Synthesis and high-throughput characterization of structural analogues of molecular glassformers: 1,3,5-trisarylbenzenes

Article abstract of DOI:10.1039/c5sm01044f

We report the synthesis and characterization of an analogous series of small organic molecules derived from a well-known glass former, 1,3-bis(1-naphthyl)-5-(2-naphthyl)benzene (α,α,β-TNB). Synthesized molecules include α,α,β-TNB, 3,5-di(naphthalen-1-yl)-1-phenylbenzene (α,α-P), 9-(3,5-di(naphthalen-1-yl)phenyl)anthracene (α,α-A), 9,9′-(5-(naphthalen-2-yl)-1,3-phenylene)dianthracene (β-AA) and 3,3′,5,5′-tetra(naphthalen-1-yl)-1,1′-biphenyl (α,α,α,α-TNBP). The design of molecules was based on increasing molecular weight with varied π-π interactions in one or more substituents. The synthesis is based on Suzuki cross-coupling of 1-bromo-3-chloro-5-iodobenzene with arylboronic acids, which allows attachment of various substituents to tailor the chemical structure. The bulk compounds were characterized using NMR spectroscopy and differential scanning calorimetry (DSC). Thin films of these compounds were produced using physical vapor deposition and were subsequently annealed above the glass transition temperatures (T_g). For each molecular glass, cooling rate-dependent glass transition temperature measurements (CR-T_g) were performed using ellipsometry as a high-throughput method to characterize thin film properties. CR-T_g allows rapid characterization of glassy properties, such as T_g, apparent thermal expansion coefficients, apparent activation energy at T_g and fragility. DSC measurements confirmed the general trend that increasing molecular weight leads to increasing melting point (T_m) and T_g. Furthermore, CR-T_g provided evidence that the introduction of stronger π-interacting substituents in the chosen set of structural analogues increases fragility and decreases the ability to form glasses, such that β-AA has the largest fragility and highest tendency to crystallize among all the compounds. These strong interactions also significantly elevate T_g and promote more harmonic intermolecular potentials, as observed by decreasing value of the apparent thermal expansion coefficient.

Full text of DOI:10.1039/c5sm01044f

Soft Matter
View Article Online
View Journal
PAPER
Synthesis and high-throughput characterization of
structural analogues of molecular glassformers:
Cite this:DOI: 10.1039/c5sm01044f
1,3,5-trisarylbenzenes†
a
a
a
a
a
Tianyi Liu, Kevin Cheng, Elmira Salami-Ranjbaran, Feng Gao, Ethan C. Glor,
ab
a
a
Mu Li, Patrick J. Walsh* and Zahra Fakhraai*
We report the synthesis and characterization of an analogous series of small organic molecules derived
from a well-known glass former, 1,3-bis(1-naphthyl)-5-(2-naphthyl)benzene (a,a,b-TNB). Synthesized
molecules include a,a,b-TNB, 3,5-di(naphthalen-1-yl)-1-phenylbenzene (a,a-P), 9-(3,5-di(naphthalen-1-
0
0
0
yl)phenyl)anthracene (a,a-A), 9,9 -(5-(naphthalen-2-yl)-1,3-phenylene)dianthracene (b-AA) and 3,3 ,5,5 -
0
tetra(naphthalen-1-yl)-1,1 -biphenyl (a,a,a,a-TNBP). The design of molecules was based on increasing
molecular weight with varied p–p interactions in one or more substituents. The synthesis is based on Suzuki
cross-coupling of 1-bromo-3-chloro-5-iodobenzene with arylboronic acids, which allows attachment of
various substituents to tailor the chemical structure. The bulk compounds were characterized using NMR
spectroscopy and differential scanning calorimetry (DSC). Thin films of these compounds were produced
using physical vapor deposition and were subsequently annealed above the glass transition temperatures
(
T
g
). For each molecular glass, cooling rate-dependent glass transition temperature measurements (CR-T
were performed using ellipsometry as a high-throughput method to characterize thin film properties.
CR-T allows rapid characterization of glassy properties, such as T , apparent thermal expansion
coefficients, apparent activation energy at T and fragility. DSC measurements confirmed the general
trend that increasing molecular weight leads to increasing melting point (T ) and T . Furthermore,
CR-T provided evidence that the introduction of stronger p-interacting substituents in the chosen set
g
)
g
g
g
m
g
Received 1st May 2015,
g
Accepted 10th August 2015
of structural analogues increases fragility and decreases the ability to form glasses, such that b-AA has the
largest fragility and highest tendency to crystallize among all the compounds. These strong interactions also
DOI: 10.1039/c5sm01044f
g
significantly elevate T and promote more harmonic intermolecular potentials, as observed by decreasing
www.rsc.org/softmatter
value of the apparent thermal expansion coefficient.
1
4–16
crystallization in the glass phase (GC crystallization).
Many
1
Introduction
factors have been shown to affect the stability of a glass
1
4,17
1,18
19
Amorphous organic compounds have important applications in including nucleation time,
fragility,
the application temperature. Specifically in GC crystal growth,
and organic electronics. Despite these and many other potential diffusion coefficient and relaxation times in the amorphous
and aging rate at
1–3
4,5
pharmaceutics, resist materials for nanoimprint lithography,
6–8
2
0–22
uses, a major limitation to their wide-spread utilization is the low state are two influential parameters.
1,5,9
stability.
of organic materials are only controlled by weak intermolecular aging and GC crystal growth rate of organic glasses is to increase
interactions. The lack of strong intermolecular networks or chain their glass transition temperature, T , relative to the operating
connectivity makes it challenging to produce organic glasses temperature (often close to room temperature). Variations
with high kinetic and thermal stability. The instability of in molecular architectures, such as chemical compositions,
organic glasses may be caused by dewetting, aging,
Unlike inorganic materials or polymers, the properties
One approach to elevate thermal stability, and to suppress
g
2
3
2
4
1
0
11–13
25
and molecular weights and symmetries are all tunable parameters
to tailor T and other physical properties of glassy materials.
g
a
Department of Chemistry, University of Pennsylvania, 231 South 34th Street,
Philadelphia, Pennsylvania 19104-6323, USA. E-mail: fakhraai@sas.upenn.edu,
pwalsh@sas.upenn.edu
Few studies have systematically investigated the connection
between chemical structure of an organic molecule and its
ability to form ‘‘good glasses’’ with high thermodynamic, kinetic
b
Department of Material Science and Engineering, University of Pennsylvania,
2
0,26–28
stability and resistance to crystallization.
While increased
3
231 Walnut Street, Philadelphia, Pennsylvania, 19104-6272, USA
2
9,30
†
Electronic supplementary information (ESI) available. See DOI: 10.1039/c5sm01044f molecular-level interactions, such as hydrogen bonding
This journal is ©The Royal Society of Chemistry 2015
Soft Matter

View Article Online
Paper
Soft Matter
g g
and p–p interactions can potentially increase the T value of Prior studies have demonstrated the advantages of CR-T in
a glass-former, they may also enhance the nucleation and reliably estimating the apparent activation barrier close to T
crystallization rates. As such, the relationship between chemical and the apparent fragility of thin polymer films as a function of
g
1
3,41–46
structure and properties of organic glasses remains complicated film thickness.
Although CR-T may not be as precise as
g
and poorly understood. To better map the relationship between calorimetry, rheology or other bulk thermodynamic and
chemical structure and material’s properties would require mechanical measurement, it provides a unique opportunity to
systematic studies using high-throughput synthesis and charac- perform high-throughput measurements on the properties of
terization methods. Furthermore, since glasses are generally these tailored molecular glasses as a function of substituents.
used in nanoscale length scales, it is also important to study This method can also be applied as a preliminary screen of a
the influence of molecular level interactions on physical proper- wide range of glass-formers with minimal synthesized material.
3
1
ties, such as mobility at the free surface, confinement effects, Once molecules of interest have been identified, additional
3
2,33
34
surface diffusion
and the ability to form stable glasses.
in-depth studies on the physical properties can be performed
Herein, we chose trisnaphthylbenzene (TNB) as a frame of with more precise techniques.
reference to synthesize structural analogues for systematic studies
In this study, DSC measurements showed that both T
m
of glass properties. TNB, an organic molecule that only contains and T
g
increased monotonically with increasing molecular
conjugated p moieties, has been extensively studied and shown to weight for compounds containing phenyl and naphthyl sub-
3
4–37
be a good glass-former.
In this study, we introduce straightfor- stituents. Interestingly, the molecules containing anthracyl
and T as a
ward and high yielding methods to synthesize an analogous series substituents showed stronger increase in both T
m
g
of TNB-like molecules. These molecules include a,a,b-TNB and function of molecular weight. Thermal expansion coefficient
four new compounds in the family: 3,5-di(naphthalen-1-yl)-1- and CR-T measured by ellipsometry showed that increased p
g
phenylbenzene (a,a-P), 9-(3,5-di(naphthalen-1-yl)phenyl)anthracene interactions can decrease the anharmonicity of intermolecular
0
(
a,a-A), 9,9 -(5-(naphthalen-2-yl)-1,3-phenylene)dianthracene (b-AA) potential, as observed by decreasing value of the expansion
0
0
0
and 3,3 ,5,5 -tetra(naphthalen-1-yl)-1,1 -biphenyl (a,a,a,a-TNBP) coefficient, increase fragility and decrease the ability of the
Fig. 1). These molecules were designed based on increasing compound to form glasses. The compound with the strongest p
molecular weight and the size of conjugated p substituents. interactions, b-AA, has the highest fragility and crystallized
The synthesized compounds were characterized using NMR rapidly around T at low cooling rates. The critical cooling rate,
(
g
spectroscopy and differential scanning calorimetry (DSC). The the cooling rate below which a super-cooled equilibrium phase
compounds were then vapor deposited to produce thin films cannot be achieved, is often used to infer the tendency of
in an ultra-high vacuum chamber for high-throughput char- a compound to crystallize. The higher the rate of cooling
4
7
acterization using cooling rate-dependent glass transition tem- required to super-cool, the higher the tendency to crystallize.
perature (CR-T ) measurements by ellipsometry. Ellipsometry is The critical cooling rate of b-AA glass was measured to be
often used to measure film thickness as well as other material greater than 10 K min , while for other compounds, it is lower
g
ꢀ
1
3
8–40
ꢀ1
properties such as the index of refraction and birefringence.
than 1 K min . These studies highlight the important role of
Measurements of thickness as a function of temperature molecular level interactions in dictating the properties of
can be used to evaluate Tg and the expansion coefficients. amorphous materials.
2
Experimental details
2.1 Synthesis of a,a,b-TNB and its structural analogues
1
-Bromo-3-chloro-5-iodobenzene was synthesized from aniline
4
8
following the procedure outlined in Gilbert and Martin.
The material was prepared by students at the University of
Pennsylvania enrolled in the Introductory Organic Chemistry
Laboratory (Chemistry 245). The trihalobenzene was further
purified as detailed in the ESI.†
The syntheses are described in the flow-chart shown in Fig. 2.
The advantage of using trihalobenzene is that each of the three
halogens has different reactivity toward palladium catalysts.
Ligands for palladium catalyzed Suzuki reactions were chosen
to maximize chemoselectivity in reactions at the C–halogen
4
9
bonds. General procedures for each of the reactions in the
1
13
1
overview are outlined in the ESI.† H NMR and C{ H} NMR
spectra were obtained on a Br u¨ ker AM-500 Fourier transform
NMR spectrometer at 500 and 125 MHz, respectively (spectra and
details shown in ESI,† Fig. S1–S7).
Fig. 1 Five molecules used in this study. Color-coding indicates the varied
substituents.
Soft Matter
This journal is ©The Royal Society of Chemistry 2015

View Article Online
Soft Matter
Paper
Fig. 2 Flowchart of the synthesis of triarylbenzenes.
ꢀ
1
2
.2 Differential scanning calorimetry (DSC)
ordinary glasses by heating at 1 K min , annealing at T
until the thickness became constant and then cooling at 1 K min
to room temperature. An example of the transformation procedure
g
+ 20 K
ꢀ
1
DSC was performed on purified compounds using a TA Q2000
DSC instrument (Thermal Instruments). Indium melting point
test was done before all measurements to calibrate the DSC. For
each run, 5–10 mg of each compound was placed in hermetically
sealed aluminum pans (Tzero pans and lids, TA Instruments).
The pans were cooled to 273 K, ramped at 10 K min to 623 K
and cooled back to 273 K with the same cooling rate. All glass
transition temperatures (T
3
6,50
is shown in Fig. S8 in ESI.†
The resulting ordinary glasses
were then used for characterization. Subsequent heating/cooling
of the samples resulted in the same measured properties in the
super-cooled liquid regime within experimental error, except for
the onset of crystallization or dewetting in some compounds, as
detailed below. This is a strong indication that the transformed
glasses maintained their bulk-like properties and were no longer
in the stable glass state.
ꢀ
1
g
) were measured upon cooling. The
melting points (T ) were determined upon heating. Fig. 3 shows
the heat capacity data for each compound in the temperature
m
range around their T (Fig. 3(a)) and T (Fig. 3(b)). Error bars in
g
m
DSC measurements were determined by the difference between 2.4 Ellipsometry measurements
repeated trails.
Each sample was mounted onto a temperature-controlled stage
Linkam THMSEL350V, temperature range between 77 K to
23 K) using a thermally conductive paste (Arctic Silver Ceramic
(
6
2.3 Thin film preparation
Thin films of the five compounds were prepared by physical polysynthetic thermal compound). The temperature during
vapor deposition (PVD) in a custom vacuum chamber with a heating/cooling ramps was controlled by the system controller
ꢀ
7
base pressure less than 2.0 ꢁ 10 Torr. Silicon (100) wafers (Linkam PE95/T95). After calibration, dilatometry measurements
Virginia Semiconductor Inc.) with B1.4 nm native oxide layer were performed using a spectroscopic ellipsometer (J.A. Woollam
were cleaned with nitrogen gas and used as substrates. The details spectroscopic ellipsometer M-2000V) as detailed in our previous
(
4
4,50
of the custom chamber and deposition methods are published publications and ESI.†
An acquisition angle of 70 degrees
5
0
in a separate manuscript. The compounds were mounted in and time of 1 second was chosen with zone averaging enabled to
a crucible and thermally heated for PVD. The deposition rate eliminate possible systematic errors due to offsets of the polarizer’s
was controlled using the input power of the thermal source angle. During measurements, nitrogen gas was flown across the
(TDK-Lambda Gen8-90-U) and monitored by a quartz crystal samples to prevent oxidation and water uptake.
microbalance (Sycon Instrument STM-1). Depositions were Fig. 4(a) shows an example of measured ellipsometry
ꢀ
1
performed at a rate of 0.20 ꢂ 0.03 nm s and all films were angles C(l) and D(l). C(l) and D(l) represent the amplitude
prepared in a thickness range of 180–210 nm. The detailed and phase change respectively of the ratio between two reflec-
deposition temperature for each film is presented in the ESI,† tion coefficients for p- and s-polarized light. C(l) and D(l) were
50
Table S1. As described in a separate publication, the as-deposited fit to a three-layer model, as schematically shown as the inset of
samples were stable glasses. The films were transformed into Fig. 4(a), to determine the thickness and the index of refraction
This journal is ©The Royal Society of Chemistry 2015
Soft Matter

View Article Online
Paper
Soft Matter
Fig. 4 (a) Measured ellipsometric angles C and D as functions of wave-
length along with the Cauchy model fits to the data for a 197 nm film of
a,a,b-TNB at room temperature. Inset shows a schematic diagram of the
three-layer model. (b) Thickness as a function of temperature for the same
Fig. 3 Heat capacity (C
p
) as a function of temperature, in the temperature
(b) for a,a-P (red), a,a,b-TNB (orange), a,a-A
yellow), b-AA (green), and a,a,a,a-TNBP (cyan). All curves were normalized
ꢀ1
film measured on cooling at 1 K min . The region highlighted in red was
used to fit a line in the super-cooled liquid (SCL) regime and the region
g m
range around T (a) and T
(
g
highlighted by blue was used to fit a line in the glassy regime. T is
between [0,1] and vertically shifted for clarity. Molecular structures are
shown as the insets.
determined by the intersection of these two linear fits (red and blue
dashed lines) as indicated by the arrow. The inset shows the molecular
structure of a,a,b-TNB.
of the film. From bottom to top, the three-layer model consists
of a temperature-dependent model of optical properties of
silicon, a 1.4 nm thick native oxide layer, and a Cauchy model
for the optical properties of the organic films. Since the PVD
films were transparent in the spectral range chosen for the
experiments, this model is suitable. In the Cauchy model, the
relationship between the real part of the index of refraction and
Fig. S8, ESI† for a 197 nm film of a,a,b-TNB). Each glass sample
was first heated at 150 K min to its DSC T + 30 K, and held
g
isothermally for 20 minutes to be completely transformed into
the super-cooled liquid state. Various cooling rates were then
ꢀ
1
g
used to cool the glass from T + 30 K to 295 K. The last repeating
ꢀ1
ramp of 150 K min was performed to ensure that the film
maintained its property over the course of the experiment. After
each cooling ramp, the sample was held at 295 K for 5 minutes
B
C
the wavelength is described as nðlÞ ¼ A þ þ l4, while the
l2
ꢀ
1
imaginary part of the index of refraction k is kept as zero. and subsequently heated to T + 30 K at 150 K min and held
g
Overall, four parameters, A, B, C and the film thickness, h, were for another 5 minutes for equilibration. Each cooling data set
fit to 486 C(l) and D(l) data points, each simultaneously was then re-plotted as thickness vs. temperature to obtain Tg
obtained at every time point. Fig. 4(b) shows an example of values as described in Fig. 4(b). Fig. 5(c) shows that the mean
the thickness as a function of temperature for a 197 nm film of square error (MSE) was independent of the temperature and
ꢀ1
a,a,b-TNB using cooling rate of 1 K min . The value of T was time during CR-T measurements, indicating that the tempera-
g
g
determined by the intersection of linear fits to the glassy (295 K ture dependent model used for the silicon substrate and the
to T
0 K to T
Cooling rate-dependent T
g
ꢀ 20 K) and the supper-cooled liquid (SCL) regions (T
g
+
thin films was adequate in predicting the optical properties of
the system at all times and temperatures.
2
g
+ 30 K).
g
(CR-T
g
) measurements were per-
Acquisition time for ellipsometry measurements was 1 second
formed on the transformed ordinary glasses (example shown in and the temperature was recorded after the measurement.
Soft Matter
This journal is ©The Royal Society of Chemistry 2015

View Article Online
Soft Matter
Paper
g
Fig. 6 (a) Glass transition temperature (T ) as a function of molecular
Fig. 5 Raw ellipsometric CR-T
g
measurements of a 197 nm a,a,b-TNB
weight for five molecules a,a-P (red); a,a,b-TNB (orange); a,a-A (yellow);
b-AA (green); a,a,a,a-TNBP (cyan). Solid symbols represent ellipsometric T
film. (a) Calculated thickness as a function of time. (b) Temperature as a
function of time. (c) Mean square error (MSE) as a function of time. The
inset shows the molecular structure of the a,a,b-TNB. The colored regions
g
ꢀ1
values obtained at 10 K min cooling rate, while open symbols represent Tg
values determined by DSC at the same cooling rate. (b) Melting point Tm as a
function of molecular weight for the same compounds in (a) as measured by
DSC. Error bars are smaller than the symbol sizes. Red frames in both figures
highlight the increasing trend in T and T for compounds that only contain
ꢀ1
ꢀ1
ꢀ1
correspond to cooling ramps of 150 K min , 120 K min , 90 K min ,
ꢀ ꢀ1 ꢀ1 ꢀ1 ꢀ1 ꢀ1
1
6
0 K min , 30 K min , 10 K min , 7 K min , 3 K min , 1 K min and
ꢀ1
1
50 K min from left to right, respectively.
g
m
phenyl and naphthyl substituents. Blue frames highlight the increasing trend
in T
g
and T
m
for compounds that contain anthracyl substituents.
As such, there is a small discrepancy between the actual tempera-
ture of the sample during the measurement and the recorded
ꢀ
1
temperature, in particular at high cooling rates (150 K min
1
,
ꢀ1
film crystallizes rapidly at 10 K min cooling, only its DSC T
g
ꢀ
1
ꢀ1
ꢀ1
20 K min , 90 K min and 60 K min ). Corrections were
value is shown. Fig. 6(b) shows the melting point (T
mined by DSC. Both T and T values were found to increase
with increasing molecular weight in similar trends. The T s of
three compounds that only contain phenyl and naphthyl sub-
stituents, a,a-P, a,a,b-TNB and a,a,a,a-TNBP, linearly increase
with molecular weight by roughly 13 K per additional benzo
ring. This observation is consistent with previous studies on
low molecular weight polymers and n-alkanols. However,
the T s and T s of the two compounds that contain stronger
p-interacting substituents, a,a-A and b-AA, deviate from this
m
) deter-
made for the fast cooling rate ramps by shifting the tempera-
tures accordingly. After the correction, the super cooled liquid
lines for all cooling rates overlap well, which is a strong indica-
tion of the validity of this correction and reproducibility of the
data after the transformation of the as-deposited glass. The error
g
m
g
in determining T
ences between repeated trials and the standard deviations in
linear fits to glassy and SCL regimes. More details of the CR-T
g
includes the shifted temperatures, the differ-
51
52
g
g
m
4
4
method can be found in our earlier publication.
trend. The additional fused benzo ring on TNB molecule gives
rise of 20 K in T
on a,a-A increases the T by another 28 K. A similar trend is
g
, while the second additional fused benzo ring
3
Results and discussions
g
Fig. 6(a) shows the results of ellipsometric (solid symbols) and observed in T s as well. These observations suggest that the
m
DSC T (open symbols) measurements for all five compounds. details of molecular level interaction potentials are important
g
The ellipsometric T
of each compound at 10 K min . T
g
was determined by cooling B200 nm films in macroscopic properties of organic compounds, and simple
ꢀ
1
g
s determined by both relationships may not hold in the presence of strong inter-
methods are the same within the experimental error. Since b-AA molecular interactions.
This journal is ©The Royal Society of Chemistry 2015
Soft Matter

View Article Online
Paper
Some studies have used reduced T
Soft Matter
g
g m
(Trg = T /T ) to predict
the glass-forming ability of organic molecules, suggesting that
larger Trg favors glass formation and that compounds with T
rg
2
9,53,54
less than 2/3 will readily crystallize.
This relationship has
been demonstrated to hold in some homologous series of
2
7
28
molecules, as well as a,a,b-TNB and its three isomers. This
empirical relationship is also valid for compounds in this
study. Among the five compounds synthesized here b-AA has
g
the lowest reduced T (Trg = 0.66). b-AA readily crystallizes, such
ꢀ
1
that it’s critical cooling rate is above 10 K min . We hypothe-
size that the introduction of stronger p-interacting substituents
in small organic molecules strengthens the molecular level
interactions and reduces the barrier to nucleation. This is seen
by a stronger increase in T
the reduction of Trg. The reduced T
pounds are Trg(a,a-P) = 0.79, Trg(a,a,b-TNB) = 0.74, Trg(a,a-A) =
m
compared to T
g
, which results in
g
for the other four com-
0.71 and Trg(a,a,a,a-TNBP) = 0.70.
In order to determine the apparent activation energy and
fragility of each synthesized compound, vapor-deposited
glasses were further characterized using CR-T measurements.
g
Increasing cooling rate causes the super-cooled liquid to fall
out of equilibrium at a higher temperature, which corresponds
to relaxation times slower than the rate of changes in tempera-
ture. Thus, cooling rate is inversely related to the relaxation
time of the SCL at T . This relationship between cooling rate
g
and relaxation time provides a convenient method to estimate
the relaxation times of the SCL as a function of temperature
close to T as well as the fragility of the compound. Fig. 7(a)
g
shows the results of CR-Tg measurements for a thin film of
a,a-P. While the thicknesses of SCL remain identical among all
cooling ramps, the glass falls out of equilibrium at different
temperatures under various cooling rates, resulting in different
densities obtained after each cooling cycle. Glass with the highest
density (lowest thickness) is attained at the lowest cooling rate.
The difference in the linear density between glasses of the same
ꢀ
1
ꢀ1
material cooled at 150 K min and 1 K min is B0.4%. Plots of
normalized thickness as a function of temperature for the other
four compounds are shown in Fig. S9 in the ESI.† We note that
ꢀ
1
measurements at cooling rates below 30 K min
were not
possible for b-AA thin films due to fast crystallization as men-
ꢀ1
tioned above. The b-AA T
Fig. 7(b) shows an Arrhenius plot of log(t) as a function of 1000/T
plot of log(CR) vs. T shown in Fig. S10 of the ESI†) for all five
g
at 10 K min was determined by DSC.
g
(
g
compounds. t was calculated using the empirical relationship Fig. 7 (a) CR-T measurements of 200 nm vapor-deposited a,a-P glasses;
g
normalized thickness is plotted as a function of temperature at various
that t = 100 seconds corresponds to the glass transition at a
ꢀ1
ꢀ1
ꢀ1
cooling rates in the range between 150 K min to 1 K min . Film thickness
is normalized to the thickness of SCL at T + 30 K. (b) Log of relaxation time
log(t)) as a function of 1000/T for a,a-P (red), TNB (orange), a,a-A (yellow),
b-AA (green), and a,a,a,a-TNBP (cyan). t is calculated by assuming that
cooling rate of CR = 10 K min (t ꢁ CR = 1000 seconds).
g
The Vogel–Tammann–Fulcher (VFT) equation is usually
(
g
5
5
used to obtain the fragility of glasses near T
g
.
However, in
ꢀ1
the limited range of cooling rates accessible in this study, the t = 100 seconds corresponds to a cooling rate of 10 K min . The solid and
ꢀ
ꢀ
ꢁꢁ
dashed lines are fits to the data. (c) Fragility (m) of vapor-deposited glasses at
; a,a-P (red), TNB (orange), a,a-A (yellow), b-AA (green open symbol), and
a,a,a,a-TNBP (cyan). Molecular structures are shown as insets.
Ea
Arrhenius relationship t ¼ t0 exp
provides a good
T
g
kT
approximation for fragility. In this equation, E is the apparent
a
activation energy, and k is the Boltzmann constant. Further-
more, using the empirical relationship between cooling rate these glasses at T
g
. As shown in Fig. S11 of the ESI,† the
and relaxation time, the CR-T
g
data can be used to estimate apparent activation energy for B200 nm a,a-P, a,a,b-TNB, a,a-A,
both the relaxation times near T
g
and the activation energy of b-AA, and a,a,a,a-TNBP glasses are 414 kJ mol , 408 kJ mol ,
ꢀ
1
ꢀ1
Soft Matter
This journal is ©The Royal Society of Chemistry 2015

View Article Online
Soft Matter
Paper
ꢀ1
ꢀ1
ꢀ1
4
26 kJ mol , 911 kJ mol and 587 kJ mol , respectively.
Activation energies calculated from the slope of Arrhenius fits
as shown in Fig. 7(b)) can be used to evaluate the dynamic
(
fragility. The dynamic fragility index (m) quantifies the extent by
which the relaxation dynamics of SCL deviate from Arrhenius
2
0
temperature dependence. The fragility, m, is defined as;
ꢂ
ꢂ
ꢂ
ꢂ
ꢂ
ꢂ
ꢂ
ꢂ
d logðtÞ
m ¼
ꢀ
ꢁ
(1)
Tg
d
T
T¼Tg
Strong liquids, with Arrhenius temperature dependence, have
small fragility values close to T . The fragility of a,a,b-TNB
g
measured in our experiment is 63 ꢂ 2, which is in agreement
with the previous reported value of 68 determined by viscosity
5
6–59
measurements.
However, dielectric studies reported a differ-
60,61
ent fragility value of a,a,b-TNB as m E 86.
We attribute this
disparity to the various accessible temperature ranges in different
instrumentations as shown in Fig. S12 (ESI†), where all three data
sets are plotted on the same scale. Fig. S12 (ESI†) shows an
g
excellent agreement between CR-T measurements and measure-
ments of viscosity in the same temperature range. It is also
important to note that different types of dynamical quantities,
probed by various experiments, may have different temperature
dependences near T
molecules. Compounds a,a-P, a,a,b-TNB, a,a-A have similar fra-
gilities around 63, while b-AA and a,a,a,a-TNBP have fragilities of Fig. 8 Thermal expansion coefficients of the five compounds; a,a-P (red),
g
. Fig. 7(c) plots the dynamic fragility of five
1
22 and 80, respectively. We observe strong correlation between TNB (orange), a,a-A (yellow), b-AA (green), and a,a,a,a-TNBP (cyan) in the
glass, a(G), (a) and super-cooled liquid a(SCL), (b) regimes.
fragility and molecular structure. We hypothesize that the large
difference in fragility between a,a-A and b-AA may be due
to molecular rearrangement in SCL. The two anthracyl groups
in b-AA may increase the chance for dimer formation or even coefficient of a,a-P, TNB, and a,a,a,a-TNBP decrease with approxi-
ꢀ
4
ꢀ1
oligomers, which can also be seen from the high apparent mately 0.06 ꢁ 10
K
decrease per additional benzo ring. In the
activation energy and rapid crystallization around T . Our obser- glassy regime the addition of p interactions, by the addition of a
g
vation is consistent with the general trend found by Qin and fused benzo ring in a,a-A, results in a larger decrease in the
2
6
ꢀ4 ꢀ1
McKenna that the apparent activation energy and fragility thermal expansion by 0.11 ꢁ 10
increases with increasing T
in organic compounds with aromatic coefficient is measured in b-AA glass to be (1.35 ꢂ 0.03) ꢁ
substituents. The apparent activation energy and fragility again 10 lower. Interest-
K . The lowest expansion
g
ꢀ
4
ꢀ1
ꢀ4 ꢀ1
K
, which is an additional 0.07 ꢁ 10
K
confirmed that p–p intermolecular interactions play an important ingly, in the super-cooled regime, as shown in Fig. 8(b), a strong
role in sculpting physical properties of organic molecular glasses. deviation from the trend is only observed in b-AA. This suggests
Fig. 8(a) shows the thermal expansion coefficients of the five that the local environment b-AA compound in the equilibrium
compounds calculated using the slope of the lines in the glassy state may be different from that of other compounds.
and SCL regions as described in the ESI.† The thermal expan-
Thermal expansion is a measure of anharmonicity in the
sion coefficient of a,a,b-TNB glass (a(G)), and super-cooled intermolecular potentials. If all molecules in a system were
ꢀ
4
ꢀ1
liquid (a(SCL)) are measured to be (1.47 ꢂ 0.03) ꢁ 10
K
bounded by perfectly harmonic potentials, the location of the
ꢀ4
ꢀ1
and (5.31 ꢂ 0.05) ꢁ 10 K , respectively. These values are molecules during the temperature change would not change,
consistent with those measured by Dalal et al. for 500 nm films resulting in zero value for the thermal expansion coefficient.
6
2
3
6
of a,a,b-TNB.
Similar to the measurements of T
As such, a lower value of thermal expansion coefficient is
g
and T
m
, the values of interpreted as the existence of more harmonic potentials. In
thermal expansion coefficient for this series of compounds do the SCL regime, the average intermolecular distance is usually
not simply correlate with the molecular weight. While generally larger, leading to a larger value of the thermal expansion. The
the thermal expansion coefficient in both SCL and glassy regimes exceptionally low value of a(SCL) in b-AA suggest that inter-
decrease with increasing molecular weight, the compounds with molecular interactions are stronger in this compound in the
stronger p interactions have lower expansion coefficients. In the SCL state, suggesting stronger p-interactions or potentially the
glassy regime, as shown in Fig. 8(a), the thermal expansion existence of dimers or oligomers.
This journal is ©The Royal Society of Chemistry 2015
Soft Matter

View Article Online
Paper
Soft Matter
In general, the thermal expansion coefficient decreases with electronics because they promote hole-mobility. However, the
increasing intermolecular forces; such that the thermal expan- role of p–p stacking in a material’s physical properties has not
sion coefficient of a crystal is lower than that of its glass received much attention. Based on the results of our study, we
counterpart. In a,a,b-TNB, the thermal expansion of the glass suggest that p–p stacking is responsible for enhanced thermal
ꢀ
1
prepared by cooling at 1 K min is about 30% higher than that stability and decreased glass-forming ability.
5
8,59,63
of the crystal.
As seen in our experiments, the introduc-
tion of stronger p-interacting side groups does significantly
lower the thermal expansion coefficient in glasses. Increasing
molecular weight generally makes the system more harmonic,
while adding p-interacting side groups may considerably pro-
mote harmonic potentials.
Acknowledgements
Z.F. acknowledges funding from the University of Pennsylvania
and seed funding by MRSEC program of the National Science
Foundation under Award No. DMR-11-20901 at the University
of Pennsylvania. P.J.W. acknowledges funding from NSF (CHE-
1
152488). F.G. thanks the Chinese Scholarship Council for financial
4
Summary and outlook
support. The authors would like to thank Kim Mullane and Sharan
Mehta for help in collecting 1-bromo-3-chloro-5-iodobenzene synthe-
sized by undergraduate students in the Chemistry 245 laboratory
course. The authors would also like to thank Tiezheng Jia and Xinyu
Cao for acquiring NMR and IR spectra.
Despite a long-standing interest in organic glass-forming mate-
rials for various applications, systematic studies on families of
molecules remain scarce. Such studies, however, hold the key
to unraveling the structural attributes that affect glass-forming
ability and the properties of a glass both in bulk and nanometer
length scales. The scarcity of studies is due, in part, to the
lack of general synthetic methods that have been applied to
glass-forming molecules and the lack of high-throughput char-
acterization methods. A high-throughput method of character-
ization, combined with an efficient synthetic tool to produce
tailored glasses can help elucidate the role of the chemical
structure in properties of glasses.
References
1 K. A. Graeser, J. E. Patterson, J. A. Zeitler, K. C. Gordon and
T. Rades, Eur. J. Pharm. Sci., 2009, 37, 492.
2 B. C. Hancock and G. Zografi, J. Pharm. Sci., 1997, 86, 1.
3 S. Yoshioka and Y. Aso, J. Pharm. Sci., 2007, 96, 960.
4 Y. Shirota, J. Mater. Chem., 2000, 10, 1.
Herein, we have introduced a chemoselective approach
based on Suzuki-Muara cross-coupling reactions employing a
versatile precursor, 1-bromo-3-chloro-5-iodobenzene. The synthesis
of 1-bromo-3-chloro-5-iodobenzene is simple and is done in
large quantities in an undergraduate laboratory setting. This
material provides an efficient starting point for the synthesis of
glasses tailored for various applications.
5 C. Neuber, A. Ringk, T. Kolb, F. Wieberger, P. Strohriegl,
H.-W. Schmidt, V. Fokkema, M. Cooke, C. Rawlings, U. D u¨ rig,
A. W. Knoll, J.-F. de Marneffe, P. De Schepper, M. Kaestner,
Y. Krivoshapkina, M. Budden and I. W. Rangelow, in SPIE Adv.
Lithogr., ed. D. J. Resnick and C. Bencher, International Society
for Optics and Photonics, 2014, p. 90491V.
6 C. W. Tang and S. A. Vanslyke, Appl. Phys. Lett., 1987,
51, 913.
Cooling rate-dependent T measurements provide a reliable
g
high-throughput characterizing tool to measure properties
of thin films of glasses and super-cooled liquids. One set of
7 Y. Shirota and H. Kageyama, Chem. Rev., 2007, 107, 953.
8 Y. Shirota, J. Mater. Chem., 2000, 10, 1.
measurements provides information on the values of T
lity and apparent expansion coefficients of the bulk and super-
g
, fragi-
9 E. A. Schmitt, D. Law and G. G. Zhang, J. Pharm. Sci., 1999,
88, 291.
cooled liquids. Using this method, we have demonstrated 10 J. G. C. Veinot and T. J. Marks, Acc. Chem. Res., 2005, 38, 632.
that four structural analogues of a good organic glass former, 11 X. Shi, A. Mandanici and G. B. McKenna, J. Chem. Phys.,
a,a,b-TNB, can all form glasses. Studies in compounds contain-
ing only phenyl and naphthyl substituents validate the general 12 R. Richert, P. Lunkenheimer, S. Kastner and A. Loidl,
trend that increasing molecular weight leads to increasing Tm, J. Phys. Chem. B, 2013, 117, 12689.
Tg, apparent activation energy, fragility, and decreasing thermal 13 G. B. McKenna and S. L. Simon, Handb. Therm. Anal.
expansion coefficients. Calorim., 2002, 49–109.
Furthermore, we have demonstrated that increasing the strength 14 Y. Sun, L. Zhu, K. L. Kearns, M. D. Ediger and L. Yu, Proc.
and number of p–p interacting substituent in organic glass Natl. Acad. Sci. U. S. A., 2011, 108, 5990.
formers can affect various physical properties of both glasses 15 K. Paeng, C. T. Powell, L. Yu and M. D. Ediger, J. Phys. Chem.
and super-cooled liquids. The molecules containing anthracyl Lett., 2012, 3, 2562.
substituents showed stronger elevation in both Tm and T and 16 M. D. Ediger and P. Harrowell, J. Chem. Phys., 2012,
2005, 123, 174507.
g
activation energies, and greater decrease in thermal expansion
coefficients. These measurements show that stronger p inter- 17 M. Yoshioka, B. C. Hancock and G. Zografi, J. Pharm. Sci.,
actions increase fragility and decrease the ability of the com- 1994, 83, 1700.
pounds to form glasses. Complexes that exhibit p–p stacking 18 M. J. Pikal, L. L. Chang and X. C. Tang, J. Pharm. Sci., 2004,
137, 080901.
are often preferred in designing materials used in organic
93, 981.
Soft Matter
This journal is ©The Royal Society of Chemistry 2015

View Article Online
Soft Matter
Paper
19 R. Surana, A. Pyne and R. Suryanarayanan, Pharm. Res., 41 Y. Yue, R. von der Ohe and S. L. Jensen, J. Chem. Phys., 2004,
2004, 21, 867.
120, 8053.
2
2
2
2
0 C. A. Angell, Science, 1995, 267, 1924.
1 K. Naito and A. Miura, J. Phys. Chem., 1993, 97, 6240.
2 S. F. Swallen and M. D. Ediger, Soft Matter, 2011, 7, 10339. 43 Z. Fakhraai and J. Forrest, Phys. Rev. Lett., 2005, 95, 025701.
3 A. P. Kulkarni, C. J. Tonzola, A. Babel and S. A. Jenekhe, 44 E. C. Glor and Z. Fakhraai, J. Chem. Phys., 2014, 141, 194505.
42 Z. P. Lu, Y. Li and S. C. Ng, J. Non-Cryst. Solids, 2000,
270, 103.
Chem. Mater., 2004, 16, 4556.
4 H. K. Huynh, T. D. Khong, S. L. Malhotra and
45 P. Badrinarayanan, W. Zheng, Q. Li and S. L. Simon,
J. Non-Cryst. Solids, 2007, 353, 2603.
2
2
2
2
2
2
3
3
3
3
3
L. P. Blanchard, Anal. Chem., 1978, 50, 976.
5 D. Huang, S. L. Simon and G. B. McKenna, J. Chem. Phys.,
46 N. Shamim, Y. P. Koh, S. L. Simon and G. B. McKenna,
J. Polym. Sci., Part B: Polym. Phys., 2014, 52, 1462.
47 D. R. Uhlmann, J. Non-Cryst. Solids, 1972, 7, 337.
2005, 122, 84907.
6 Q. Qin and G. B. McKenna, J. Non-Cryst. Solids, 2006, 48 J. Gilbert and S. Martin, Experimental Organic Chemistry,
52, 2977. Thomson Brook/Cole, Belmont, CA, 4th edn, 2006.
7 W. Ping, D. Paraska, R. Baker, P. Harrowell and C. A. Angell, 49 R. Martin and S. L. Buchwald, Acc. Chem. Res., 2008,
J. Phys. Chem. B, 2011, 115, 4696. 41, 1461.
8 K. Dawson, L. Zhu, L. A. Kopff, R. J. McMahon, L. Yu and 50 T. Liu, K. Cheng, E. Salami-Ranjbaran, F. Gao, C. Li,
3
M. D. Ediger, J. Phys. Chem. Lett., 2011, 2, 2683.
9 C. Alba-Simionesco, J. Fan and C. A. Angell, J. Chem. Phys.,
X. Tong, Y.-C. Lin, Y. Zhang, W. Zhang, L. Klinge,
P. J. Walsh and Z. Fakhraai, J. Phys. Chem., 2015, DOI:
10.1063/1.4928521.
1999, 110, 5262.
0 O. Lebel, T. Maris, M.-E. Perron, E. Demers and J. D. Wuest, 51 A. L. Agapov and A. P. Sokolov, Macromolecules, 2009,
J. Am. Chem. Soc., 2006, 128, 10372. 42, 2877.
1 C. R. Daley, Z. Fakhraai, M. D. Ediger and J. A. Forrest, Soft 52 L.-M. Wang and R. Richert, J. Phys. Chem. B, 2007, 111, 3201.
Matter, 2012, 8, 2206. 53 D. Turnbull, Contemp. Phys., 1969, 10, 473.
2 S. F. Swallen, K. Traynor, R. J. McMahon, M. D. Ediger and 54 W. Kauzmann, Chem. Rev., 1948, 43, 219.
T. E. Mates, J. Phys. Chem. B, 2009, 113, 4600.
3 L. Zhu, C. W. Brian, S. F. Swallen, P. T. Straus, M. D. Ediger
and L. Yu, Phys. Rev. Lett., 2011, 106, 256103.
55 M. D. Ediger, C. A. Angell and S. R. Nagel, J. Phys. Chem.,
1996, 100, 13200.
56 J. H. Magill and D. J. Plazek, J. Chem. Phys., 1967, 46, 3757.
4 S. F. Swallen, K. L. Kearns, M. K. Mapes, Y. S. Kim, R. J. 57 M. D. Ediger, P. Harrowell and L. Yu, J. Chem. Phys., 2008,
McMahon, M. D. Ediger, T. Wu, L. Yu and S. Satija, Science,
007, 315, 353.
5 K. Dawson, L. A. Kopff, L. Zhu, R. J. McMahon, L. Yu,
R. Richert and M. D. Ediger, J. Chem. Phys., 2012, 136, 094505. 59 C. M. Whitaker and R. J. McMahon, J. Phys. Chem., 1996,
6 S. S. Dalal, A. Sepulveda, G. K. Pribil, Z. Fakhraai and
M. D. Ediger, J. Chem. Phys., 2012, 136, 10.
7 A. Sep u´ lveda, S. F. Swallen, L. A. Kopff, R. J. McMahon and
M. D. Ediger, J. Chem. Phys., 2012, 137, 204508.
8 D. Yokoyama, Y. Setoguchi, A. Sakaguchi, M. Suzuki and
C. Adachi, Adv. Funct. Mater., 2010, 20, 386.
9 D. Yokoyama and C. Adachi, J. Appl. Phys., 2010, 107, 7.
0 D. Yokoyama, Z. Qiang Wang, Y.-J. Pu, K. Kobayashi, J. Kido 63 J. H. Magill and A. R. Ubbelohde, Trans. Faraday Soc., 1958,
and Z. Hong, Sol. Energy Mater. Sol. Cells, 2012, 98, 472. 54, 1811.
128, 034709.
2
58 C.M. Whitaker and R.J. McMahon, Magill in 1958 and
Plazek in 1966 misidentified aab-TNB as aaa-TNB.
3
3
3
3
100, 1081.
60 R. Richert, K. Duvvuri and L. T. Duong, J. Chem. Phys., 2003,
118, 1828.
61 A. Sep u´ lveda, M. Tylinski, A. Guiseppi-Elie, R. Richert and
M. D. Ediger, Phys. Rev. Lett., 2014, 113, 045901.
62 N. W. Ashcroft and N. D. Mermin, Solid State Physics, Holt,
Rinehart and Winston, New York, 1976.
3
4
This journal is ©The Royal Society of Chemistry 2015
Soft Matter