Polymer 143 (2018) 336e342

Contents lists available at ScienceDirect

Polymer
journal homepage: www.elsevier.com/locate/polymer

Thermal transitions in semi-crystalline polymer thin films studied via

spectral reflectance
Giovanni M. Kelly a, James F. Elman b, Zhang Jiang c, Joseph Strzalka c, Julie N.L. Albert a, *
a
Department of Chemical and Biomolecular Engineering, Tulane University, New Orleans, LA, 70118, United States
b
Filmetrics Applications Lab, Rochester, Fairport, NY, 14450, United States
c
X-Ray Science Division, Argonne National Laboratory, Argonne, Illinois, 60439, United States

a r t i c l e i n f o a b s t r a c t

Article history: Herein we report the finding that spectral reflectance (SR) can be used to identify thermal transitions in
Received 4 January 2018 semi-crystalline polymer thin films. By studying film thickness as a function of temperature, we found
Received in revised form that semi-crystalline polymers exhibit characteristic expansion and contraction profiles during the
29 March 2018
melting (Tm) and crystallization (Tc) transitions, respectively. Prior to this discovery, studies on the
Accepted 3 April 2018
Available online 4 April 2018
crystalline Tm in thin films have involved more expensive and complex techniques such as atomic force
microscopy (AFM), ellipsometry, and grazing-incidence wide-angle X-ray scattering (GIWAXS). We
correlate Tm and Tc as measured by SR with differential scanning calorimetry (DSC) in the bulk and
Keywords:
Crystallization
temperature-controlled GIWAXS or AFM in thin films. We show that SR is accurate for measuring
Melting temperature changes in films with thicknesses of 500 nm down to 21 nm and for detecting melting point depression
Spectral reflectance due to thin film confinement. Thus, we demonstrate that SR is a powerful tool for measuring thermal
transitions in semi-crystalline polymer films with single-degree resolution.
© 2018 Elsevier Ltd. All rights reserved.

1. Introduction [4,31e33]. Semi-crystalline polymers, such as poly(caprolactone)

(PCL), polyethylene, poly(ethylene glycol) (PEG), and poly(lactic
Characterization of the physical properties and morphologies of acid) (PLA), have a second characteristic thermal transition, the
polymers confined to thin and ultrathin film geometries is an melting point [34e36]. Most of these polymers are used at tem-
important area of research in polymer science and engineering peratures well above their glass transition where crystallinity
[1e7] due to the wide applicability of such materials to a variety of rather than glassiness dictates chain mobility; thus, the melting
technologies, including lithography [6,8e11], water purification point (Tm) rather than Tg determines material stability and appli-
[12,13], fuel cells [14,15], photovoltaics [16e19], and biological cability for these polymers. In thin films, much work has been done
studies [20,21]. The pace of study and interest in polymer thin films to understand the effect that blending with both polymeric and
has driven researchers to develop new methods to probe their non-polymeric additives, as well as the thin film geometry itself,
many defining physical characteristics. For example, in amorphous has on crystallization. The groups of T. Ogihara [37], Q. Fu [38e40],
polymers, the glass transition temperature (Tg) dictates the onset of T. Nishi [41], R.E. Prud'homme [42,43], and others have shown that
chain mobility, which is an important physical change that can the interaction between miscible and immiscible components can
determine annealing kinetics [10,22e24], affect polymer-drug increase or decrease crystallization rates and thermal transition
composite stability [25e27], and govern the ductility of consumer temperatures, and that the substrate and second component have
plastics [28e30], among other properties. In substrate-supported large effects on both overall morphology and substrate wettability
thin films, many studies have focused on how the glass transition [5,44e48].
is affected by the substrate chemistry. This body of work has shown There are a number of methods for probing the thermal tran-
that attractive versus repulsive interactions between the substrate sitions of polymers. Differential scanning calorimetry (DSC) has
and polymer can lead to Tg elevation or depression, respectively been used for decades to indirectly quantify thermal transitions in
both semi-crystalline and amorphous polymers in the bulk. How-
ever, thin films ranging from tens to hundreds of nanometers do
* Corresponding author. not contain sufficient material for this technique to maintain
E-mail address: [email protected] (J.N.L. Albert).

https://doi.org/10.1016/j.polymer.2018.04.017
0032-3861/© 2018 Elsevier Ltd. All rights reserved.
 G.M. Kelly et al. / Polymer 143 (2018) 336e342 337

accuracy [49e56]. Several companies and research groups have 2. Experimental section
pioneered and utilized advanced calorimetry techniques, including
“micro-DSC” and “nano-DSC” as well as differential fast scanning 2.1. Materials
calorimetry (DFSC) units; however, these units come with a sig-
nificant cost and most prior studies have used orders of magnitude Hydroxyl-terminated linear poly(ε-caprolactone) (PCL) (Sigma
more material than is available in nanometer thin films [57e60]. Aldrich, Mn ¼ 45 kDa, Ð ¼ 1.5) as well as poly(ethylene glycol) (PEG)
Even for units that require only nanograms of material, the film (Sigma Aldrich, Mn ¼ 20 kDa) were used as received.
must still be removed from the substrate prior to measurement, as
detailed in the experimental work by Balko et al. in which they
2.2. Film preparation
studied the crystallinity of poly(3-hexylthiophene) in films down to
26 nm thickness using DFSC [61]. Unfortunately, removal of the film
PCL and PEG films were cast via spincoating from dilute toluene
from the substrate prior to measurement renders studies on the
or chloroform solutions, respectively, onto ultra-violet ozone
effect of surface interactions on melting and crystallization transi-
cleaned silicon wafers. Both solvents were ACS grade and used as
tions in thin films impossible by this method. Alternatively, poly-
received. Solution concentrations ranged from 0.25 wt. % to 7.5 wt.
mer can be deposited and melted directly on the measurement cell,
% depending on the final film thickness desired as well as the
as was done by Tardif et al. in studying the crystallization of poly-
polymer molecular weight. After casting, all films were heated to
ether ether ketone (PEEK) [62], but this technique results in micron
100
C for 10 min to erase thermal history and subsequently
rather than nanometer thick films. Several groups have utilized
quenched to room temperature (25
C) where they were allowed to
fluorescence, Brillouin light scattering (BLS), and ellipsometry to
crystallize for several hours prior to performing thermal charac-
study the glass transition temperature in substrate-supported and
terization experiments.
free standing thin films of polystyrene (PS), PS/poly(2-vinyl pyri-
dine), poly(methyl methacrylate) (PMMA), and cyclic polystyrene
[4,32,33,63e69]. Kim et al. used ellipsometry to study melting 2.3. Characterization
point depression in poly[ethylene-co-(vinyl acetate)] thin films
[70]. Thermal transitions also can be probed using shear force Bulk thermal transitions were studied via differential scanning
modulus (SFM) measurements as was done by Wang et al. although calorimetry (DSC) (TA Instruments Q200) in an inert nitrogen at-
this can be a destructive and expensive method [34]. mosphere with heating and cooling rates of 1
C/min. This data was
In this work, we describe our finding that spectral reflectance analyzed using the TA Instruments Universal Analysis software.
(SR), a technique for measuring the film thickness of polymeric and To probe the thin film melting point on a molecular level, PCL
non-polymeric materials, can be used to identify the thermal thin films were studied via temperature-controlled grazing inci-
transitions of semi-crystalline PCL and PEG thin films due to the dence wide-angle X-ray scattering (GIWAXS). GIWAXS measure-
thermal expansion or contraction of these films during melting or ments were performed at beamline 8-ID-E of the Advanced Photon
crystallization, respectively. PCL serves as a model polymer for Source at Argonne National Laboratory [76] with a heater/cooler
crystallization studies due to the similarity of its crystal structure to stage constructed in-house. An X-ray wavelength of l ¼ 1.1417 Å
that of polyethylene, a widely used commodity plastic [71], its (energy 10.86 keV) was used. The sample was measured under
accessible melting temperature (59e64
C) [72], and its relatively vacuum at an incident angle q ¼ 0.14
and exposure time of 1 s.
slow crystal growth rate (e.g., it is orders of magnitude slower than Starting from 25
C, the temperature was increased at a rate of 1
C/
PEG [42]), which facilitates study of morphology development and min up to 85
C and subsequently cooled at 1
C/min to 25
C, with
kinetics. PEG, typically used in biomedical, solution, and drug de- measurements taken every minute. Due to the overall exposure
livery applications, was chosen for comparison to PCL because it has time (120 s total) we observed some radiation damage on the
a similar bulk melting temperature (64e66
C) [73] but exhibits exposed sample section. However, it is likely that this damage did
different thin film properties that highlight the accuracy of this not occur until the polymer film reached the melt state (~35 s total
technique. Spectral reflectance has been used previously by Lu et al. exposure) based on a systematic study by Stein and coworkers [77].
to investigate the stress relaxation and glass transition temperature Additionally, these sections were not used in spectral reflectance
(Tg) of polystyrene films on ionic liquids [74]. To the best of the measurements. The acquired GIWAXS data (as two-dimensional
authors' knowledge, this work represents the first time that spec- images) were further treated and analyzed using GIXSGUI soft-
tral reflectance has been used to probe the melting and crystalli- ware package [78]. The degree of crystallinity was determined by
zation of semi-crystalline polymers and its first application to integrating the 2D GIWAXS data to generate a 1D Intensity vs. q
identifying thermal transitions in substrate-supported polymer curve; fitting this curve with four Gaussian peaks, one representing
thin films. Spectral reflectance offers a variety of benefits over the the amorphous component and three representing the crystalline
previously discussed techniques. These units are tens of thousands peaks ((003), (110), and (200)) [79,80]; and then taking the ratio of
of dollars less expensive than ellipsometers and advanced calo- the crystalline peak area to the overall area (amorphous plus
rimetry units, making them much more accessible to researchers. crystalline peak fits) as the degree of crystallinity (see Supporting
Unlike calorimetry experiments, spectral reflectance measure- Information, Fig. S1). Although GIWAXS is not the preferred
ments can be performed in situ without removing the film from the method to calculate true crystallinity, this analysis enabled us to
surface, which means that these measurements will capture the correlate the film thickness trend measured by SR with well-
effect of substrate surface interactions on the melting and crystal- defined characteristics of the melting transition.
lization transitions in thin films, if present. Compared to SFM and X- Temperature-controlled atomic force microscopy (AFM) exper-
ray studies, spectral reflectance is less destructive to the sample. iments were conducted on a Bruker Dimension ICON AFM oper-
The small footprint and equipment simplicity make spectral re- ating in PeakForce Quantitative Nanomechanical Mapping
flectometers mobile and easily-modified for non-ambient experi- (PFQNM) mode and using ScanAsyst-Air tips (k ¼ 0.4 N/m; resonant
ments such as the ones described in this work. Finally, the frequency ¼ 70 kHz). The temperature was increased manually in
resolution afforded by spectral reflectance makes nanometer 1
C increments and rapid (~30 s) scans were taken after each step
changes in film thickness readily quantifiable [75]. change to mimic an approximate 1
C/min heating rate used in the
SR experiment.
338 G.M. Kelly et al. / Polymer 143 (2018) 336e342

2.4. Spectral reflectance

Spectral reflectance measurements were taken using the Fil-

metrics F20-UV system. This system is able to resolve polymeric
and non-polymeric film thicknesses from 1 nm to 40 mm (with
resolution the greater of 1 nm or ± 0.2%). The instrument measures
the intensity of sample-reflected UV and visible light
(200e1100 nm) and the FILMeasure software is used to determine
the film thickness by fitting a calculated optical model spectrum to
the observed reflectance spectrum. This model is constructed based
on the properties of the substrate, the thin film material, and the
measurement medium. The thin film material properties used in
the model include the wavelength dependence for the refractive
index (n) and extinction coefficient (k), which were calculated by
the FILMeasure software. For both PCL and PEG, the calculated
extinction coefficient was negligible (k < 0.0001) signifying no ab-
sorption of light energy which is typical for non-aromatic, non-
conjugated polymers [81,82]. However, because the FILMeasure
software can be used to fit n and k for materials exhibiting strong
absorptions in the visible spectrum [75], the approach described
herein is expected to be readily extendable to light-absorbing
polymers. Additionally, the dependence of refractive index on
temperature is assumed to be negligible for PCL and PEG under
these conditions, as several groups have shown that refractive in-
dex decreases by less than 1% over the span of many tens of degrees
[83,84]. We also observed no evidence that the degree of crystal-
linity or increase in amorphous fraction through melting affected
the calculated refractive index model. Details on the film thickness
models used in this study are provided in the Supporting Infor-
mation, Fig. S2 and pp. 3e4. A deeper discussion of the funda-
mentals of spectral reflectance is beyond the scope of this
manuscript, and more exhaustive application notes are available
from Filmetrics Inc. Although the full range of UV and visible light
wavelengths were used in this study, measurements can be con-
ducted using only visible light wavelengths (i.e., without the UV
lamp) for samples that are UV-sensitive. Ambient measurements
were conducted on a simple sample stage. Measurements at
elevated temperature were conducted with the sample held on a
separate Linkam heater/cooler stage (Linkam THMS600E) so that
film thickness could be measured accurately as a function of
temperature.
In a typical melting point experiment, the initial temperature
was 25
C, the heating rate was 1
C/min, and film thickness was
measured in triplicate every 5
C from 25
C to 50
C, every 1
C
from 50
C to 65
C, and every 5
C from 65
C to 85
C. This heating
rate was used to correspond to the heating rate capabilities of both
in situ elevated temperature GIWAXS as well as DSC. In crystalli-
zation experiments, the film was heated to 100
C, held for 10 min,
and then the film was cooled at a rate of 1
C/min down to 25
C. In
these experiments, measurements were taken every 5
C from
100
C to 50
C and every 1
C from 50
C to 25
C.

3. Results and discussion

Fig. 1. Melting point determination for a 500 nm 45 kDa PCL thin film: a) Thickness vs.
temperature (open diamond symbols) and degree of crystallinity vs. temperature as
3.1. Tm of thin film PCL via spectral reflectance (SR) determined from GIWAXS (black X's). This film exhibits characteristic expansion upon
melting with a transition at approximately 61
C. Blue, purple, and red shading indicate
The film thickness as measured by SR during melting of a the semi-crystalline, melting transition, and melt state regions, respectively. b) Melting
point studied via elevated temperature GIWAXS showing loss of crystalline order
500 nm thick film of PCL is presented in Fig. 1a (open diamond
between 60
C and 61
C c) Analysis of slope of the thickness versus temperature curve
symbols). Corresponding to the shading in Fig. 1a, the film exhibits from (a) showing a primary peak value at 57
C. This value corresponds to when the
three distinct thermal expansion phenomena in different temper- degree of crystallinity has decreased by half during melting and can be considered the
ature ranges. From 25
C to approximately 51
C, the film expands midpoint of the melting transition region. (For interpretation of the references to
colour in this figure legend, the reader is referred to the Web version of this article.)
almost-linearly due to the thermal expansion of the amorphous
regions of the semi-crystalline PCL. During the melting transition,
we observe a large increase in thickness as melting destroys the
dense crystalline lamellae, which leads to a sharp rise in material
 G.M. Kelly et al. / Polymer 143 (2018) 336e342 339

volume. It is well-known that there is a substantial increase in the film PEG samples. While melting point depression has been shown
specific volume upon melting [85]. In the fully melted state, the to occur in polyethylene using SFM [34] and poly[ethylene-co-(vi-
film again expands linearly, as it is entirely amorphous and of nyl acetate)] thin films using ellipsometry [70], to the best of our
uniform density. The boundary between the melting transition knowledge no previous studies have focused on Tm depression in
region and the fully melted state is readily identified by the sharp PCL or PEG thin films. Knowledge of the existence and magnitude of
change in the slope of the film thickness versus temperature curve Tm depression is vital for kinetics studies, as any deviation from the
and occurs at 61
C ± 1
C, which is in good agreement with the bulk and equilibrium Tm also shifts the crystallization degree of
bulk Tm observed via DSC (60
C ± 1
C) (see Supporting Informa- supercooling, which greatly affects nucleation and growth phe-
tion, Fig. S3). Repeated trials exhibit the same characteristic curve nomena [48]. From SR measurements on the PCL films, we found
and transition point accurate to within 1
C regardless of the that PCL in this film thickness range did not exhibit appreciable Tm
starting film thickness (see Supporting Information, Fig. S4). depression (see Supporting Information, Fig. S6). However, our data
Additional experiments were conducted at heating rates of 5
C/ show that SR is sensitive enough to measure the characteristic
min and 10
C/min to illustrate the resolution of SR with faster thermal expansion upon melting despite the very low amount of
heating rates (see Supporting Information, Fig. S5). With increasing material in these films.
heating rates, the Tm shifts to slightly lower values (59
C at 5
C/ To validate SR as a valuable technique for measuring the thermal
min and 57
C at 10
C/min). The phenomenon of decreasing Tm transitions of various polymer systems while retaining the accuracy
with increasing heating rate has been shown in polyethylene necessary to examine Tm depression, we then studied four PEG
terephthalate (PET) and is caused by polymer crystals reorganizing films. The bulk film (1130 nm thickness) had a melting temperature
during heating to form higher melting point crystals with slower of 64
C as determined by SR (see Fig. 2a), which is in good
heating rates [86]. agreement with the bulk Tm reported in the literature of 64e66
C
We next verified the melting temperature of 500 nm thick PCL [73]. Films at 659 nm and 316 nm exhibited melting points of 63
C
films using GIWAXS (see Fig. 1b). This technique probes the mo-
lecular level organization of materials making it a highly sensitive
technique for detecting the disappearance of PCL crystals upon
melting, and because GIWAXS measurements were taken every
minute during the temperature ramp, we obtained excellent res-
olution of the melting temperature from this study. Fig. 1b shows
the intensity versus q plots as a function of temperature for the
sample depicted in Fig. 1a. As can be seen in Fig. 1b, the sharp peaks
that are characteristic of the crystalline regions in PCL [72,73]
disappear at approximately 61
C, which is in good agreement with
DSC and our determination of Tm from SR data.
The above analysis identifies Tm as the temperature above which
crystallinity is no longer detectable in the polymer thin film.
However, it is difficult to characterize the onset of melting from
visual analysis of the intensity vs. q curves in Fig. 1b. Thus, we
calculated the degree of crystallinity as a function of temperature
from the GIWAXS data for comparison with SR data, as described in
the Experimental section. This analysis enabled us to correlate the
film thickness trend measured by SR with three additional char-
acteristics of the melting transition. First, the most dramatic film
thickness increase (primary peak in the slope of the film thickness
versus temperature curve, see Fig. 1c) at 57
C corresponds to
where the degree of crystallinity has decreased by half and can be
considered the midpoint of the melting transition. Second, we see
that the onset of melting is gradual from z50 to 53
C and then the
degree of crystallinity drops off sharply; this sharp decrease is
consistent with the beginning of the melting transition region we
would calculate from SR data (53
C) based on identification of the
end-point (61
C) and the midpoint (57
C). Finally, we observe two
distinct peaks in the slope of film thickness vs. temperature curve
(Fig. 1c), signifying two sharp increases in the film thickness with
temperature, and therefore two separate melting events during the
overall melting of the PCL crystals. Correspondingly, the melting
peak in DSC has a low shoulder (see Supporting Information,
Fig. S3). The phenomenon of multiple melting in PCL has been
observed by Cordova et al. [87] and can be attributed to partial
reorganization of PCL crystals during the slow heating ramp as
described above.

3.2. Ultrathin PCL films and PEG Fig. 2. Film thickness vs. temperature for: a) “bulk” 1130 nm 20 kDa PEG film; and b)
147 nm 20 kDa PEG thin film. This comparison shows that SR can be used to detect Tm
depression in PEG due to thin film confinement. Blue, purple, and red shading indicate
To test the robustness of SR for even thinner films and other the semi-crystalline, melting transition, and melt state regions, respectively. (For
polymers, we studied additional PCL thin and ultrathin films interpretation of the references to colour in this figure legend, the reader is referred to
ranging from 107 nm down to 21 nm as well as bulk film and thin the Web version of this article.)
340 G.M. Kelly et al. / Polymer 143 (2018) 336e342

and 58
C, respectively (see Supporting Information, Fig. S7).

Finally, the thinnest sample (147 nm) was an order of magnitude
lower in film thickness as compared to the bulk sample (1130 nm)
and exhibited a melting temperature of 53
C (see Fig. 2b). To verify
the accuracy of the melting point determined by SR for the 147 nm
PEG film, we additionally studied this film using elevated-
temperature atomic force microscopy (AFM) (see Supporting In-
formation, Fig. S8). By capturing images of the PEG film morphology
as a function of temperature, we identified the melting tempera-
ture as 53
C, the same as determined by SR.
As shown in Fig. 3, by comparing the melting points of PCL (open
squares) and PEG (open diamonds) as a function of film thickness, it
is apparent that these two materials experience different
confinement-driven melting point depression. Even in ultrathin
films at thicknesses as low as 21 nm, the observed Tm for PCL re-
mains within the temperature range measured via DSC in the bulk
(60
C ± 1
C), whereas the Tm of PEG begins to decrease at film Fig. 4. Thickness vs. temperature of 500 nm 45 kDa PCL for a 1
C/min recrystallization
thicknesses in the hundreds of nanometers. A deeper analysis and from 100
C. This film exhibits a sharp decrease in film thickness starting at 47
C,
discussion of the possible chemical and physical origins of this which is within a few degrees of the onset of crystallization in the DSC endotherm.
Red, purple, and blue shading indicate the melt state, crystallization transition, and
difference is beyond the scope of this manuscript. However, these
semi-crystalline regions, respectively. (For interpretation of the references to colour in
results verify that the SR technique is sensitive to quantifying Tm this figure legend, the reader is referred to the Web version of this article.)
depression (if present) for changes that are greater than 1
C and
illustrate its extension to another polymer system.
degrees of the onset of the exothermic trace observed in DSC
(45
C) (see Supporting Information, Fig. S3).
3.3. Crystallization temperature via SR

Finally, we aimed to correlate the film thickness vs. temperature 4. Conclusion

data measured by SR with the crystallization of PCL from the melt.
We hypothesized that since there is a characteristic expansion upon In this work, we showed that spectral reflectance, a method of
melting, there will be corresponding contraction when crystalli- determining the film thickness of polymeric and non-polymeric
zation occurs and more densely packed chains order from the materials, can be used to detect thermal transitions in semi-
amorphous melt. Using a cooling rate of 1
C/min, we again crystalline polymer thin films based on the thermal expansion of
observed three regions in the film thickness versus temperature the films during melting and contraction of the films during crys-
plots: first, a linear trend as the amorphous melt contracts; second, tallization from the melt. One of the novelties of spectral reflec-
a sharp drop in film thickness through the crystallization transi- tance is its experimental simplicity. With knowledge of the
tion; and third, a nearly-linear trend below the crystallization substrate, film material, and measurement medium (usually air),
temperature as the amorphous regions of the semi-crystalline film the user can easily measure film thicknesses ranging from single
continue to contract (see Fig. 4). However, the crystallization nanometers to many microns. Through careful temperature studies
transition does not occur at the observed melting point. Instead, the using SR measurements, we showed that semi-crystalline polymer
onset of crystallization as determined from SR occurs at approxi- thin films exhibit characteristic expansion and contraction profiles
mately 47
C, where the linear decrease in film thickness through around the melting and crystallization transitions, respectively, and
the melt state turns into a sharp drop in film thickness through the we correlated these profiles with DSC, temperature-controlled
crystallization transition. This temperature value is within a few GIWAXS, and temperature-controlled AFM data. Specifically, we
showed that from film thickness versus temperature curves con-
structed from SR data, the melting temperature corresponds to the
point at which the thermal expansion becomes linear following the
melting transition, and the midpoint of the melting transition
corresponds to a maximum in the slope the curve. Similarly, the
onset of crystallization as determined by the point at which the
thermal contraction occurs correlates well with DSC measurement.
Additionally, we showed that this method is capable of quantifying
the significant melting point depression of semi-crystalline PEG
films due to thin film geometry. Thus, SR is a simple, low-cost, and
easily accessible technique that can be used to determine the
melting and crystallization temperatures of ultrathin films in situ,
without removal from the substrate, and for which the small
amount of material renders other laboratory techniques infeasible.
Fig. 3. Observed Tm vs film thickness for 45 kDa PCL (red open squares) and 20 kDa
PEG (black open diamonds). This comparison illustrates the confinement driven
melting point depression in these two materials. No appreciable Tm depression is Acknowledgements
observed in 45 kDa PCL, even in films as thin as 21 nm. In 20 kDa PEG films, the onset of
Tm depression can be seen in the 659 nm film (Tm ¼ 63
C) and becomes more dramatic The authors gratefully acknowledge financial support from the
as the film thickness decreases. Error bars represent the ±1
C window we report as
the uncertainty associated with determining Tm from SR data. (For interpretation of
Louisiana Board of Regents (G. M. Kelly, Graduate Fellowship,
the references to colour in this figure legend, the reader is referred to the Web version Contract Number 017GF-13) and the National Academies Gulf
of this article.) Research Program (J. N. L. Albert, Early-Career Research Fellowship,
 G.M. Kelly et al. / Polymer 143 (2018) 336e342 341

Grant Number 2000005803). Also, this material is based on work block copolymer films zone annealed below the order disorder transition
temperature, Nano Letters 7 (9) (2007) 2789e2794.
supported by the National Science Foundation under Grant Number
[23] J.N.L. Albert, W.-S. Young, R.L. Lewis III, T.D. Bogart, J.R. Smith, T.H. Epps III,
1554555 (J. N. L. Albert, NSF CBET CAREER). We thank Dr. Curtis Systematic study on the effect of solvent removal rate on the morphology of
Jarand and the PolyRMC laboratory (Tulane University) for assis- solvent vapor annealed ABA triblock copolymer thin films, ACS Nano 6 (1)
tance in conducting DSC experiments. GIWAXS experiments were (2011) 459e466.
[24] J.E. Seppala, R.L. Lewis III, T.H. Epps III, Spatial and orientation control of cy-
conducted on beamline 8-ID-E of the Advanced Photon Source lindrical nanostructures in ABA triblock copolymer thin films by raster solvent
(APS) at Argonne National Laboratory under Proposal GUP-49891. vapor annealing, ACS Nano 6 (11) (2012) 9855e9862.
Use of the Advanced Photon Source, an Office of Science User Fa- [25] G. Van den Mooter, M. Wuyts, N. Blaton, R. Busson, P. Grobet, P. Augustijns,
R. Kinget, Physical stabilisation of amorphous ketoconazole in solid disper-
cility operated for the U.S. Department of Energy (DOE) Office of sions with polyvinylpyrrolidone K25, Eur. J. Pharmaceut. Sci. 12 (3) (2001)
Science by Argonne National Laboratory, was supported by the U.S. 261e269.
DOE under Contract No. DE-AC02-06CH11357. Finally, we [26] A.A. Ambike, K. Mahadik, A. Paradkar, Spray-dried amorphous solid disper-
sions of simvastatin, a low Tg drug: in vitro and in vivo evaluations, Phar-
acknowledge Dr. Brian S. Mitchell and Dr. Sourav Chatterjee (Tulane maceut. Res. 22 (6) (2005) 990e998.
University), who provided invaluable discussion of the data. [27] T. Vasconcelos, B. Sarmento, P. Costa, Solid dispersions as strategy to improve
oral bioavailability of poor water soluble drugs, Drug Discovery Today 12 (23)
(2007) 1068e1075.
Appendix A. Supplementary data [28] D. Legrand, Crazing, yielding, and fracture of polymers. I. Ductile brittle
transition in polycarbonate, J. Appl. Polym. Sci. 13 (10) (1969) 2129e2147.
[29] S.M. Aharoni, Ductile and brittle behavior of amorphous polymers. Relation-
Supplementary data related to this article can be found at ship with activation energy for glass transition and mechanical fracture,
https://doi.org/10.1016/j.polymer.2018.04.017. J. Appl. Polym. Sci. 16 (12) (1972) 3275e3284.
[30] S.M. Aharoni, Correlations between chain parameters and failure character-
istics of polymers below their glass transition temperature, Macromolecules
References 18 (12) (1985) 2624e2630.
[31] J.L. Keddie, R.A.L. Jones, R.A. Cory, Size-Dependent depression of the glass-
[1] J.A. Thornton, D. Hoffman, Stress-related effects in thin films, Thin Solid Films transition temperature in polymer-films, Europhys. Lett. 27 (1) (1994) 59e64.
171 (1) (1989) 5e31. [32] J.A. Forrest, K. Dalnoki-Veress, J.R. Stevens, J.R. Dutcher, Effect of free surfaces
[2] G. Reiter, Dewetting of thin polymer films, Phys. Rev. Lett. 68 (1) (1992) on the glass transition temperature of thin polymer films, Phys. Rev. Lett. 77
75e78. (10) (1996) 2002e2005.
[3] F. Bruder, R. Brenn, Spinodal decomposition in thin films of a polymer blend, [33] J.A. Forrest, K. DalnokiVeress, J.R. Dutcher, Interface and chain confinement
Phys. Rev. Lett. 69 (4) (1992) 624. effects on the glass transition temperature of thin polymer films, Phys. Rev. E
[4] A. Richard, Interface and surface effects on the glass-transition temperature in 56 (5) (1997) 5705e5716.
thin polymer films, Faraday Discuss 98 (1994) 219e230. [34] Y. Wang, M. Rafailovich, J. Sokolov, D. Gersappe, T. Araki, Y. Zou, A. Kilcoyne,
[5] Y.-X. Liu, E.-Q. Chen, Polymer crystallization of ultrathin films on solid sub- H. Ade, G. Marom, A. Lustiger, Substrate effect on the melting temperature of
strates, Coord. Chem. Rev. 254 (9) (2010) 1011e1037. thin polyethylene films, Phys. Rev. Lett. 96 (2) (2006) 028303.
[6] J.N.L. Albert, T.H. Epps III, Self-assembly of block copolymer thin films, Mater. [35] I. Yang, C.Y. Liu, Real-time SAXS and WAXS study of the multiple melting
Today 13 (6) (2010) 24e33. behavior of poly (ε-caprolactone), J. Polym. Sci. B Polym. Phys. 48 (16) (2010)
[7] G.M. Kelly, J.N.L. Albert, Ultrathin film crystallization of poly (ε-caprolactone) 1777e1785.
in blends containing styrene-isoprene block copolymers: the nano-rose [36] H.H. Su, H.L. Chen, A. Diaz, M.T. Casas, J. Puiggali, J.N. Hoskins, S.M. Grayson,
morphology, Polymer 117 (2017) 295e305. R.A. Perez, A.J. Muller, New insights on the crystallization and melting of cyclic
[8] M. Park, C. Harrison, P.M. Chaikin, R.A. Register, D.H. Adamson, Block copol- PCL chains on the basis of a modified Thomson-Gibbs equation, Polymer 54
ymer lithography: periodic arrays of similar to 10(11) holes in 1 square (2) (2013) 846e859.
centimeter, Science 276 (5317) (1997) 1401e1404. [37] G. Jimenez, N. Ogata, H. Kawai, T. Ogihara, Structure and thermal/mechanical
[9] J.N.L. Albert, M.J. Baney, C.M. Stafford, J.Y. Kelly, T.H. Epps III, Generation of properties of poly(ε-caprolactone)-clay blend, J. Appl. Polym. Sci. 64 (11)
monolayer gradients in surface energy and surface chemistry for block (1997) 2211e2220.
copolymer thin film studies, ACS Nano 3 (12) (2009) 3977e3986. [38] M. Ma, Z. He, J. Yang, F. Chen, K. Wang, Q. Zhang, H. Deng, Q. Fu, Effect of film
[10] J.N.L. Albert, T.D. Bogart, R.L. Lewis, K.L. Beers, M.J. Fasolka, J.B. Hutchison, thickness on morphological evolution in dewetting and crystallization of
B.D. Vogt, T.H. Epps III, Gradient solvent vapor annealing of block copolymer polystyrene/poly(epsilon-caprolactone) blend films, Langmuir 27 (21) (2011)
thin films using a microfluidic mixing device, Nano Letters 11 (3) (2011) 13072e13081.
1351e1357. [39] M. Ma, Z. He, J. Yang, Q. Wang, F. Chen, K. Wang, Q. Zhang, H. Deng, Q. Fu,
[11] J.Y. Kelly, J.N. Albert, J.A. Howarter, C.M. Stafford, T.H. Epps III, M.J. Fasolka, Vertical phase separation and liquid-liquid dewetting of thin PS/PCL blend
Manipulating morphology and orientation in thermally responsive block films during spin coating, Langmuir 27 (3) (2011) 1056e1063.
copolymer thin films, J. Polym. Sci. B Polym. Phys. 50 (4) (2012) 263e271. [40] M. Ma, Z. He, Y. Li, F. Chen, K. Wang, Q. Zhang, H. Deng, Q. Fu, Surface phase
[12] B.H. Jeong, E.M.V. Hoek, Y.S. Yan, A. Subramani, X.F. Huang, G. Hurwitz, separation, dewetting feature size, and crystal morphology in thin films of
A.K. Ghosh, A. Jawor, Interfacial polymerization of thin film nanocomposites: a polystyrene/poly(epsilon-caprolactone) blend, J. Colloid Interface Sci. 387 (1)
new concept for reverse osmosis membranes, J. Membr. Sci. 294 (2007) 1e7. (2012) 262e269.
[13] B. Chakrabarty, A.K. Ghoshal, M.K. Purkait, Effect of molecular weight of PEG [41] Z. Qiu, T. Ikehara, T. Nishi, Miscibility and crystallization of poly (ethylene
on membrane morphology and transport properties, J. Membr. Sci. 309 (2008) oxide) and poly (ε-caprolactone) blends, Polymer 44 (10) (2003) 3101e3106.
209e221. [42] V.H. Mareau, R.E. Prud'Homme, In-situ hot stage atomic force microscopy
[14] T. Ioroi, K. Yasuda, Z. Siroma, N. Fujiwara, Y. Miyazaki, Thin film electrocatalyst study of poly (ε-caprolactone) crystal growth in ultrathin films, Macromole-
layer for unitized regenerative polymer electrolyte fuel cells, J. Power Sources cules 38 (2) (2005) 398e408.
112 (2) (2002) 583e587. [43] V.H. Mareau, R.E. Prud'homme, Crystallization of ultrathin poly(ε-capro-
[15] F.Q. Liu, B.L. Yi, D.M. Xing, J.R. Yu, H.M. Zhang, Nafion/PTFE composite mem- lactone) films in the presence of residual solvent, an in situ atomic force
branes for fuel cell applications, J. Membr. Sci. 212 (2003) 213e223. microscopy study, Polymer 46 (18) (2005) 7255e7265.
[16] K. Chopra, P. Paulson, V. Dutta, Thin-film solar cells: an overview, Prog. [44] K.L. Beers, J.F. Douglas, E.J. Amis, A. Karim, Combinatorial measurements of
Photovoltaics Res. Appl. 12 (2e3) (2004) 69e92. crystallization growth rate and morphology in thin films of isotactic poly-
[17] G. Li, Y. Yao, H. Yang, V. Shrotriya, G. Yang, Y. Yang, “Solvent annealing” effect styrene, Langmuir 19 (9) (2003) 3935e3940.
in polymer solar cells based on poly (3-hexylthiophene) and methanofuller- [45] Q. Zong, Z. Li, X.M. Xie, Inversion of phase morphology in polymer-blend thin
enes, Adv. Funct. Mater. 17 (10) (2007) 1636e1644. films on glass substrates, Macromol. Chem. Phys. 205 (8) (2004) 1116e1124.
[18] F.C. Chen, C.J. Ko, J.L. Wu, W.C. Chen, Morphological study of P3HT: PCBM [46] C.D. Qiao, J.C. Zhao, S.C. Jiang, X.L. Ji, L.J. An, B.Z. Jiang, Crystalline morphology
blend films prepared through solvent annealing for solar cell applications, Sol. evolution in PCL thin films, J. Polym. Sci. B Polym. Phys. 43 (11) (2005)
Energy Mater. Sol. Cell. 94 (12) (2010) 2426e2430. 1303e1309.
[19] Z. Xiao, Q. Dong, C. Bi, Y. Shao, Y. Yuan, J. Huang, Solvent annealing of [47] G. Reiter, G.R. Strobl, Progress in Understanding of Polymer Crystallization,
perovskite-induced crystal growth for photovoltaic-device efficiency Springer, 2007.
enhancement, Adv. Mater. 26 (37) (2014) 6503e6509. [48] E. Piorkowska, G.C. Rutledge, Handbook of Polymer Crystallization, John Wiley
[20] Y.-C. Tsai, S.-C. Li, J.-M. Chen, Cast thin film biosensor design based on a nafion & Sons, 2013.
backbone, a multiwalled carbon nanotube conduit, and a glucose oxidase [49] R. Fava, Differential scanning calorimetry of epoxy resins, Polymer 9 (1968)
function, Langmuir 21 (8) (2005) 3653e3658. 137e151.
[21] K.S. Tiaw, S.H. Teoh, R. Chen, M.H. Hong, Processing methods of ultrathin poly [50] M.-J. Richardson, N. Savill, Derivation of accurate glass transition tempera-
(ε-caprolactone) films for tissue engineering applications, Biomacromolecules tures by differential scanning calorimetry, Polymer 16 (10) (1975) 753e757.
8 (3) (2007) 807e816. [51] J. Hay, P. Mills, The use of differential scanning calorimetry to study polymer
[22] B.C. Berry, A.W. Bosse, J.F. Douglas, R.L. Jones, A. Karim, Orientational order in crystallization kinetics, Polymer 23 (9) (1982) 1380e1384.
342 G.M. Kelly et al. / Polymer 143 (2018) 336e342

[52] B. Fillon, J. Wittmann, B. Lotz, A. Thierry, Self-nucleation and recrystallization [69] M.Y. Efremov, A.V. Kiyanova, J. Last, S.S. Soofi, C. Thode, P.F. Nealey, Glass
of isotactic polypropylene (a phase) investigated by differential scanning transition in thin supported polystyrene films probed by temperature-
calorimetry, J. Polym. Sci. B Polym. Phys. 31 (10) (1993) 1383e1393. modulated ellipsometry in vacuum, Phys. Rev. E 86 (2 Pt 1) (2012) 021501.
[53] E. Freire, Differential scanning calorimetry, Protein Stability and Folding: [70] J.H. Kim, J. Jang, W.C. Zin, Thickness dependence of the melting temperature of
Theor. Pract. (1995) 191e218. thin polymer films, Macromol. Rapid Commun. 22 (6) (2001) 386e389.
[54] I. Okazaki, B. Wunderlich, Reversible melting in polymer crystals detected by [71] V.H. Mareau, R.E. Prud'Homme, Growth rates and morphologies of miscible
temperature-modulated differential scanning calorimetry, Macromolecules PCL/PVC blend thin and thick films, Macromolecules 36 (3) (2003) 675e684.
30 (6) (1997) 1758e1764. [72] M.A. Woodruff, D.W. Hutmacher, The return of a forgotten polymer-
[55] A. Zen, M. Saphiannikova, D. Neher, J. Grenzer, S. Grigorian, U. Pietsch, dpolycaprolactone in the 21st century, Prog. Polym. Sci. 35 (10) (2010)
U. Asawapirom, S. Janietz, U. Scherf, I. Lieberwirth, G. Wegner, Effect of mo- 1217e1256.
lecular weight on the structure and crystallinity of poly(3-hexylthiophene), [73] B.K. Money, J. Swenson, Dynamics of poly (ethylene oxide) around its melting
Macromolecules 39 (6) (2006) 2162e2171. temperature, Macromolecules 46 (17) (2013) 6949e6954.
[56] H.Z. Li, Y.J. Wu, H. Sato, L. Kong, C.F. Zhang, K. Huang, D.L. Tao, J. Chen, X.X. Liu, [74] H. Lu, W. Chen, T.P. Russell, Relaxation of thin films of polystyrene floating on
Y. Zhao, Y.Z. Xu, J.G. Wu, Y. Ozaki, A new facile method for preparation of ionic liquid surface, Macromolecules 42 (22) (2009) 9111e9117.
nylon-6 with high crystallinity and special morphology, Macromolecules 42 [75] M. Chapman, M. Mullen, E. Novoa-Ortega, M. Alhasani, J.F. Elman, W.B. Euler,
(4) (2009) 1175e1179. Structural evolution of ultrathin films of rhodamine 6G on glass, J. Phys. Chem.
[57] S. Feng, Z. Li, R. Liu, B. Mai, Q. Wu, G. Liang, H. Gao, F. Zhu, Glass transition of C 120 (15) (2016) 8289e8297.
polystyrene nanospheres under different confined environments in aqueous [76] Z. Jiang, X. Li, J. Strzalka, M. Sprung, T. Sun, A.R. Sandy, S. Narayanan, D.R. Lee,
dispersions, Soft Matter 9 (18) (2013) 4614e4620. J. Wang, The dedicated high-resolution grazing-incidence X-ray scattering
[58] H. Wang, C.-L. Liu, G. Wu, S.-C. Chen, F. Song, Y.-Z. Wang, Temperature beamline 8-ID-E at the Advanced Photon Source, J. Synchrotron Radiat. 19 (4)
dependent morphological evolution and the formation mechanism of aniso- (2012) 627e636.
tropic nano-aggregates from a crystalline-coil block copolymer of poly (p- [77] S.A. Vaselabadi, D. Shakarisaz, P. Ruchhoeft, J. Strzalka, G.E. Stein, Radiation
dioxanone) and poly (ethylene glycol), Soft Matter 9 (36) (2013) 8712e8722. damage in polymer films from grazing-incidence X-ray scattering measure-
[59] S. Feng, Y. Chen, B. Mai, W. Wei, C. Zheng, Q. Wu, G. Liang, H. Gao, F. Zhu, Glass ments, J. Polym. Sci. B Polym. Phys. 54 (11) (2016) 1074e1086.
transition of poly (methyl methacrylate) nanospheres in aqueous dispersion, [78] Z. Jiang, GIXSGUI: a MATLAB toolbox for grazing-incidence X-ray scattering
Phys. Chem. Chem. Phys. 16 (30) (2014) 15941e15947. data visualization and reduction, and indexing of buried three-dimensional
[60] S. Kunugi, T. Tada, N. Tanaka, K. Yamamoto, M. Akashi, Microcalorimetric periodic nanostructured films, J. Appl. Crystallogr. 48 (3) (2015) 917e926.
study of aqueous solution of a thermoresponsive polymer, poly (N-vinyl- [79] H. Hu, D.L. Dorset, Crystal structure of poly (ε-caprolactone), Macromolecules
isobutyramide)(PNVIBA), Polym. J. 34 (5) (2002) 383e388. 23 (21) (1990) 4604e4607.
[61] J. Balko, A. Rinscheid, A. Wurm, C. Schick, R.H. Lohwasser, M. Thelakkat, [80] N. Sanandaji, L. Ovaskainen, M.K. Gunnewiek, G.J. Vancso, M.S. Hedenqvist,
T. Thurn-Albrecht, Crystallinity of poly (3-hexylthiophene) in thin films S. Yu, L. Eriksson, S. Roth, U.W. Gedde, Unusual crystals of poly (ε-capro-
determined by fast scanning calorimetry, J. Polym. Sci. B Polym. Phys. 54 (18) lactone) by unusual crystallisation: the effects of rapid cooling and fast sol-
(2016) 1791e1801. vent loss on the morphology, crystal structure and melting, Polymer 54 (5)
[62] X. Tardif, B. Pignon, N. Boyard, J.W. Schmelzer, V. Sobotka, D. Delaunay, (2013) 1497e1503.
C. Schick, Experimental study of crystallization of PolyEtherEtherKetone [81] V. Loiko, P. Maksimenko, A. Konkolovich, A method for calculating the light
(PEEK) over a large temperature range using a nano-calorimeter, Polym. Test. extinction coefficients of a polymer layer with small nematic liquid-crystal
36 (2014) 10e19. droplets, Optic Spectrosc. 105 (5) (2008) 791e797.
[63] M.D. Ediger, J.A. Forrest, Dynamics near free surfaces and the glass transition [82] M.S. Vezie, S. Few, I. Meager, G. Pieridou, B. Do €rling, R.S. Ashraf, A.R. Gon
~ i,
in thin polymer films: a view to the future, Macromolecules 47 (2) (2014) H. Bronstein, I. McCulloch, S.C. Hayes, Exploring the origin of high optical
471e478. absorption in conjugated polymers, Nat. Mater. 15 (2016) 746e753.
[64] C.J. Ellison, R.L. Ruszkowski, N.J. Fredin, J.M. Torkelson, Dramatic reduction of [83] A. Priyadarshi, L. Shimin, E. Wong, R. Rajoo, S. Mhaisalkar, V. Kripesh,
the effect of nanoconfinement on the glass transition of polymer films via Refractive indices variation with temperature and humidity of optical adhe-
addition of small-molecule diluent, Phys. Rev. Lett. 92 (9) (2004) 095702. sive, J. Electron. Mater. 34 (11) (2005) 1378e1384.
[65] C.B. Roth, K.L. McNerny, W.F. Jager, J.M. Torkelson, Eliminating the enhanced [84] R. Moshrefzadeh, M. Radcliffe, T. Lee, S. Mohapatra, Temperature dependence
mobility at the free surface of polystyrene: fluorescence studies of the glass of index of refraction of polymeric waveguides, J. Lightwave Technol. 10 (4)
transition temperature in thin bilayer films of immiscible polymers, Macro- (1992) 420e425.
molecules 40 (7) (2007) 2568e2574. [85] J.D. Hoffman, J.J. Weeks, Specific volume and degree of crystallinity of semi-
[66] S. Kim, S. Hewlett, C. Roth, J. Torkelson, Confinement effects on glass transition crystalline polychlorotrifluoroethylene, and estimated specific volumes of the
temperature, transition breadth, and expansivity: comparison of ellipsometry pure amorphous and crystalline phases, J. Research Natl. Bur. Standards 60
and fluorescence measurements on polystyrene films, The European Physical (1958) 465.
Journal E: Soft Matter and Biological Physics 30 (1) (2009) 83e92. [86] A. Miyagi, B. Wunderlich, Superheating and reorganization on melting of poly
[67] L. Zhang, R. Elupula, S.M. Grayson, J.M. Torkelson, Major impact of cyclic chain (ethylene terephthalate), J. Polym. Sci. B Polym. Phys. 10 (7) (1972)
topology on the T g-confinement effect of supported thin films of polystyrene, 1401e1405.
Macromolecules 49 (1) (2015) 257e268. [87] M.E. Co rdova, A.T. Lorenzo, A.J. Muller, J.N. Hoskins, S.M. Grayson,
[68] L. Zhang, R. Elupula, S.M. Grayson, J.M. Torkelson, Suppression of the fragility- A comparative study on the crystallization behavior of analogous linear and
confinement effect via low molecular weight cyclic or ring polymer topology, cyclic poly(caprolactones), Macromolecules 44 (7) (2011) 1742e1746.
Macromolecules 50 (3) (2017) 1147e1154.