www.sciencemag.

org/cgi/content/full/321/5891/939/DC1

Supporting Online Material for

Graphoepitaxy of Self-Assembled Block Copolymers on
Two-Dimensional Periodic Patterned Templates
Ion Bita, Joel K. W. Yang, Yeon Sik Jung, Caroline A. Ross,*
Edwin L. Thomas,* Karl K. Berggren*
*To whom correspondence should be addressed. E-mail: caross@mit.edu (C.A.R.),
elt@mit.edu (E.L.T.), berggren@mit.edu (K.K.B.)
Published 15 August 2008, Science 321, 939 (2008)
DOI: 10.1126/science.1159352

This PDF file includes:

Materials and Methods
SOM Text
Figs. S1 and S2
Tables S1 and S2
References

Supplemental information
Graphoepitaxy of Self-assembled Block Copolymers on Two-Dimensional Periodic
Patterned Templates
Ion Bita1,3, Joel K.W. Yang2,3, Yeon Sik Jung1,3,
Caroline A. Ross1, Edwin L. Thomas1, Karl K. Berggren2
1

Department of Materials Science and Engineering, Massachusetts Institute of Technology,

77 Massachusetts Avenue, Cambridge, MA 02139, USA

Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology,

77 Massachusetts Avenue, Cambridge, MA 02139, USA
3

These authors contributed equally to the reported work.

Section S1. Experimental Methods

The templates were fabricated using electron beam patterning of hydrogen
silsesquioxane (HSQ), a negative-tone electron resist. HSQ films were spun coated on
silicon substrates, and single-pixel dots were exposed in a Raith 150 electron-beam
lithography tool at 30 kV acceleration voltage. The samples were developed in a highcontrast developer system as described previously [S1], and further treated with O2/He
plasma (50W, 2min) to remove possible organic residues.
The topographically patterned substrates were chemically functionalized with
either PS or PDMS brushes by spin coating a 30nm thick layer of hydroxyl-terminated
homopolymer (10 kg/mol PS-OH, or 5kg/mol PDMS-OH), annealing at 170C for 15
hours, and removing unreacted homopolymer in toluene. 50 nm thick thin PS-b-PDMS
(51.5kg/mol, volume fraction fPDMS=16.5%, PDI=1.04) films were spin-coated from a
toluene solution (2.5% by weight of the PS-PDMS block copolymer) on the
functionalized substrates, then the samples were vacuum annealed at 200C for 12 hours.

S1

The annealed film was treated with a 5s, 50 W CF4 plasma to remove the PDMS surface
layer, and then with a 30s, 90W O2 plasma to selectively remove the PS domains and
form oxidized PDMS dots on the substrate. The resulting surface morphology was
observed using a Raith 150 SEM operated with an acceleration voltage of 10kV. A thin
layer of Au-Pd alloy was sputter-coated on samples in order to avoid charging effects.

Section S2. Analysis of the BCP sublattice orientations templated by a post lattice
We consider a template consisting of cylindrical posts arranged on a hexagonal
lattice with period Lpost and derive the possible configurations where the post superlattice
is commensurate with a 2D hexagonal lattice of spatial period L < Lpost corresponding to
the BCP spherical microdomain array.

r
In general, commensurate configurations can be obtained by using any vector rij
in the BCP lattice as one of the basis vectors of the post lattice: (Fig. S1)
r
L post b1 = rij

(S-1)

where b1 is one of the basis vectors of the post lattice (together with a second vector
! generated by a 60 rotation of b1 ), and
r
rij = L (ia1 + ja 2 )

(S-2)

where i and j are integers, and a1 and a 2 are the unit vectors of the BCP lattice (defined

at 60 from each other).

Fig. S1. A post lattice vector (red dashed
Lpost b1

arrow) is equal to the sum of BCP vectors

L a 2

3L a1

60

(blue solid arrows). This illustrates the <3 1>

lattice.

The post lattices that correspond to each unique pair i and j are defined by
r
L post = rij = L ia1 + ja 2 = L i 2 + j 2 + ij .
(S-3)

S2

We use the notation ij to index the possible BCP lattice configurations with respect to
the hexagonal post lattice.
The orientation of the post lattice with respect to the BCP lattice can be described
r
by using the angle formed between rij and a1 :
&
#
2i + j
$
!
' = arccos
$
!
2
2
% 2 i + j + ij "

(S-4)

The table below lists all the possible orientation angles for the case when i, j 5.
A map of the lattice orientations is shown in Fig. 2A of the main text for Lpost/L < 5. For
each angle, a family of lattices exists corresponding to multiples of i and j, e.g. at = 0
the family is 10, 20, 30....

Lattice

Lpost/L

[]

10

1 = 1.00

0.0

11

3 = 1.73

30.0

21

7 = 2.65

19.1

31

13 = 3.61

13.9

32

19 = 4.36

23.4

41

21 = 4.58

10.9

51

31 = 5.57

9.0

43

37 = 6.08

25.3

52

39 = 6.24

16.1

53

49 = 7.00

21.8

54

61 = 7.81

26.3

TABLE S1: ij commensurate post lattice spacing and orientations corresponding to the 11

unique rij directions in a 2D hexagonal BCP lattice with i, j 5 and spacing L.

We may now consider how many BCP domains are templated by a given post for
each of the BCP lattices ij. The table below shows for some of the lattices the number n
of domains templated by each post and the number m of unit cells of the BCP lattice per
unit cell of the post lattice. Note that m = n+1 = i2 + j2 +ij.

S3

Lattice

10
20
30
40
50
11
22
33
21

n = number of
BCP domains
templated by
each lattice
post
0
3
8
15
24
2
11
26
6

m =number
of BCP unit
cells per unit
cell of the
post lattice
1
4
9
16
25
3
12
27
7

i2 + j2 +ij
1
4
9
16
25
3
12
27
7

TABLE S2: The number of BCP domains templated by each post of the lattice, and the number of
BCP unit cells per lattice unit cell, for several BCP lattices ij.

Section S3. Free energy model for BCP sublattice formation onto a post lattice
We present a simple free energy model developed to describe a BCP lattice that is
allowed to undergo in-plane strain in order to achieve commensuration on a given
periodic template. The previous section and Fig. 2A described the case of perfect
commensuration, where for most Lpost/L values there is only one possible orientation
between the BCP lattice and the post lattice. However, the BCP lattice may be able to
adopt multiple orientations for certain Lpost values by slightly expanding or compressing
itself so that commensuration is achieved at particular values of the Lpost/L ratio, where L,
the strained BCP lattice spacing, differs from the equilibrium spacing, Lo, adopted in the
absence of the template.
For a given <ij> configuration, the templated BCP lattice spacing is:

L=

L post

(S-5)

i + j 2 + ij

In order to derive a free energy model, F, we assume an affine deformation

where the relative deformation of an individual BCP chain is the same as the macroscopic
strain, , of the BCP lattice. For hexagonal symmetry in 2D, the elastic modulus is
isotropic, justifying the use of a 1D strain model. The free energy change for an A-B
diblock copolymer chain can be defined as:
!Fchain = !H " T!S = # AB A " T!Sconf

S4

(S-6)

where AB = interfacial tension between the A and B blocks, A = interface area per chain,
T = temperature, and Sconf = change in conformational entropy per chain.
The change in conformational entropy for a strained ideal polymer network is:
k
"Sconf = ! (#2x + #2y + #2z ! 3)
2

(S-7)

where k = Boltzmann constant, and i are the strains in the i = x, y, z directions. Further
assuming an incompressible system (i=1) and that x = , the entropy term becomes:
k
2
"Sconf = ! (#2 + ! 3)
2
#

(S-8)

The strain is calculated with respect to a Gaussian coil reference state:

!=

L/2

(S-9)

b M

where b = statistical segment (Kuhn) length, and M = number of statistical steps.

The enthalpic contribution to Fchain is related to the change in interfacial energy
upon deformation of the polymer chain. The interfacial tension between the A and B
blocks of an A-B diblock copolymer can be estimated from the Helfand-Tegami
expression for the interfacial energy between A and B homopolymers [S2]:
kT
" AB = 2
b

! AB
6

(S-10)

where AB is the Flory-Huggins interaction parameter between the A and B blocks of the
diblock copolymer.
The interface area for the microphase separated chain can be estimated as:

V
2 Mb3
A ! chain =
L/2
L

(S-11)

where half the BCP spacing is chosen to correspond to the length of one chain.
Thus, by using Eq. S8-S11, we can derive an expression for the free energy:
(Fchain / kT =

#
2 Mb ) AB 1 &$ L2
4b M
!
+
+
'
3
!
L
6
2 $% 4 Mb2
L
"

(S-12)

In order to use this model to compare with the experimental results shown in
Figure 3A, we first minimize Eq. (S-11) with respect to L and determine an equilibrium
spacing, Lo, for this model using AB = 0.18 (at 200C, the sample anneal temperature),
S5

M=231, and b = 0.56nm (estimated as a weighted mean of the Kuhn steps for PS and
PDMS reported elsewhere [S3, S4]). For each experimental post spacing, we calculate
the templated BCP spacing (Eq. S-5), and then the free energy corresponding to that
spacing is calculated from Eq. (S-12). The free energy is plotted versus Lpost/Lo in Fig. 3A
for a direct comparison with the experimental results of Fig. 3B.

Section S4. Procedure for SEM image analysis and determination of BCP grain
orientations
A computer program written in Matlab was developed to analyze the electron
microscope images. The steps of the analysis process are described in Fig. S2.
(A)

(B)

200 nm

(C)

Fig. S2. (A) Scanning electron microscope image; (B) The center for each dot (oxidized
PDMS spherical microdomains and template posts) in the image is identified by
correlation of the image with a disk the size of a dot; (C) A Voronoi diagram is
generated, and the orientation of the BCP dot lattice with respect to the post lattice
calculated at each point. In this case, blue = +/-19, red = 30, and green = 0. Cells

S6

without color are either defects or have vertices outside the frame (e.g. near the edges of
the image).
An approximate defect density was estimated within a 1.5 by 1.5 micron area. In the
templated samples with the lowest defect levels, in which 2D commensuration was
achieved, the percentage of defects (5-fold or 7-fold coordinated spheres) was less than
2%. In these instances, the orientational and translational order of the BCP was preserved
over the entire templated area despite the small number of defects. The translational and
orientational order is preserved over large distances due to the pinning action of the
lithographically defined posts which themselves also had excellent long range order.

References
S1. J. K. W. Yang, K.K. Berggren, Journal of Vacuum Science & Technology B, 25, 2025
(2007)
S2. E. Helfand, Accounts of Chemical Research, 8, 295 (1975)
S3. R. Matsuno, H. Otsuka, A. Takahara, Soft Matter, 2, 415 (2006)
S4. T.J. Senden, J.M. di Meglio, P. Auroy, European Physical Journal B 3, 211 (1998)

S7