You are on page 1of 7

Article

pubs.acs.org/crystal

Geometric Design of Heterogeneous Nucleation Sites on


Biocompatible Surfaces
Vilmalí López-Mejías, Allan S. Myerson, and Bernhardt L. Trout*
Department of Chemical Engineering, Massachusetts Institute of Technology, 50 Ames Street, Cambridge, Massachusetts 02139,
United States
*
S Supporting Information

ABSTRACT: Biocompatible polymer surfaces imprinted with nanopores of


See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

various geometries were utilized as heteronucleants during crystallizations of


Downloaded via UNIV FED DE SANTA CATARINA on August 8, 2022 at 17:14:08 (UTC).

mefenamic acid (MA). MA, a nonsteroidal anti-inflammatory drug (NSAID) that


possesses two structurally characterized polymorphs, is utilized as a model
compound. By combining geometric nanoconfinement and favorable surface−
solute interactions, it was possible to influence the nucleation kinetics of this system
and harvest the metastable form of MA, MA form II, on the square nanopores. An
exploration of the relationship between the preferred orientation found by powder
X-ray diffraction (PXRD) and the angular matching based on the intrinsic angles,
determined by the predicted morphology, was used to provide insights into the
mechanism for the observed nucleation enhancement and polymorph selection.
The results presented here might lead to a rational design strategy of surfaces to
control nucleation and, therefore, polymorphism.

T here are many challenges that arise during the


manufacturing of crystalline materials; one of these is
controlling the product quality. Since crystallization is often
rational design of a surface to control nuclei formation and,
therefore, polymorphism. By combining geometric nano-
confinement and favorable surface-solute interactions, it should
used as the last step in the production of the final product as a be possible to influence not only nucleation but also crystal size
solid form, it is of considerable importance to design distribution, crystallization yield, and morphology.
crystallization processes that control not only purity but other In the present investigation, a range of nanopore geometries
properties of the solid, such as size distribution, morphology, were employed as heteronucleants, choosing pore geometries
and polymorphism, to ensure the product quality and therefore which contain angles that were encountered most often among
its performance. The latter task is complicated by the fact that the major faces of the calculated morphology in structurally
there are many compounds exhibiting multiple crystalline similar compounds. The geometries explored here are
phases that differ in the arrangement and/or conformations of imprinted on a biocompatible polymer surface, presenting a
molecules within a crystal lattice. In recent years, hetero- high degree of hydrophilicity. This strategy aims to facilitate
nucleation, the process by which molecules aggregate to form a nuclei formation by providing favorable molecular interactions,
crystal nuclei at a solid−liquid interface, has emerged as a as well as angular matching between the surface and the
valuable method to control nucleation and therefore poly- crystallizing molecules in solution. By providing kinetic
morphism. However, the current fundamental understanding of resolution at the corners of the nucleating nanopore, it might
this process does not provide a rational design of hetero- be possible to provide access to specific polymorphs of a given
nucleants. Previous reports have implied the role of the surface compound if the crystal structures are known. Similarly, it
in liquid−solid phase transformations, as well as the impact of might be possible to produce a never observed solid form for a
the surface morphology on the crystallization outcome.1−5
compound by screening different angles or shapes, choosing
Recent investigations have shown that crystal nucleation under
either from a general library or from targeted guesses based on
porous confinement can alter nucleation kinetics, phase
structural resemblance. The first possibility is explored in this
stabilization, polymorphism, and crystal orientation.6−17 On
smooth polymer surfaces, polymer-induced heteronucleation investigation, and the second one would be the focus of future
has allowed access to an unprecedented number of structurally experiments.
characterized polymorphs based on complementary interac- Mefenamic acid (MA, Figure 1) is a potent nonsteroidal anti-
tions at the polymer−crystal interface.18−21 At the computa- inflammatory drug (NSAID), which possesses two structurally
tional level, there have been studies which relate crystal characterized polymorphs by single crystal X-ray diffraction.
nucleation to the geometry of the surface.16 Moreover, at the Elucidation of the first crystal structure (form I) occurred in
experimental level, the presence of a nanoimprinted polymer
surface has been demonstrated to alter nucleation kinetics in a Received: June 28, 2013
solution.22 These investigations provided the first insights into a Published: July 3, 2013

© 2013 American Chemical Society 3835 dx.doi.org/10.1021/cg400975b | Cryst. Growth Des. 2013, 13, 3835−3841
Crystal Growth & Design Article

Figure 1. Molecular structures of NSAIDs mefenamic acid (MA), tolfenamic acid (TA), flufenamic acid (FFA), and other structural analogues
employed in the morphology prediction calculation and determination of the most common intrinsic angles among these structurally similar
molecules. The numbers of structurally characterized forms for each molecular structure presenting polymorphism are in parentheses.

1976,23 and nearly thirty years later the structure of a second


metastable form (form II) showing disorder was reported.24
The study of the crystallization behavior in this compound is of
particular interest since many members of the femanic acid
series present polymorphism, including flufenamic acid (FFA,
Figure 1), the most polymorphic compound known with eight
structurally characterized forms,21 and tolfenamic acid (TA,
Figure 1),18 another highly polymorphic compound that
possesses five structurally characterized forms.
In order to design the nanopore geometries investigated
here, the predicted morphology based on the crystal structures
of MA, its analogues, and polymorphs thereof was calculated
using BFDH prediction methods.25 In this investigation, we
have assumed that structurally similar molecules will give rise to
isostructural or structurally similar crystal forms, therefore the Figure 2. Histogram of number count of intrinsic angles (deg) in MA
intrinsic angles among these should be similar. The single and its structural analogues that fall within a ± 5° range.
crystals structures were retrieved from the Cambridge
Crystallographic Database (CSD). The reference codes for hypothesize that for a given crystal, some of the major faces for
the structures used in the morphology prediction can be found the predicted morphology will be manifested, even as a minor
in the Supporting Information. The major facets of these face experimentally. Therefore, even a coarse morphology
predicted morphologies were identified and used in the prediction method such as BFDH can be used to determine the
determination of the intrinsic angle between (adjacent) major most common intrinsic angles in a predictive way and guide the
crystal facets. It was observed that 60°, 70°, 80°, and 90° were design of the geometric nanopores.
the most commonly observed intrinsic angles among these In order to imprint nanopores on a surface containing the
structurally similar compounds (Figure 2 and the Supporting 60° and 80° angles, it was necessary to form parallelograms
Information). Given the assumptions of BFDH, the predicted with complementary angles of 120° and 100°, respectively.
angles are only at best a guide and should not be taken as Since these complementary angles were not observed in the
quantitative predictions of actual experimental morphologies. investigation of the most commonly observed intrinsic angles
Nevertheless, the intrinsic angles among major faces (Figure 2 and the Supporting Information), they should have
determined via BFDH prediction methods, will not vary minimal effect on the nucleation kinetics of this system. Three
significantly based on morphology. As long as at least two of different nanopatterns were developed for this investigation:
the adjacent major facets appear, even as minor faces, these the type A pattern containing nanopores with 60° and 120°
angles will be present in the observed morphology. We angles, the type B pattern containing nanopores with 80° and
3836 dx.doi.org/10.1021/cg400975b | Cryst. Growth Des. 2013, 13, 3835−3841
Crystal Growth & Design Article

Figure 3. AFM images of (a) type A, (b) type B, (c) type C, and (d) spherical patterns imprinted on hydroxylpropylmethyl cellulose (HPMC), as
well as (e) a smooth HPMC surface.

Figure 4. AFM images of type C (square nanopores) on a HPMC surface: (a) 5 × 5 μm area prior to exposure to crystallization solvent and heat,
(b) 1 × 1 μm and (c) 5 × 5 μm areas after exposure to anisole at 65 °C in the absence of solute, and (d) 1 × 1 μm and (e) 5 × 5 μm areas
containing MA form II crystals deposited by cooling crystallization method using anisole as a solvent.

100° angles, and the type C pattern containing nanopores of The nanopatterned surfaces were then imprinted onto a
only 90° angles (Figure 3, panels a−c). For each pattern, a 4” biocompatible polymer film by solvent casting a 10% (w/w)
silicon master comprised of pillars in the shape of the solution of hydroxylpropylmethyl cellulose (HPMC) in a 1:1
parallelogram was developed through interference lithography mixture of EtOH:H2O on top of the mold. Since it was
(made in collaboration with the Nanostructures Laboratory at observed that the majority of the faces covering the largest
MIT). A silicon master containing angles of 70° and 110° was percent (%) area in the predicted morphology presented
not fabricated. The fabrication of the silicon masters is time- hydrogen bonding opportunities, this highly hydrophilic
consuming and expensive. At the onset of this investigation, it polymer was chosen for the study (Supporting Information).
was decided to work only with three masters. Silicon masters The films were peeled off the mold and utilized as
containing the 60°, 80°, and 90° angles were chosen because heteronucleants after drying at 70 °C for 2 h in a vacuum
oven. The smooth polymer surface was developed by following
we hypothesize that these were the most effective for our
a similar solvent casting and drying procedure in the presence
system of interest, MA, according to our preliminary angle-
of an unpatterned 4” silicon wafer.
matching analysis. An additional control surface with spherical Biocompatible organic solvents were screened in order to
pores of 100 nm was produced by patterning a 2” quartz disk find an appropriate crystallization solvent. Ethanol, acetonitrile,
(Figure 3d). The disk was treated with O2 plasma to enrich the ethyl acetate, heptane, anisole, and cumene were tested. When
surface with hydroxyl groups, then 200 μL of a 5% (w/w) heated to 65 °C in the absence of solute, only films exposed to
colloidal silica particles in water were spread and allowed to anisole and cumene retained the imprinted pattern on the
self-assemble by slow evaporation. The self-assembled SiO2 and surface as confirmed by AFM imaging (Figure 4, panels a−c
the quartz slide were then sintered at 800 °C in a high and the Supporting Information). The solubility at equilibrium
temperature oven. The absence of nanopores (a smooth of MA in both anisole and cumene was determined at three
surface, Figure 3e) and the absence of a polymer surface were different temperatures (30, 45, and 65 °C). The solubility of
also tested as controls. MA in anisole was determined to be 4.6 mg/mL at 30 °C, 5.8
3837 dx.doi.org/10.1021/cg400975b | Cryst. Growth Des. 2013, 13, 3835−3841
Crystal Growth & Design Article

mg/mL at 45 °C, and 29.5 mg/mL at 65 °C. The solubility of induction time when compared to the results observed in the
MA in cumene was determined to be 0.6 mg/mL at 30 °C, absence of the polymer. Additionally, a higher percentage of
which is not high enough to proceed with experiments using vials crystallized in the presence of a polymer surface (95−
this solvent. Therefore, only anisole was utilized as a solvent in 65%), with the exception of the spherical pattern, in which the
this investigation because it does not dissolve the patterns number of vials crystallized as a function of time is slightly less
imprinted on these films and provides reasonable solubility (40%) than in the absence of a polymer (55%) (Supporting
under the experimental conditions utilized. A supersaturation of Information).
1.5× was determined to be the optimal supersaturation level in The surface of each polymer film after crystallization
order to minimize homogeneous/bulk nucleation while occurred was studied through atomic force microscopy
maintaining reasonable induction times. It is at this low (AFM). It was observed that on the surfaces patterned with
supersaturation that heterogeneous nucleation can be truly geometrical nanopores, the pores contained a significant
explored; at higher supersaturation homogeneous/bulk nucle- number of the crystals that nucleated on the surface (Figure
ation dominates the crystallization events with little effect from 4, panels d−e, and the Supporting Information). Specifically, it
the surface. At this supersaturation level, relatively large spatial was observed that the crystals mostly form at the corners of the
fluctuations are likely needed for nucleation to occur. It was nanopores (Figure 4, panels d−e, and the Supporting
assumed that these fluctuations occur at a scale larger than the Information). Generally, for the surfaces used as controls, the
>5 nm radius of curvature resulting after the polymer spherical nanopores and the smooth polymer surfaces, the
imprinting process. A 7 mg/mL solution of MA in anisole number of crystals covering the examined area were fewer than
was delivered (750 μL) into 78 (1 mL) vials and crashed cooled those of the experimental nanopatterned surfaces.
from 65 to 30 °C to achieve a supersaturation of 1.5×. Further examination of each polymer surface was performed
The crystal nucleation induction time for the vials was using powder X-ray diffraction (PXRD). Diffractograms for
monitored continuously with an inverted microscope in order each of the surfaces examined during this investigation present
to determine the average induction time of nucleation. Six different and highly discernible preferred orientation (PO); this
different samples were prepared: bulk samples without polymer is indicative of different intermolecular interactions occurring at
films, samples with polymer films imprinted with types A−C, the polymer−crystal interface during nucleation (Figure 5). A
and spheres, as well as polymer films without any pattern. For high density of crystals aligned along a specific crystallographic
each type of samples, 13 vials were prepared and loaded onto plane (or face) will produce a high intensity peak or a preferred
the temperature controlled block. The heating and cooling orientation peak in the resulting PXRD, which indicates that
procedures were repeated to provide statistically significant data the majority of the molecules aggregate and further nucleate
for 78 samples. The samples were crash-cooled from 65 to 30 from this specific face and allow for Bragg’s Law to be satisfied.
°C to achieve a 1.5× supersaturated solution. The onset of each When crystals that grew in the absence of a surface were
crystallization event was detected optically; when the first examined through PXRD, the resulting diffractogram presented
crystal was observed, the frame number was recorded. Frames only a small degree of PO along the (100) face of MA form I
were taken at 5 min intervals. The average nucleation induction (Figure 5a). For the type A pattern, only one single major
time, τ, was determined from a statistical analysis of the reflection was observed, corresponding to crystals of MA form I
induction time data based on the knowledge that nucleation nucleating along the (100) face with a high degree of PO
follows a Poisson distribution, P(t) = e(‑t/τ), where P is the (Figure 5b). For the type B pattern, two major reflections were
probability that no nucleation event occurs within time, t2. The observed, corresponding also to MA form I crystals nucleating
total number of vials as a function of time was also monitored from the (100) and the (02-2) faces (Figure 5c). When the
during the experiment. type C pattern was utilized during this crystallization
Nucleation induction time experiments revealed that at the experiment, the presence of MA form II was established by
same supersaturation level, the presence of type C or square examination of the diffractogram, which indicated the presence
nanopores are able to enhance the nucleation kinetics 5× faster of three major reflections corresponding to PO along the (01-
than in the absence of the polymer (Table 1 and the 1), (02-1), and (012) faces (Figure 5d). On the other hand,
when the crystals nucleated in the absence of nanopores, the
Table 1. Summary of Average Nucleation Induction Times crystals examined display a high degree of PO along the slow
for MA Crystals Nucleated on the Various Nano-Imprinted growing (100) face of form I MA (Figure 5b). The molecular
Polymer Surfaces arrangement for all of the PO faces found by PXRD when the
heteronucleant type induction time (τ, h) error (h) linearity (h)
different nanopores were utilized as heteronucleants is shown
in Figure 6. Examination of these PO faces indicate that each of
type A (60° and 120°) 9.6 0.1 0.99
these faces provides hydrogen bonding between the hetero-
type B (80° and 100°) 13.6 0.3 0.98
nucleant and the molecules in contact with the PO face. This
type C (90°) 8.8 0.1 0.99
finding supports the notion that favorable surface−solute
no polymer 42.7 0.8 0.99
interactions are required to effectively enhance the nucleation
no pattern 15.8 0.3 0.96
induction time.
spheres 36.5 0.7 0.95
The dimensions of the crystals nucleated on the nanopores
are on the nanometer range, therefore, the use of optical
Supporting Information). The patterns of types A and B microscopy (hot-stage microscopy) or differential scanning
provide nucleation induction enhancement but to a lesser calorimetry (DSC) to determine phase purity were unsuccess-
extent when compared to the type C pattern. An even lesser ful, due to the limit of detection of these techniques. When the
effect was observed for the samples in which crystals grew in samples were examined by Raman spectroscopy, three major
the absence of a pattern (smooth surface). The presence of regions (1605−1570, 1250−1035, and 1000−770 cm−1) aided
spherical pores seems to have minimal effect on the average the identification of the polymorphs (see the Supporting
3838 dx.doi.org/10.1021/cg400975b | Cryst. Growth Des. 2013, 13, 3835−3841
Crystal Growth & Design Article

Figure 5. PXRD of MA crystals nucleated in the presence and absence of a heteronucleant. (a) Crystals that grew in the absence of a surface do not
express a high degree of preferred orientation (black ●) relative to the simulated PXRD of MA form I (red star). (b) MA form I crystals grown on a
type A nanopore and smooth surface (both represented by the PXRD under the black ●) align along the (100) face with a high degree of preferred
orientation. (c) MA form I crystals grown on type B nanopore (black ●) align along the (100) and the (02-2) faces with a high degree of preferred
orientation, and (d) MA form II crystals selectively grown on a type C nanopore (dark circle) align along the (01-1), (02-1), and (012) faces with a
high degree of preferred orientation relative to the simulated PXRD of MA form II (orange hexagon).

Figure 6. Observed molecular arrangement of MA crystals as determined by the PO faces on the diffractogram for each heteronucleant type. (a) MA
form I crystals aligned along the (100) face, possibly forming hydrogen bonds with the HPMC polymer surface in both the absence of pores (a
smooth surface) and the presence of the type A pattern. (b) MA form I crystals aligned along the (100) and (02-2) faces possibly forming hydrogen
bonds with the HPMC polymer surface imprinted with the type B pattern. (c) MA form II crystals aligned along the (01-1), (02-1), and (012) faces,
possibly forming hydrogen bonds with the HPMC polymer surface imprinted with the type C pattern.

Information). Crystals that grew in the absence of a polymer, in types A and B correspond to those of the thermodynamically
the presence of a smooth polymer surface, a spherical polymer stable form, MA form I, whereas crystals that grew in the
surface, or an imprinted polymer surface with the patterns of presence of an imprinted polymer surface with the type C
3839 dx.doi.org/10.1021/cg400975b | Cryst. Growth Des. 2013, 13, 3835−3841
Crystal Growth & Design Article

Figure 7. Illustrations of the relationship between the PO face found by PXRD and the angular matching based on the angles present on the (a−c)
type A, (d) type B, and (e−f) type C patterns.

pattern correspond to those of the metastable form, MA form absence of a heteronucleant. For the type B pattern, there is just
II. It is worth mentioning that the quality of the Raman spectra one possible combination of faces laying almost perpendicular
is not good, due to the presence of the polymer film and the to the PO face, the (02-2) face, and containing an angle that
size of the nanocrystals. Nevertheless, unique Raman shifts closely matches that of the nanopore imprinted on this surface.
corresponding to either MA form I or MA form II are observed, This combination is that of the (100) face and the (0-10) face
in addition to large and broad Raman shifts that correspond to which forms an intrinsic angle of 80.4° (Figure 7d). A second
the polymer heteronucleant. The determination of the phase PO face is presented in the diffractogram of MA crystals
purity by Raman spectroscopy rules out the possibility that nucleated on the type B pattern (Figure 5c). This corresponds
some crystals might be oriented in directions that do not to the (100) face, which is one of the slow-growing faces in MA
diffract (aligned along space-group defined systematic absen- form I (49.7% of the total facet area in the predicted
ces) and supports our observation that only polymers morphology); therefore, it is possible that the crystals might
imprinted with the type C pattern provide phase selectivity present this orientation regardless of the imprinted nanopore
for MA form II, in contrast to the other crystallization present on the surface. This might also explain the greater
conditions used in this investigation, which only provide access number of possible combinations of faces that would lead to
to the MA form I. the PO along the (100) face when a type A pattern is utilized.
In order to determine the role of the pores during nucleation For this reason, only the combination between the (100) face
and phase selection of MA crystal forms, a possible relationship and the (0-10) face providing an intrinsic angle of 80.4° should
between the PO found by PXRD and the angular matching be responsible for the decrease in the nucleation induction time
based on the intrinsic angles determined by the predicted observed when the type B pattern is used as a heteronucleant.
morphology was investigated. The PO faces found by PXRD For the type C pattern, there are two possible combinations of
were modeled, and faces that were perpendicular to the PO face faces laying almost perpendicular to the PO faces observed in
were determined from these; faces that formed intrinsic angles the diffractogram of MA form II and containing an angle that
that match the nanopore geometry utilized during the matches that of the square nanopore (Figure 7, panels e−f).
crystallization experiment were identified. To facilitate the The first combination involves PO along the (02-1) face to
search, only those faces that contributed significant area to the which the (001) and the (100) faces are almost perpendicular
predicted morphology were examined. For the type A pattern, and make an intrinsic angle of 84.5° (Figure 7e). The second
there are three possible combinations of faces laying almost combination presents PO along the (012) face to which the
perpendicular to the PO face, the (100) face, and containing an (01-1) and the (100) faces are almost perpendicular and make
angle that closely matches that of the nanopore imprinted on an intrinsic angle of 83.6° (Figure 7f). The third PO face
this surface (Figure 7, panels a−c). The first intrinsic angle of present in the diffractogram of MA form II crystals grown on a
56.0° is formed between the (0-11) face and the (0-10) face type C pattern, the (01−1) face, is one of the slow-growing
(Figure 7a). The second intrinsic angle of 56.1° is formed faces in the predicted morphology (13.4% of the total facet
between the (010) face and the (01-1) face (Figure 7b). The area). Therefore, it is also possible that MA form II crystals
third intrinsic angle of 63.2° is formed between the (010) face might present this orientation regardless of the imprinted
and the (−101) face (Figure 7c). The angular matching nanopore present on the surface. This would explain the
between the nanopore and the intrinsic angle explain the absence of an intrinsic angle matching this PO face. Although
decrease in the nucleation induction time when this type of these possible combinations do not provide an intrinsic angle
nanopore is utilized relative to crystallizations happening in the that exactly matches the 90° angle present in the square
3840 dx.doi.org/10.1021/cg400975b | Cryst. Growth Des. 2013, 13, 3835−3841
Crystal Growth & Design Article

nanopores imprinted on this surface, they provide a close (7) Yu, L. Acc. Chem. Res. 2010, 43, 1257.
enough match to the pore geometry. Moreover, the faster (8) Beiner, M.; Rengarajan, G. T.; Pankaj, S.; Enke, D.; Steinhart, M.
crystallization kinetics provided by the type C pattern when Nano Lett. 2007, 7, 1381.
compared to any of the other crystallization conditions might (9) Rengarajan, G. T.; Enke, D.; Steinhart, M.; Beiner, M. J. Mater.
Chem. 2008, 18, 2537.
also be responsible for the appearance of the metastable form, (10) Diao, Y.; Helgeson, M. E.; Myerson, A. S.; Hatton, T. A.; Doyle,
form II, of MA. P. S.; Trout, B. L. J. Am. Chem. Soc. 2011, 133, 3756.
The use of geometrically imprinted biocompatible surfaces as (11) Brás, A. R.; Merino, E. G.; Neves, P. D.; Fonseca, I. M.;
heteronucleants has successfully established that nanopores can Dionísio, M.; Schönhals, A.; Correia, N. T. J. Phys. Chem. C 2011, 115,
alter the nucleation kinetics, depending on the angle. It was 4616.
observed that in the presence of type C or square nanopores, (12) Hamilton, B. D.; Ha, J.-M.; Hillmyer, M. A.; Ward, M. D. Acc.
nucleation occurred faster than during any of the other Chem. Res. 2012, 45, 414.
crystallization conditions utilized in this investigation, possibly (13) Hamilton, B. D.; Hillmyer, M. A.; Ward, M. D. Cryst. Growth
due to the angle matching between the intrinsic angle present Des. 2008, 8, 3368.
in the growing crystals and the angle at the corners of the (14) Ha, J.-M.; Hamilton, B. D.; Hillmyer, M. A.; Ward, M. D. Cryst.
Growth Des. 2009, 9, 4766.
nanopore. Furthermore, the appearance of the metastable form,
(15) Hamilton, B. D.; Wassbuch, I.; Lahav, M.; Hillmyer, M. A.;
MA form II, was observed only in the presence of the square Ward, M. D. J. Am. Chem. Soc. 2009, 131, 2588.
nanopores, which indicated that, under this crystallization (16) Page, J. A.; Sear, P. R. J. Am. Chem. Soc. 2009, 131, 17550.
condition, the type C pore is able to provide kinetic control (17) Rengarajan, G. T.; Enke, D.; Beiner, M. Open Phys. Chem. J.
over the formation of this polymorph versus the thermody- 2007, 1, 18.
namically more stable form, MA form I. Additionally, all of the (18) López-Mejı ́as, V.; Kampf, W. J.; Matzger, A. J. J. Am. Chem. Soc.
PO faces determined by PXRD show hydrogen-bonding 2009, 131, 4554.
interactions demonstrating that favorable surface−solute (19) McClelland, A. A.; López-Mejias, V.; Matzger, A. J.; Chen, Z.
interactions are required to promote crystal nucleation. Finally, Langmuir 2011, 27, 2162.
this investigation revealed that a relationship exists between the (20) López-Mejı ́as, V.; Knight, J. L.; Brooks, C. L., III; Matzger, A. J.
Langmuir 2011, 27, 7575.
pore angle and the PO faces found in the crystals grown on this (21) López-Mejı ́as, V.; Kampf, W. J.; Matzger, A. J. J. Am. Chem. Soc.
nanopore; this relationship might be responsible for affecting 2012, 134, 9872.
the nucleation kinetics and the phase selection through the (22) Diao, Y.; Harada, T.; Myerson, A. S.; Hatton, T. A.; Trout, B. L.
alteration of the orientation order at the nanometer scale in the Nat. Mater. 2011, 10, 867.
corners of the nanoindentation. (23) McConnell, J. F.; Company, F. Z. Cryst. Struct. Commun. 1976,


*
ASSOCIATED CONTENT
S Supporting Information
5, 861.
(24) Lee, E. H.; Byrn, S. R.; Carvajal, T. M. Pharm. Res. 2006, 23,
2375.
(25) Donnay, J. D. H.; Harker, D. Am. Mineral. 1937, 22, 446.
Experimental details, morphology prediction, nucleation
induction time measurement, Raman spectroscopy, powder
X-ray diffraction analysis, atomic force microscopy images, and
angular-matching analysis. This material is available free of
charge via the Internet at http://pubs.acs.org.

■ AUTHOR INFORMATION
Corresponding Author
*E-mail: trout@mit.edu.
Notes
The authors declare no competing financial interest.

■ ACKNOWLEDGMENTS
We thank the Novartis-MIT Center for Continuous Manu-
facturing for the continuous funding of the research presented
here. We also thank Dow Chemical Company and Dow Wolff
Cellulosic for kindly providing Methocel LV free of charge and,
finally, Dr. Tim Savas and the Nanostructures Laboratory at
MIT for the fabrication of the types A−C silicon masters.

■ REFERENCES
(1) Ward, M. D. Chem. Rev. 2001, 101, 1697.
(2) Diao, Y.; Myerson, A. S.; Hatton, T. A.; Trout, B. L. Langmuir
2011, 27, 5324.
(3) Cacciuto, A.; Auer, S.; Frenkel, D. Nature 2004, 428, 404.
(4) Holbrough, J. L.; Campbell, J. M.; Meldrum, F. C.; Christenson,
H. K. Cryst. Growth Des. 2012, 12, 750.
(5) Ha, J. M.; Hillmyer, M. A.; Ward, M. D. J. Phys. Chem. B 2005,
109, 1392.
(6) Ha, J. M.; Wolf, J. H.; Hillmyer, M. A.; Ward, M. D. J. Am. Chem.
Soc. 2004, 126, 3382.

3841 dx.doi.org/10.1021/cg400975b | Cryst. Growth Des. 2013, 13, 3835−3841

You might also like