

90.

91.

92.

93.

94.

Y. Lin, H. Skaff, T. Emrick, A. D. Dinsmore, T. P. Russell, Science 299, 226
(Jan 10, 2003).

J. M. Bertolo, A. Bearzotti, P. Falcaro, E. Traversa, P. Innocenzi, Sensor Lett.
1, 64 (Dec, 2003).

X. Pages, V. Rouessac, D. Cot, G. Nabias, J. Durand, Sepa. Puri. T echn. 25,
399 (Oct 1, 2001).

K. B. Blodgett, J.Am.Chem.Soc. 57, 1007 (Jan, 1935).
G. Decher, Science 277, 1232 (Aug 29, 1997).
Y. Lvov, G. Decher, H. Mohwald, Langmuir 9, 481 (Feb, 1993).

Z. Y. Tang, N. A. Kotov, S. Magonov, B. Ozturk, Nat. Mater. 2, 413 (Jun,
2003)

T. P. Bigioni et al., Nat. Mater. 5, 265 (Apr, 2006).

K. Stratford, R. Adhikari, I. Pagonabarraga, J. C. Desplat, M. E. Cates,
Science 309, 2198 (Sep 30, 2005).

J. X. Huang, F. Kim, A. R. Tao, S. Connor, P. D. Yang, Nat. Mater. 4, 896
(Dec, 2005).

M. E. Mackay et al., Science 311, 1740 (Mar 24, 2006).
A. Yethiraj, Phys. Rev. Lett. 74, 2018 (Mar 13, 1995).

S. Gupta, Q. L. Zhang, T. Emrick, A. C. Balazs, T. P. Russell, Nat. Mater. 5,
229 (Mar, 2006).

Z. B. Guan, P. M. Cotts, E. F. McCord, S. J. McLain, Science 283, 2059 (Mar
26,1999)

112

 

95. T. G. Fox, P. J. Flory, J.Am.Chem.Soc. 70, 2384 (1948).

96. J. H. Xavier et al., Macromolecules 39, 2972 (Apr 18, 2006).

113

List of Publications

Krishnan, R. S.; Mackay, M. E.; Hawker, C. J .; Van Horn, B. “Influence
of Molecular Architecture on the Dewetting of Polystyrene Films”
Langmuir 2005, 21, (13), 5770-5776.

Mackay, M. E.; Tuteja, A.; Duxbury, P. M.; Hawker, C. J .; Van Horn, 3.;
Guan, Z. B.; Chen, G. H.; Krishnan, R. S. “General Strategies for
Nanoparticle Dispersion” Science 2006, 311, 1740-1743.

Krishnan, R. S.; Mackay, M. E.; Hawker, C. J .; Van Horn, 3.; Duxbury, P.
M.; Wong, M. S.; Asokan, S.; Goyette, R.; Thiyagarajan, P. “Nanoparticle
Induced Wetting of Polymer Thin Films” Manuscript for Macromolecules
2006

Krishnan, R. S.; Mackay, M. E.; Pastor, A.; Hawker, C. J .; Van Horn, 8.;
Wong, M. Asokan, S. “Self-Assembled Multilayers of Nanocomponents”
Submitted to Science. 2006

Krishnan, R. S.; Mackay, M. E.; Hawker, C. J .; "Entropic phase separation
of organic nanoparticles," Polymer Preprints, 46 (2005) 547-548.

114

 

 

111311111211111)le11111111