Linking CO2 Sorption Performance to Polymer Morphology in Aminopolymer/Silica Composites through Neutron Scattering

Article abstract of DOI:10.1021/jacs.5b06823

Composites of poly(ethylenimine) (PEI) and mesoporous silica are effective, reversible adsorbents for CO₂, both from flue gas and in direct air-capture applications. The morphology of the PEI within the silica can strongly impact the overall carbon capture efficiency and rate of saturation. Here, we directly probe the spatial distribution of the supported polymer through small-angle neutron scattering (SANS). Combined with textural characterization from physisorption analysis, the data indicate that PEI first forms a thin conformal coating on the pore walls, but all additional polymer aggregates into plug(s) that grow along the pore axis. This model is consistent with observed trends in amine-efficiency (CO₂/N binding ratio) and pore size distributions, and points to a trade-off between achieving high chemical accessibility of the amine binding sites, which are inaccessible when they strongly interact with the silica, and high accessibility for mass transport, which can be hampered by diffusion through PEI plugs. We illustrate this design principle by demonstrating higher CO₂ capacity and uptake rate for PEI supported in a hydrophobically modified silica, which exhibits repulsive interactions with the PEI, freeing up binding sites.

Full text of DOI:10.1021/jacs.5b06823

This is an open access article published under an ACS AuthorChoice License, which permits
copying and redistribution of the article or any adaptations for non-commercial purposes.
Article
pubs.acs.org/JACS
Linking CO Sorption Performance to Polymer Morphology in
2
Aminopolymer/Silica Composites through Neutron Scattering
Adam Holewinski, Miles A. Sakwa-Novak, and Christopher W. Jones*
School of Chemical & Biomolecular Engineering, Georgia Institute of Technology, 311 Ferst Drive, Atlanta, Georgia 30332, United
States
*
S Supporting Information
ABSTRACT: Composites of poly(ethylenimine) (PEI) and
mesoporous silica are effective, reversible adsorbents for CO2,
both from flue gas and in direct air-capture applications. The
morphology of the PEI within the silica can strongly impact
the overall carbon capture efficiency and rate of saturation.
Here, we directly probe the spatial distribution of the
supported polymer through small-angle neutron scattering
(
SANS). Combined with textural characterization from
physisorption analysis, the data indicate that PEI first forms
a thin conformal coating on the pore walls, but all additional polymer aggregates into plug(s) that grow along the pore axis. This
model is consistent with observed trends in amine-efficiency (CO /N binding ratio) and pore size distributions, and points to a
2
trade-off between achieving high chemical accessibility of the amine binding sites, which are inaccessible when they strongly
interact with the silica, and high accessibility for mass transport, which can be hampered by diffusion through PEI plugs. We
illustrate this design principle by demonstrating higher CO capacity and uptake rate for PEI supported in a hydrophobically
2
modified silica, which exhibits repulsive interactions with the PEI, freeing up binding sites.
INTRODUCTION
loadings. Thus, the utilization of amine sites increases as the
pores are filled to higher fractions.
A maximum in the amine efficiency is set by the reaction
■
12,20,21,31
Supported amines are a valuable class of materials for numerous
1
−3
4−11
catalysis and separation applications.
Most notably, they
12−15
chemistry between amines and CO , where the primary
2
are effective in reversible CO adsorption.
consensus on the link between anthropogenic CO and climate
change continues to solidify, these materials are increasingly
being incorporated into scalable carbon capture processes.
Mesoporous silica materials are versatile supports for amine
moieties and can be synthesized with optimal pore sizes to
balance gas transport and amine dispersion. A particularly
useful model silica system is SBA-15, as it possesses well-
ordered and monodisperse pores, arranged as a hexagonal
As scientific
2
product is believed to be alkylammonium carbamate ion
pairs, formed through a zwitterionic mechanism. This chemistry
2
32
16,17
is well established in liquid solutions, and it has been
33−36
extrapolated to solid systems with spectroscopic support.
Under anhydrous conditions, two amines are required in the
immobilization of one CO , setting the amine efficiency limit at
2
one-half. Sorbents also usually exhibit a plateau in capacity with
respect to PEI loading because the diffusivity of CO through
2
18,19
the polymer slows dramatically during the course of
lattice of parallel cylinders (depicted in Figure 1).
In studies
37,38
39,40
saturation,
likely due to CO -induced cross-linking.
2
using well-defined mesoporous oxide supports, the role of
support pore size, composition and particle size on the
adsorption capacity and efficiency have been studied in
Ultimately, it is expected that the adsorption capacity and
kinetics will depend heavily on the morphology of the polymer,
i.e., whether it prefers to coat the pore walls or form
agglomerates, and/or the extent to which each occurs. Thus,
an informed performance optimization requires a detailed
understanding of the interrelation between polymer and
support morphologies. Despite this critical need, there are no
published reports that differentiate and directly probe the
morphology of the impregnated polymer phase in amine/oxide
2
0−30
detail.
Amino-polymers, most often poly(ethylenimine) (PEI), are
the most studied and arguably the most effective amine-
containing additives in these systems, as they have low volatility
and possess a high density of amine sites for CO adsorption.
As would intuitively be expected, the CO capacity (mmol
2
2
CO /g sorbent) of amino-polymer/silica composites generally
2
composite CO adsorbents.
increases with polymer loading. This relationship is shown in
Figure 1 for a typical series of PEI/SBA-15 composites
2
In the present contribution, we use small angle neutron
scattering (SANS) to probe the morphology of PEI in the pore
adsorbing CO from simulated dry air. The figure also shows
2
the capacity is not directly proportional to the amine loading,
and the important metric of amine efficiency (mol CO /mol
Received: July 1, 2015
2
N) tends to rise when moving from low to intermediate amine
©
XXXX American Chemical Society
A
DOI: 10.1021/jacs.5b06823
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
equilibrate under acidic conditions (forming an organic template for
silica condensation), prior to addition of tetraethylorthosilicate
(
TEOS, Sigma-Aldrich) and subsequent silica condensation, hydro-
thermal treatment, and calcination steps. For SBA15−100 preparation,
4 g of P123 were first dissolved in a 2L Erlenmeyer with 636 g of DI
water and 120 mL of fuming HCl (BDH) and stirred for 3 h. Next,
2
5
2
1
2.6 g of TEOS was added dropwise, and the solution stirred another
4 h at 40 °C. Stirring was then stopped and the material was held at
00 °C for 24 h. The SBA15−130 was prepared in a sealed 700 mL
Berghof synthesis reactor equipped with an impeller. Here, 12 g of
P123 was dissolved in 61.7 g of fuming HCl and 388.3 g of H O. The
2
solution was stirred at 600 rpm for 3 h at 40 °C. Then, 25.5 g of TEOS
was poured into the container and the mixture stirred at 600 rpm for 5
min, then stopped. The material was then statically aged at 40 °C for
2
4 h before hydrothermal treatment at 130 °C for 48 h. After the
hydrothermal treatments, all SBA-15 samples were filtered and rinsed
with water, then dried overnight at 75 °C. All samples were then
calcined at 550 °C for 6 h.
The surface of a portion of the SBA-15 was modified with
trimethylsilyl groups via the grafting of hexamethyledisilazane (HMDS,
Sigma-Aldrich). Here, 2 g of dried SBA-15 was dispersed in 100 mL of
anhydrous toluene in a sealed flask and allowed to equilibrate for 1 h.
Then, 1.2 g of HMDS was added via syringe and the solution was
stirred at 80 °C for 24 h before filtering, washing with toluene, and
vacuum drying at ∼80 °C.
Figure 1. Top: Schematic process for CO adsorption and recovery.
The hexagonally packed cylinder structure is exhibited by SBA-15 silica
2
supports. Bottom: CO capacity for a typical PEI/SBA-15 composite
2
as a function of PEI (M ∼ 800 Da) loading, in 400 ppm of CO /He
w
2
at 25 °C, with corresponding amine efficiencies. Inset: low molecular-
weight poly(ethylenimine) in a silica pore.
Poly(ethylenimine) Synthesis. Synthesis of poly(ethylenimine)-
d5 (d-PEI) is described in detail along with characterization in the
Supporting Information. Briefly, d-PEI was synthesized by the acid-
catalyzed polymerization of ethylenimine-d4 in the presence of
ethylenediamine-d4 (CDN Isotopes) as a capping agent. The
ethylenimine monomer was prepared by bromination of ethanol-
amine-d (CDN Isotopes) to bromoethylamine-d by heating in HBr,
space of two SBA-15 type silicas as a function of PEI loading.
We examine the scattering profiles in the context of other
morphological characterizations, including N physisorption
2
isotherms and electron microscopy images, as well as simulated
scattering models. Our analysis suggests that PEI initially forms
a thin coating, adhering to the surface of the pore walls at low
loadings, while increasing the PEI loading results in plug-like
agglomerates forming after the initial coating. We also find that
the plugs of PEI are narrower (on the order of 1 nm) than the
pore diameters derived from physisorption, in accord with the
proposition of a pore-perimeter-region containing silica of
4
4
followed by base-activated ring closure and concentration by
distillation. Final sample impregnation on the SBA-15 support was
done in MeOD (CDN) to achieve H/D exchange for the labile
hydrogen. Hydrogenated PEI (h-PEI) was purchased (Mw ∼ 800,
Sigma-Aldrich).
Silica/PEI Composite Preparation. PEI/SBA-15 composites
were prepared by mixing, equilibrating and drying solutions of PEI
and SBA-15 in MeOD (CDN) with a rotatory evaporator. For each
41,42
reduced density, often referred to as a corona.
The
sample, 30 mL MeOD, 250 mg SiO and a chosen amount of PEI were
2
evidence suggests that the corona region contains a boundary
with significant open volume that can mimic a wider pore when
used. PEI and SiO were added to methanol separately, stirred for 1 h
2
and then combined and further stirred for 3 h before solvent removal.
Final degassing was performed in a Schlenk flask on a high vacuum
probed by N , but that PEI does not necessarily permeate this
2
space; i.e., the corona can be dominated by either surface
line (∼10 mTorr) under a vacuum at 110 °C to remove any CO and
2
4
3
44−47
corrugation or narrow micropores,
synthetic conditions.
depending on
residual solvent. Samples were transported under a vacuum into a
helium glovebox for further handling.
Chemical Structure Characterization. Poly(ethylenimine)-d
Finally, we compare the CO adsorption performance of
5
2
and synthetic intermediates were probed by 2H NMR and
13
C
these materials against the polymer morphology model. We
infer that at low polymer loadings, oxide surface-binding
restricts the chemical accessibility of the amines for CO2
adsorption, leading to low amine efficiencies. Conversely,
NMR on a Bruker DRX 500 spectrometer. All samples were prepared
13
in distilled H O solvent. For the C measurements, all T relaxation
2
1
times were first determined, and scan times were allotted to yield
quantitative spectra. No decoupling sequences were needed as C/D
coupling only yielded slightly broadened peaks. Spectra demonstrating
product purity and distribution of primary, secondary, and tertiary
amines (defining branching rate) from the 13C spectra are shown in
the Supporting Information. The molecular weight was estimated by
gel-permeation chromatography using a Shimadzu HPLC system with
refractive index detector (RID-10A) and Viscotek TSK Viscogel
PWXL Guard, G3000, G4000, and G6000 columns (in series).
Samples were prepared at ∼30 mg/mL in an eluent phase of 0.1 M
sodium nitrate +0.1 M acetic acid and flowed at 1 mL/min. The
synthetic PEI elution time was compared to commercial samples of
branched PEI at MW 800, 1200, 1800, 2000, and 25 000. Elution of the
synthetic PEI came between the 800 and 1200 Da reference samples.
Physisorption and Analysis. All nitrogen physisorption iso-
therms were collected on a Micromeritics Tristar 3020 instrument at
77K. Samples were degassed at 110 °C for 12 h prior to measurement.
Pore volume and pore size distributions were calculated from the
higher loadings exhibit increased CO diffusional barriers due
2
to formation of plugs of PEI within the pores. These chemical
and diffusional accessibility factors must be balanced for
optimal adsorbent performance. As a demonstration of this
design principle, we synthesize an SBA-15 variant with a
trimethylsilyl-functionalized surface, for which we hypothesize
PEI to exhibit weaker silica surface adsorption. Adsorbents
prepared from this support material demonstrate faster CO2
uptake rates and higher total capacities than their surface
silanol-containing counterparts.
EXPERIMENTAL SECTION
■
(
(
Silica Synthesis. SBA-15 was prepared by adapted protocols from
2
,22
46,48
our prior work
(SBA15−100), and from the literature
SBA15−130). In each case, block copolymer Pluronic 123
EO PO EO , Sigma-Aldrich) was dissolved and allowed to
49,50
adsorption isotherms using the NLDFT equilibrium model
in the
2
0
70
20
B
DOI: 10.1021/jacs.5b06823
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Quantachrome VersaWin software package. Surface areas were
determined by the BET method.
CO Adsorption and Amine Loading. Dry CO capacities were
2
2
measured using a TA Instruments Q500 TGA. Samples were exposed
to 400 ppm of CO /He at 30 °C for 12 h and the corresponding
2
weight gain was recorded and normalized by the dry weight of sample
used. Samples were pretreated under He flow at 110 °C for 3 h prior
to exposure to CO containing gas. Amine content of the samples was
2
measured using a Netzch STA409PG TGA. Weight loss from 120 to
9
00 °C under a flow of nitrogen diluted air was recorded and
normalized by the residual mass at 900 °C.
Imaging. Transmission electron microscopy was performed on an
FEI Tecnai F30 field emission gun microscope at 300 keV accelerating
voltage. Samples were prepared by sonicating 5 mg/mL silica powder
in methanol and pipetting a drop onto a Cu grid with Formvar coating.
Scanning electron microscopy was performed on a Hitachi SU8010 at
5
keV. Samples were prepared on a carbon tape substrate.
Scattering Experiments. All small-angle neutron scattering
Figure 2. Representative SEM (above) and TEM (below) images for
SBA15−100 (left) and SBA15−130 (right).
samples were dry powders in gastight aluminum holders with 1 mm
path length and 1 in. diameter quartz windows. Sample weights were
chosen to maintain a constant 100 mg silica basis (e.g., a 50% wt. PEI
sample would be 200 mg total sample mass). All assembly was
performed under helium in a glovebox. SANS was performed on the
EQ-SANS instrument at the Spallation Neutron Source at Oak
Ridge National Lab. Two instrument configurations were used to
access a wide q-range and establish an incoherent background intensity
for subtraction. Low q data were acquired with a detector distance of 5
m and neutron wavelength of 6 Å. High q data were acquired with a
Region ii (intermediate pressure) of the isotherms corresponds
mainly to multilayer adsorption of N ; a higher slope in this
2
region should correspond to increased surface roughness (more
53
51
area) along the mesopores, as implied by the BET equation.
The isotherms together indicate that the SBA15−130 contains
higher specific pore volume and decreased microporosity
compared to SBA15−100, making it a more ideal cylindrical
array and a presumptively better model system for scattering
studies.
1
.3 m detector distance and 1 Å neutron wavelength. These
configurations had a q-uncertainty of less than 5% for all q > 0.01
Å . The two spectra for each sample were stitched with an overlap
region from 0.13 to 0.14 Å and calibrated to absolute intensity using
a silica standard (Porasil). No adjustments for SBA-15 packing fraction
were made.
Upon introduction of PEI, a very rapid drop in surface area
−1
51
(
BET method) is seen at small loadings, as shown in Figure 3b.
−1
49,50,54
Likewise, pore volumes (NLDFT method)
show an
initial decrease that is disproportionately large relative to the
volume of added PEI itself. This suggests that smaller intrawall
pores or surface corrugations may become blocked off, rather
than filled. This blocked intrawall porosity is less prevalent on
the SBA15−130 material, likely because it has larger and less
RESULTS
■
Morphology Analysis of PEI/SBA-15 by Physisorption.
Nitrogen physisorption isotherms were used to assess textural
characteristics of the PEI/silica composites including surface
area, pore volume, and pore-size-distribution. PEI weight-
loadings ranging from 0 to 50% (nominal) were investigated
44,46,47
numerous intrawall pores.
It is noteworthy that the
surface areas converge to similar values once the smallest
micropores are blocked. Corresponding isotherms (used to
determine the surface area and pore volume) for the full PEI-
loading series on both support materials are contained in Figure
S1.
We note that all physisorption data presented are from
samples used for scattering characterization, which required
fully deuterated PEI (d-PEI, synthesized in-house). Composite
sorbents prepared from this surrogate polymer showed CO2
adsorption capacity that was comparable, but ∼20% lower than
analogues prepared from commercial PEI, as shown in Figure
S2. The d-PEI had similar molecular weight to commercial
(
exact loadings in Table S1). Two separate material loading-
series were prepared using two batches of SBA-15, chosen to
assess the impact of textural differences such as particle size,
pore dimensions, and degree of intrawall porosity or surface
roughness. These supports were synthesized as described
1
8,46,48
elsewhere,
with the main differentiating factor being
hydrothermal treatment at 100 °C for 24 h (denoted SBA15−
1
00) and 130 °C for 48 h (denoted SBA15−130). In the case
of SBA15−130, synthesis conditions were selected to optimize
the mesostructure with minimal nonidealities (intrawall pores
etc.).
PEI’s used in composite sorbents (M ∼ 1000), while the ratio
w
4
6,48
In contrast, the SBA15−100 was prepared as a
of 1°:2°:3° amines was 20:48:32, compared to 44:33:23 in
55
13
2
standard to match many previous reports by our group and
others on PEI/SBA-15 adsorbents. In Figure 2, we show SEM
images of the silica particles comprising each material, as well as
TEM images depicting their internal mesopore structure. It can
be seen that the SBA15−130 generally exhibits larger, more
uniform particles with wider pores compared to the standard
SBA15−100.
commercial PEI (quantitative C NMR and H NMR in
Figures S7−S8). Primary amines are the most effective at CO
capture, largely accounting for the capacity difference.
2
Further morphological information can be extracted from
pore-size distribution (PSD) analyses, shown in Figure 4. Using
the NLDFT equilibrium PSD method, the bare SBA15−100
shows distinct volume contributions from pores with diameters
below 3 nm. These pores disappear at 10% wt PEI loading,
which corresponds to roughly a monolayer relative to the
mesopore surface area. Additional increases in PEI loading
appear to cause a contraction of the apparent diameter from
∼8.2 nm to ∼7 nm, after which the pore volume decreases with
no change in diameter. Although the calculated size distribution
N physisorption isotherms for each of the bare SBA-15
2
support materials are presented in Figure 3a. The pressure
regime of largest volume uptake (region (iii)) corresponds to
capillary condensation in the primary mesopores of the
material, while the sharp condensation feature at low pressure
(
region (i)) is generally attributed to narrow, intrawall
52
micropores, illustrated schematically in the inset of the figure.
is predicated on numerous assumptionsmainly that N
2
C
DOI: 10.1021/jacs.5b06823
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 3. (a) N physisorption isotherms for the each SBA-15 support at 77K, showing differences in microporosity and pore volume. Inset:
2
Schematic two-dimensional cross-section of SBA-15 (b) Summary of BET surface area and NLDFT pore volume vs PEI-loading on each material.
In the case of SBA15−130, the shift in pore diameter at low
PEI loadings is even more pronounced than on SBA15−100.
We suspect that surface corrugation or larger intrawall porosity
(
the distinction is not necessarily apparent) is more dominant
than narrow micropores in this material, as evidenced
additionally by larger N volumes adsorbing in the multilayer
2
pressure regime of the isotherms. Such corrugation could be
manifested as a larger apparent pore radius in the PSD rather
than distinct smaller pores. As PEI begins to coat (and possibly
fill) the corrugations, the apparent diameter drops significantly
from ∼10.8 to ∼8 nm. It is also worth noting that due to the
higher pore volume of SBA15−130, 50% wt nominal loading of
PEI only achieves a fill fraction of ∼60% of the total measurable
void space in this material. Higher PEI loading tests confirm a
similar trend in PSD evolution to the SBA15−100 materials,
and the apparent pore diameter does not fall below 7 nm.
Scattering Properties of PEI/SBA-15 Systems. Small
angle neutron scattering is well suited to probe the geometric
properties of each phase in PEI/SBA-15 composites due to
differences in the neutron scattering-length density (SLD) of
the polymer and that of the silica matrix. However, achieving a
strong contrast requires replacement of the hydrogen atoms on
PEI with deuterium, which scatters neutrons more strongly and
coherently (reducing noise). Thus, composites for this study
were prepared with fully deuterated PEI (d-PEI), synthetically
tailored to possess similar properties to commercial PEI
(characterization in SI).
Figure 4. Pore size distributions calculated from N physisorption
2
isotherms using the NLDFT method and a cylindrical silica pore
model for samples of SBA15−100 (a) and SBA15−130 (b) with d-PEI
loadings ranging from 0 to 50% wt. Introduction of the polymer causes
an initial drop in apparent radius, which levels off at higher loading.
In Figure 5 we show the scattering intensity of the bare SBA-
1
5 supports as a function of the magnitude of neutron
momentum transfer, q = (4π sin(θ)/λ), where θ is the
scattering angle and λ is the wavelength. Intensities are given in
adsorption is occurring in perfect cylinders of SiO with no
2
polymer coatingone could tentatively interpret the PSD to
indicate that PEI deposition begins as a conformal coating of
the mesopore (having an annular shape). After reaching some
critical thickness, PEI may then begin to aggregate (likely due
to less attraction to the polymer coated wall than the bare,
silanol-laden wall) taking the approximate shape of a cylindrical
plug that grows in the axial direction of the pore with further
amine loading. Such a deposition mechanism would permit a
decrease in volume with no change in diameter distribution
after a critical polymer loading. While the aforementioned
physisorption trends have been reported in numerous instances
−1
absolute units of differential cross-section per volume (m )
and were calibrated from a standard sample. To a reasonable
approximation, structural motifs of vastly different length scales
can be considered to dominate scattering within their own
characteristic q-regimes, where q ∼ (2π/d), and d is the
distance in real space. For both silica samples, the particle size
corresponds to q-values well below the range of measurement,
but scattering from the external surface of the particles still
contributes intensity at the lowest measured q-values and is
proportional to q− (Porod’s law, illustrated in Figure 5). At
intermediate q-values, the particle surface contribution decays
below that of the internal structure, which should in principle
be controlled by scattering of mesoscale features. However, it is
commonly observed in amorphous silica systems that a
relatively high concentration of smaller scale (and very
4
2
0,55,56
throughout the adsorption literature,
interpretation of sequential coating and plug formation
supported further below with scattering data) is the first
the present
(
clear explanation of the observed threshold diameter
phenomenon.
D
DOI: 10.1021/jacs.5b06823
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
relative locations of the scattering bodies. These quantities are
given respectively by the squared magnitude of the Fourier
transform (FT) of the SLD distribution defining the shape, and
the FT of a lattice describing the locations of the centers of
mass across the ensemble of shapes. The present scattering
system may be represented as an array of unit cells that (i) form
a hexagonal lattice, and (ii) are hierarchically arranged in the
form of a particle. In this general description, we may posit a
scattering law (derived in the Supporting Information) given by
N
p
I(q) = V {N P (q)·[S(q) + N P (q)] + i (q)}
(1)
C C
C P0
D
where P (q) is the form factor of the unit cell, N is the number
C
C
Figure 5. SANS patterns of SBA15−100 (blue) and SBA15−130
(
red), along with calculated scattering characteristics of an ideal
mesoporous particle with smooth boundaries at low and intermediate
q (dashed curves). The SBA15−130 contains larger and fewer
intrawall pores or corrugations than SBA15−100, and consequently
shows less diffuse scattering at high q.
polydisperse) inhomogeneities creates a broad, smooth
scattering characteristic that masks the mesopore scattering,
with the exception of diffraction peaks (the collective
interference pattern of the mesopores), which rise above the
57−61
baseline intensity.
We demonstrate this in Figure 5 with an
overlay of the calculated scattering pattern of a smooth 80 Å
pore inside a solid 120 Å hexagonal lattice primitive cell (details
in Supporting Information). A rapid decay into a second Porod
form factor and diffuse function both technically represent
“diffuse” scattering, but for clarity we refer to only the latter as
such.
−
4
(q ) regime, corresponding to internal surface scatter, would
The weak scattering by the bare mesostructure relative to the
smaller scale imhomogeneities in SBA-15 and other hexagonal
silicas has often relegated determination of structural
parameters (e.g., pore size or density gradients) with SANS
or SAXS to comparison of relative diffraction peak
be expected at much lower q-values than seen in experiment
due to smaller features such as micropores.
To adopt a more quantitative framework for discussion, we
consider that a simple scattering pattern may be described by
the product of a form factor, P(q), related to the shape of a
scattering body, and a structure factor, S(q), related to the
42,45,62−72
heights.
However, full-pattern fitting has been of use
in modeling characteristics of adsorbed films or pore-filling
Figure 6. (a) SANS patterns collected for both d-PEI/SBA-15 systems ranging from 0 to 50 wt % nominal loading of polymer (0−90% fill fraction
for SBA15−100 and 0−60% fill fraction for SBA15−130). (b) Bragg regions of the same patterns, shifted for visibility (c) Calculated form factors for
SBA-15 unit cell with varying fill fraction for each general motif depicted in Figure 7 and discussed in the text. An ideal hexagonal lattice structure
factor is overlaid for reference. Parameters used in the calculation were a = 120 Å, R = 40 Å, R = 50 Å, and L = 10 000 Å. Relative intensities are
p
c
self-consistent within each cluster of curves, but each cluster is shifted vertically for visibility.
E
DOI: 10.1021/jacs.5b06823
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
58,60,73−78
57,58,60,62,67,70,72,73
agents, particularly with SANS.
Addition of a high-
may contain some structural complexity,
it
contrast pore occupant such as d-PEI intensifies the
was modeled as a single annular region with an average density
mesostructure scattering. On the basis of component atomic
to avoid excessive free parameters. For each filling motif, the
79
scattering-lengths, we estimate the SLD of d-PEI to be ∼8.2
form factor, P
polymer fill fraction, f, pore diameter, R
as well as a corona thickness, t = R − R
. Analytical expressions for the unit cell scattering amplitude
C
(q), was investigated as a function of the
, and lattice constant, a,
, and corona density,
−
6
−2
−6 −2
×
(
10 Å , compared to ∼3.5 × 10
Å
for amorphous SiO2
p
nonporous), yielding a scattering cross-section ratio (squared
c
p
c
ratio of SLDs) of at least fivelarger in practice due to small or
ρ
c
occluded porosity in the silica phase.
were constructed from the superposition principle of waves and
approximation of each density region with a geometric shape
with a known scattering law. These expressions are deferred to
the Supporting Information, but we note they were validated
against direct numerical Fourier transform and orientational
averaging of 3-dimensional density grids for numerous
representative cases (also in the SI).
The clearest qualitative impact of changes to the unit cell
mesostructure can be observed from its effect on low-order
diffraction intensities, which rise above the diffuse background.
In Figure 6a, we show SANS patterns for the full 0−50% wt
d-PEI-loading series on SBA15−100 and SBA15−130. Spectra
were taken with an equal silica-mass basis so that the differences
in intensity directly reflect changes to the volumetric fill fraction
of the PEI polymer. The incremental addition of PEI raises the
total scattered intensity and generates several unique changes to
the shape of the scattering pattern. To the left of the [10] Bragg
peak, the intensity takes on an increasingly pronounced linear
−1
characteristic (intensity ∝ q ), suggestive of high aspect-ratio
80
The effect originates in the product P (q)·S(q) in eq 1, which
prism structures such as a cylinder or elongated shell. The
10] peak concomitantly takes on an increasingly asymmetric
C
modulates the Bragg peak intensities (structure factor) by the
magnitude of the form factor at a given q. In Figure 6b, we
show the Bragg regions of the experimental scattering patterns
for both sets of PEI/SBA-15 systems. These plots highlight the
characteristic filling trends mentioned previouslynamely, the
growing asymmetry of the [10] peak and the relative intensity
inversion of the [11] and [20] peaks in both systems. From eq
[
shape. These features are both direct manifestations of the
mesoscale unit cell scattering, which was not directly observable
from the empty SBA-15. It can also be seen that the [11] and
[
20] Bragg peaks undergo a relative inversion in intensity.
Collectively, these changes provide a signature of the PEI
deposition morphology, which we make use of and discuss
below in the context of modeling and data fitting.
1
, it is expected that the magnitude of the unit cell form factor
should exhibit corresponding functionality. Form factors for the
various morphological motifs discussed above, calculated as a
function of polymer filling, are thus shown in Figure 6c. For
illustration, we have simply chosen a representative set of unit
cell parameters and overlaid the structure factor of the
corresponding hexagonal lattice (to be multiplied by the form
factor point-wise). The figure shows (i) plug and (ii) conformal
annular coating models, as well as (iii) a hybrid model, whereby
the first 20% vol of polymer (∼1−2 monolayers depending on
roughness) is allocated to a surface coating with further
addition yielding a plug in the remaining central space. It
should be noted that although the proposed polymer plugs
have lengths corresponding to very low q-values, a growing
fraction of plug occupation along the pore is still manifested
across the measured q-range: the spectrum roughly behaves as a
weighted average between the two-dimensional cross-sectional
spectra of regions with and without occupation by the plug.
It can be seen that in the case of a pure plug model, the
asymmetry of the [10] peak would be expected to set in rapidly
after relatively low filling (pending diffuse background level),
while for a conformal coating or sequential coating-plug model
this asymmetry would be delayed until higher fillings (as
observed). Shifting focus to the higher order reflections, strong
minima are evident at intermediate fill fractions in the plug and
hybrid models, which will lower the [11] peak relative to the
SANS Model for PEI Deposition. A variety of
morphological filling motifs for PEI within the SBA-15 pores
can be envisioned based on the interaction of the polymer with
the pore wall. A strongly attractive interaction would be
expected to yield an adsorbed coating, and this interaction
could potentially extend to many conformal layers. Conversely,
inert or repulsive walls should lead to preferential aggregation
of the polymer as it minimizes interfacial area.
To assess the possible modes of polymer filling, including
various intermediate cases, we explicitly modeled the PEI as (i)
a conformal polymer coating (an annular shell of growing
thickness at the pore wall, surrounding an open core), (ii) a
single polymer plug (same diameter as pore, growing
lengthwise with filling), and (iii) various hybrid structures
where polymer filling creates both a plug and a conformal
coating. The hybrid case was examined both sequentially
(
shifting to a plug after reaching a critical shell thickness) and
simultaneously (both structures growing in different propor-
tions at the same time). Schematics of each general filling motif
are shown in Figure 7. We also considered the possibility of a
reduced-density corona region spanning an annular space of
unknown thickness at the pore perimeter. Though this region
[20] peak; a strict conformal deposition will not reproduce the
inversion until nearly complete filling, thus pointing to the
hybrid case. Naturally, the models converge in the limit of
completely full (or empty) pores.
Quantitative Fitting and Interpretation. Full-pattern
fitting with eq 1 was used to estimate characteristic dimensions
of the PEI/SBA-15 systems. Unit cell form factors were
determined as described above. The structure factor was that of
a 2-D hexagonal lattice, parametrized with a Debye−Waller
Figure 7. Schematic of the proposed SBA-15 morphology and various
PEI-filling motifs. The pore is modeled as containing an empty
cylindrical region of radius R and an annular region of reduced density
p
(
relative to the bulk SiO ), with outer radius R . PEI may fill the pore
2
c
2
factor ⟨u⟩ to account for variable displacement from ideal
conformally (i), as an aggregate (ii), or as a mixture of these modes
iii), either sequentially (coating up to critical thickness, then forming
a plug) or simultaneously (both regions coexisting at all fill fractions).
(
lattice planes (see SI). To model the myriad array of small-scale
inhomogeneities contributing to i (q), we empirically selected a
D
F
DOI: 10.1021/jacs.5b06823
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 8. (a) Example fits of SANS patterns for bare SBA15−130 and 40 wt % PEI/SBA15−130. Component functions include: particle form factor
−
4
P (q) at low q (∝q here), unit cell form factor P (q) at intermediate q, and diffuse scattering function i (q) at high q. These are not strictly
P0
C
D
additive, per eq 1. The red line represents the product P (q)·S(q), with adjustments for polydispersity and resolution (see Supporting Information).
C
(
b) Mean-squared density fluctuations vs fill fraction, obtained from the diffuse scattering function. (c) Schematic of initial PEI coating modes
whereby a distinct polymer layer may or may not be seen depending on dimensions of surface corrugations. (d) Corona density vs fill fraction for
each SBA-15 material, shown for the two best-fit fill motifs. Error bars give the range of values that shift the SSR by <5% in (b) and (d).
a
Table 1. Structural Parameters Derived for SBA-15 Samples
PEI/SBA15−100
PEI/SBA15−130
2
2
2
2
model
Rp (Å)
tc (Å)
⟨u ⟩ (Å )
SSR
Rp (Å)
tc (Å)
⟨u ⟩ (Å )
SSR
plug
31
33
33
32
32
12
10
12
14
12
100
160
200
260
240
1.15
1.09
1.04
1.00
1.11
41
41
36
10
8
100
120
280
1.00
1.05
1.23
sequential 10/90
sequential 20/80
sequential 30/70
simultaneous 50/50
conformal
16
unstable
35
18
140
1.14
unstable
unstable
a
On the basis of mimimization of collective error in I(q) for a PEI loading series based on different potential filling morphologies. Polymer fill
2
models are described in the main text. Parameters are pore radius (R ), corona thickness (t ), Debye−Waller factor (⟨u ⟩), and sum of squared
p
c
2
residuals (SSR), which is normalized to the best fit. Since I(q) spans multiple orders of magnitude, squared residuals were weighted by 1/|I(q)| to
make all points contribute equally.
two-phase random-medium model, first described by Debye et
included three “sequential hybrids” whereby either the first
10, 20, or 30% vol were allocated to a conformal coating
(setting a critical thickness) while additional polymer was
allocated to a plug in the center of the annular space. A final
8
1,82
al.
The scattering of such a system yields a squared-
Lorentzian distribution defined by an effective correlation
2
2
length (ξ) and the mean-squared SLD fluctuations η = ρ −
2
ρ :
“
simultaneous hybrid” model placed equal volumes of polymer
onto the pore wall and plug region, irrespective of the total fill
i.e., fill fraction was divided by two and this amount was
3
2
8
πξ (η )V
S
(
iD(q) =
(2)
2
2 2
(
1 + q ξ )
allocated to each region). In all, ∼500 000 form factor
candidates were generated.
This contribution is considered to be purely additivei.e.,
no coherent interferenceand its intensity scales with
illuminated volume, which we have defined as the solid skeletal
To keep results physically transparent, we chose to fix
parameters that could be determined a priori with reasonable
accuracy (polymer fill fraction from physisorption, lattice
constant from Bragg peak positions). We then fit parameters
that are intrinsic to a given support (average pore radius,
corona radius, and Debye−Waller factor) through an aggregate
error minimization across each complete loading series (i.e.,
minimizing the sum of the sums of squared error for all patterns
on a given material). For this initial step, the remaining
variables (corona density, diffuse scattering parameters) were
allowed to vary across fill fractions. Two sample fits derived in
this manner are shown in Figure 8a, along with overlays of
volume, V (the nonmesoporous volume of the particle).
s
2
Likewise, the values obtained for η only correspond to
fluctuations on the length scale of ξ rather than the global
mean-squared fluctuations, often obtained from total spectrum
6
0
integration. This modeling choice is discussed further in the
Supporting Information, along with treatment of corrections for
instrumental resolution, polydispersity, and orientational
averaging.
Because of the complexity of the total scattering law, data
were fit by numerical error minimization based on a series of
constrained factorial parameter sweeps. To reduce computation
time, a database of unit cell form factors was first generated,
incrementally spanning all realistic dimensions (pore radius,
corona thickness and density etc.) for each filling motif (cf.
Figure 7). Six models were considered, including the plug
model, annular coating model, and four hybridsthese
several component functions for visual aid. The optimum
2
values for R , t , and ⟨u⟩ achieved in the above manner are
P
C
shown in Table 1 for each filling model considered. Summed
squares of residual error (SSR), normalized to the best fit, are
also shown to provide a gauge for how uniquely each model fits
the data.
G
DOI: 10.1021/jacs.5b06823
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 9. (a) Amine efficiency for untreated and HDMS capped PEI/SBA-15 composites. (b) CO uptake over time for PEI/SBA15−100 materials
2
before and after HDMS capping. Kinetic data on semilog axes available in Figure S5.
The fits reveal that the collective data set for the PEI/
SBA15−100 series is most consistent with a sequential fill
motif, in which the first 30% vol of PEI is deposited as a coating
and further addition yields an aggregated plug. Conversely, it
appears that the PEI/SBA-130 forms a much thinner (if any)
surface coating, with the plug model yielding a marginally better
fit than the thinnest layer sequential layer model (10% vol.
coating). Thinner coating cases were examined for SBA15−130
but were negligibly better in fit qualitythese would
correspond to subnanometer features that are not resolvable.
It should be noted that the fitting parameters in Table 1
represent clear optima for each filling motif (SSR vs each
variable is parabolic), and that the next-best fits represent small
perturbations to the optimized topology; i.e., the sequential
propose that this layer is manifested as a continuous high SLD
coating on SBA15−100 because the corona is relatively
impermeable to PEI, save for perhaps individual side-chains
that may serve as anchor points. Such an interaction would
result in polymer chains that are tethered, but extend into the
pore rather than lay flat. In contrast, the corona of SBA15−130
possesses larger, smooth corrugations, which can be directly
occupied by PEI molecules. The initial coating will be more
discontinuous in this case and have a lower average density
(moving in radial increments). We display a schematic of the
envisioned interactions between the two pore/corona top-
ologies and PEI in Figure 8c. This picture is consistent with the
more dramatic drop in apparent pore size in the PSD of
SBA15−130 (vs SBA15−100), since wider corrugations are
more likely to cause delayed condensation in the mesopore
(increasing apparent diameter) than to condense N2
independently.
CO2 Adsorbent Design and Performance. The
sequential model of pore filling inferred from the physisorption
and scattering data (surface coating followed by plug
formation) suggests there is in general very little freely
accessible PEI in contact with the void space of the pore. At
low PEI loadings, amine sites are largely unavailable due to
strong interaction with the pore wall, while at higher loadings,
1
0−90% model is the next-best relative to the plug model for
SBA15−130, while a 20−80% sequential model is second in
quality to the 30−70% sequential option for SBA15−100.
Values obtained for the diffuse scattering parameters (ξ and
2
η ) and the corona SLD (ρ ) as a function of PEI fill fraction
c
shed additional light onto the polymer−silica interaction. We
2
first note in Figure 8b that η values, which represent the
magnitude of spatial SLD fluctuations at the micropore scale,
are effectively constant for SBA15−100 (becoming drowned
out by mesopore scatter at the highest fillings), while they
exhibit a clear increase with fill fraction on SBA15−130.
Correlation lengths did not appreciably vary with filling (6−8 Å
CO must diffuse through a significant volume of PEI to reach
2
the majority of unsaturated amine sites. To date, there remain
conflicting reports as to whether the pore dimensions of a
mesoporous support will independently improve the amine
in SBA15−100 and 10−12 Å in SBA15−130), and the trends in
2
η
provide initial evidence that the intrawall porosity is
2
0,21,23−27
accessible to the polymer only in the case of the larger,
smoother pore SBA15−130 material. This picture is corrobo-
rated by the corona SLDs, shown in Figure 8d. In the figure, we
efficiency of impregnated PEI.
Reports on the
importance of PEI/surface interactions are less numer-
22,28,29
ous,
and ambiguities along both avenues stem in part
show ρ values (normalized to the wall density) obtained for
from difficulty in fixing one textural property of a support while
systematically changing another. The present work implies that
an optimal support (maximizing amine efficiency) may be
characterized by having (i) a high specific pore volume and
short pore length to minimize diffusion lengths through PEI
plugs, (ii) weak PEI/surface interactions and (iii) a surface area
approaching that of the geometric area of the pores (minimal
corrugation) to minimize competition between CO2 and
surface sites for binding of PEI.
To evaluate the proposed design hypotheses, we post-
synthetically modified the surfaces of our SBA15 materials with
hexamethyldisilazane (HMDS) to convert the hydrophilic
silanols to hydrophobic trimethylsilyl groups. This was
c
each SBA-15 material using both the optimal fill motif as well as
the “next-best” motif. In general, small increases to the corona
SLD are observed upon introduction of PEI, and in most cases
the SLD increase is larger and present at lower loadings for
SBA15−130, as would be expected from PEI occupying the
2
corona and raising η . More interestingly, by comparing to the
less optimal fit motifs, a compensation effect can be seen
whereby a thicker conformal coating tracks with a lower corona
density, and vice versa (e.g., plug model showing higher corona
SLD than the sequential model for SBA15−130). Since the
pore/corona boundary is not in reality a discrete transition, this
convergent behavior provides confidence in the hypothesis that
an initial conformal layer is formed in both materials. We
expected to raise the CO capacity and uptake rate by changing
2
H
DOI: 10.1021/jacs.5b06823
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
the attractive interaction of the surface with PEI to a repulsive
interaction, freeing up previously bound amines. It was also
forms a conformal coating on the pore walls, then begins to
form plug-like aggregates at higher loading. The preferred filling
motif likely relates to a relatively high surface tension of PEI
and preference for the molecules to remain in a bulk-polymer-
like state when they cannot directly contact the strongly
interacting pore wall. Scattering length density profiles fit to the
data indicate that the nature of the coating may be highly
compact or more diffuse depending on the roughness and
intrawall porosity of a given silica material. The filling model
ultimately points to a trade-off between high chemical
accessibility of the amine binding sites, which are inaccessible
when they bind to silica, and high accessibility for mass
transport, which can be hampered by diffusion through PEI
plugs. We have illustrated this design principle by demonstrat-
anticipated that higher total CO capacity could highlight mass-
2
transport limited regimes, supporting the formation of plugs.
We note that while the grafted coating may close off very small
intrawall pores, PSDs of the HDMS-SBA-15 materials (Figure
S3) show very minor changes, with loss of volume fractions
from pores only <2 nm. The present data indicate PEI does not
appreciably enter such spaces. Thus, the accessibility of amine
sites is more likely to reflect PEI-surface interactions.
Figure 9 shows amine efficiencies collected at 400 ppm of
CO for a PEI-loading series (commercial h-PEI) of SBA15−
2
1
00 and SBA15−130 with and without HDMS capping, as well
as uptake vs time (normalized to equilibrium capacity) for the
SBA15−100 materials. In both cases the HDMS treatment
ing higher CO capacity and uptake rate for PEI supported in a
2
clearly increases the total CO capacity, with post-treatment
hydrophobically coated silica, yielding repulsive wall inter-
actions. While such materials may introduce new complications
(e.g., stability loss from PEI leaching), appropriate responses
(e.g., mixed-functionality coatings with anchoring sites) can
also be envisioned. Such architectures may permit design of
2
capacities being very similar for both materials at low loadings.
The SBA15−130 has a less dramatic enhancement, as this
substrate is inherently less prone to inhibition due to lower
surface-area. The performance of SBA15−130 also appears to
converge with that of its HDMS-capped counterpart at higher
PEI loading, likely because the bound amines represent a
smaller fraction of the total available sites. Despite the overall
enhancements, the two HDMS-capped materials show a
divergence in capacity in the high-loading limit. This limit
represents nearly filled pores for the SBA15−100 while there is
still substantial free space in the SBA15−130. Thus, we
hypothesize that the PEI plug component reaches a critical
length in SBA15−100, at which point true adsorption
equilibration becomes unachievable on the experimental time
scale. We suggest that this regime could set in rather abruptly
since there may exist a compound effect between increased
diffusion length and decreased diffusivity, caused by polymer
sorbent systems with faster CO concentration capability and
2
improved energy efficiency.
ASSOCIATED CONTENT
Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/jacs.5b06823.
■
*
S
Supplementary adsorbent characterization, including
summary of textural properties for all PEI/SBA-15
samples, physisorption isotherms of d-PEI/SBA-15
adsorbents, comparison of CO adsorption on SBA-15
2
composites with h-PEI vs d-PEI, pore size distributions
39,40
of HDMS-capped PEI/SBA-15 adsorbents, and CO
2
cross-linking in the course of saturation.
adsorption vs time on PEI/SBA15−130 and PEI/
SBA15−130-HDMS. SANS modeling details, and syn-
thesis and characterization of d-PEI are also provided.
The trends in CO capacity are consistent with rates of
2
pseudoequilibration, demonstrated for SBA15−100 in Figure
9
b. We first note that the low PEI-loading samples show a
(PDF)
significant increase in equilibration rate upon functionalization
with HDMS. The equilibration rates increase for each sample
up to and including 40 wt %, while the fully pore-blocked 50 wt
AUTHOR INFORMATION
Corresponding Author
*cjones@chbe.gatech.edu
■
%
sample shows essentially identical performance before and
after capping. We suggest that while the HDMS-capped sorbent
should have more free-amines available for binding, a greater
Notes
degree of saturation with CO may lower the diffusivity of gas
2
The authors declare no competing financial interest.
through PEI and undermine the increased site availability. It is
also conceivable that the microporous corona region of the
SBA15−100 could actually aid in gas transport and that this
benefit is lost upon closure of the micropores with HDMS. This
notion would be supported by the fact that the uncapped
SBA15−130, which possesses less microporosity than SBA15−
ACKNOWLEDGMENTS
■
Research supported as part of UNCAGE-ME, an Energy
Frontier Research Center funded by the U.S. Department of
Energy, Office of Science, Basic Energy Sciences, under Award
DE-SC0012577 (neutron scattering studies), and by the U.S.
Office of Naval Research contract # N00014-14-1-0704
adsorption studies). A portion of this research at ORNL’s
Spallation Neutron Source was sponsored by the Scientific User
Facilities Division, Office of Basic Energy Sciences, US
Department of Energy. We also thank Dr. Gernot Rother for
discussions in planning the scattering experiments and Chun-
Jae Yoo for acquiring the SEM images.
#
1
00, already shows a drop in saturation rate at 50 wt %, which
represents only about 60% pore filling for the material (kinetic
data for SBA15−130 in Figure S4). While the diffusional
impediments at higher amine-loadings are modest in our model
silica, these materials also possesses relatively short, linear
transport paths. Commercial silicas that are practical for scale-
up generally possess longer, tortuous pores, and in applications
such as membrane separations, diffusion paths can be as much
(
40,83
as 1−2 orders of magnitude greater.
REFERENCES
■
(
1) Long, W.; Brunelli, N. A.; Didas, S. A.; Ping, E. W.; Jones, C. W.
ACS Catal. 2013, 3, 1700.
(2) Brunelli, N. A.; Didas, S. A.; Venkatasubbaiah, K.; Jones, C. W. J.
Am. Chem. Soc. 2012, 134, 13950.
CONCLUSIONS
Small angle neutron scattering and physisorption analyses have
suggested that in PEI/silica composite adsorbents, PEI initially
■
I
DOI: 10.1021/jacs.5b06823
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(
(
(
3) Das, D. D.; Sayari, A. J. J. Catal. 2007, 246 (1), 60.
4) Choi, S.; Drese, J. H.; Jones, C. W. ChemSusChem 2009, 2, 796.
5) Drese, J. H.; Talley, A. D.; Jones, C. W. ChemSusChem 2011, 4,
(36) Espinal, L.; Green, M. L.; Fischer, D. A.; Delongchamp, D. M.;
Jaye, C.; Horn, J. C.; Sakwa-Novak, M. A.; Chaikittisilp, W.; Brunelli,
N. A.; Jones, C. W. J. Phys. Chem. Lett. 2015, 6, 148.
(37) Bollini, P.; Brunelli, N. A.; Didas, S. A.; Jones, C. W. Ind. Eng.
Chem. Res. 2012, 51, 15153.
379.
(
(
6) Sayari, A.; Hamoudi, S.; Yang, Y. Chem. Mater. 2005, 17 (1), 212.
7) Serna-Guerrero, R.; Sayari, A. Environ. Sci. Technol. 2007, 41 (13),
(38) Wilfong, W. C.; Chuang, S. S. C. Ind. Eng. Chem. Res. 2014, 53,
4
(
(
761.
8) Nomura, A.; Jones, C. W. ACS Appl. Mater. Interfaces 2013, 5
12), 5569.
9) Belmabkhout, Y.; Serna-Guerrero, R.; Sayari, A. Ind. Eng. Chem.
Res. 2010, 49, 359.
10) Tailor, R.; Abboud, M.; Sayari, A. Environ. Sci. Technol. 2014, 48
3), 2025.
11) Xu, X.; Song, C.; Andresen, J. M.; Miller, B. G.; Scaroni, A. W.
Energy Fuels 2002, 16 (18), 1463.
12) Bollini, P.; Didas, S. A.; Jones, C. W. J. Mater. Chem. 2011, 21,
5100.
13) Dutcher, B.; Fan, M.; Russell, A. G. ACS Appl. Mater. Interfaces
015, 7 (4), 2137.
14) Chen, C.; Kim, J.; Ahn, W.-S. Korean J. Chem. Eng. 2014, 31
11), 1919.
15) Hedin, N.; Chen, L.; Laaksonen, A. Nanoscale 2010, 2 (10),
819.
16) McNutt, M. K.; Abdalati, W.; Caldeira, K.; Doney, S. C.;
4224.
(39) Wilfong, W. C.; Srikanth, C. S.; Chuang, S. S. C. ACS Appl.
Mater. Interfaces 2014, 6, 13617.
(40) Kim, H.; Chaikittisilp, W.; Jang, K.-S.; Didas, S. A.; Johnson, J.
R.; Koros, W. J.; Nair, S.; Jones, C. W. Ind. Eng. Chem. Res. 2015, 54,
4407.
(41) Blin, J. L.; Imper
4071.
(42) Solovyov, L. A. Chem. Soc. Rev. 2013, 42, 3708.
(43) Liu, Z.; Terasaki, O.; Ohsuna, T.; Hiraga, K.; Shin, H. J.; Ryoo,
R. ChemPhysChem 2001, 2 (4), 229.
(
(
(
́
or-Clerc, M. Chem. Soc. Rev. 2013, 42 (42),
(
(
1
(
2
(
44) Ryoo, R.; Ko, C. H.; Kruk, M.; Antochshuk, V.; Jaroniec, M. J. J.
Phys. Chem. B 2000, 104, 11465.
45) Imperor-Clerc, M.; Davidson, P.; Davidson, A. J. J. Am. Chem.
Soc. 2000, 122 (16), 11925.
46) Galarneau, A.; Cambon, H.; Di Renzo, F.; Fajula, F. Langmuir
001, 17 (26), 8328.
47) Kruk, M.; Jaroniec, M.; Ko, C. H.; Ryoo, R. Chem. Mater. 2000,
2 (7), 1961.
48) Lee, H. I.; Kim, J. H.; Stucky, G. D.; Shi, Y.; Pak, C.; Kim, J. M.
J. Mater. Chem. 2010, 20, 8483.
(
(
(
(
1
(
2
(
1
(
Falkowski, P. G. Climate Intervention: Carbon Dioxide Removal and
Reliable Sequestration; National Academies Press: Washington, D.C.,
(
2
(
015; p 141.
17) Eisenberger, P. M.; Cohen, R. W.; Chichilnisky, G.; Eisenberger,
N. M.; Chance, R. R.; Jones, C. W. Energy Environ. 2009, 20, 973.
18) Zhao, D.; Feng, J.; Huo, Q.; Melosh, N.; Fredrickson, G.;
Chmelka, B.; Stucky, G. Science (Washington, DC, U. S.) 1998, 279,
48.
19) Zhao, D.; Huo, Q.; Feng, J.; Chmelka, B. F.; Stucky, G. D. J. J.
Am. Chem. Soc. 1998, 120 (5), 6024.
20) Zhang, H.; Goeppert, A.; Czaun, M.; Prakash, G. K. S.; Olah, G.
A. RSC Adv. 2014, 4 (37), 19403.
21) Wang, D.; Wang, X.; Ma, X.; Fillerup, E.; Song, C. Catal. Today
014, 233, 100.
22) Kuwahara, Y.; Kang, D. Y.; Copeland, J. R.; Brunelli, N. A.;
(
49) Thommes, M.; Cychosz, K. A. Adsorption 2014, 20 (2−3), 233.
(50) Landers, J.; Gor, G. Y.; Neimark, A. V. Colloids Surf., A 2013,
4
(
1
37, 3.
51) Zhao, J. K.; Gao, C. Y.; Liu, D. J. J. Appl. Crystallogr. 2010, 43,
068.
(
5
(
(52) Thommes, M. Chem. Ing. Tech. 2010, 82 (7), 1059.
(53) Brunauer, S.; Emmett, P. H.; Teller, E. J. J. Am. Chem. Soc. 1938,
6
0 (1), 309.
(
(54) Monson, P. A. Microporous Mesoporous Materials 2012, 16, 47.
(55) Drese, J. H.; Choi, S.; Lively, R. P.; Koros, W. J.; Fauth, D. J.;
(
2
(
Gray, M. L.; Jones, C. W. Adv. Funct. Mater. 2009, 19, 3821.
56) Sanz, R.; Calleja, G.; Arencibia, A.; Sanz-Perez, E. S. Microporous
Mesoporous Mater. 2012, 158, 309.
57) Pollock, R. A.; Walsh, B. R.; Fry, J.; Ghampson, I. T.;
(
́
Didas, S. A.; Bollini, P.; Sievers, C.; Kamegawa, T.; Yamashita, H.;
Jones, C. W. J. J. Am. Chem. Soc. 2012, 134, 10757.
(
Melnichenko, Y. B.; Kaiser, H.; Pynn, R.; Desisto, W. J.; Wheeler, M.
C.; Frederick, B. G. Chem. Mater. 2011, 23, 3828.
(
23) Vilarrasa-Garcia, E.; Moya, E. M. O.; Cecilia, J. a.; Cavalcante, C.
L.; Jimenez-Jimenez, J.; Azevedo, D. C. S.; Rodríguez-Castellon, E.
Microporous Mesoporous Mater. 2014, 209, 172.
24) Yan, X.; Zhang, L.; Zhang, Y.; Yang, G.; Yan, Z. Ind. Eng. Chem.
Res. 2011, 50, 3220.
25) Wang, L.; Yao, M.; Hu, X.; Hu, G.; Lu, J.; Luo, M.; Fan, M.
Appl. Surf. Sci. 2015, 324, 286.
26) Heydari-Gorji, A.; Yang, Y.; Sayari, A. Energy Fuels 2011, 25,
206.
27) Yan, X.; Komarneni, S.; Yan, Z. J. J. Colloid Interface Sci. 2013,
90 (1), 217.
28) Heydari-Gorji, A.; Belmabkhout, Y.; Sayari, A. Langmuir 2011,
7, 12411.
29) Yue, M. B.; Chun, Y.; Cao, Y.; Dong, X.; Zhu, J. H. Adv. Funct.
Mater. 2006, 16 (13), 1717.
30) Cogswell, C. F.; Jiang, H.; Ramberger, J.; Accetta, D.; Willey, R.
J.; Choi, S. Langmuir 2015, 31, 4534.
31) Sakwa-Novak, M. A.; Jones, C. W. ACS Appl. Mater. Interfaces
014, 6, 9245.
32) Donaldson, T. L.; Nguyen, Y. N. Ind. Eng. Chem. Fundam. 1980,
9, 260.
33) Pinto, L.; Mafra, L.; Guil, J. M.; Pires, J.; Rocha, J. Chem. Mater.
011, 23, 1387.
34) Danon, A.; Stair, P. C.; Weitz, E. J. J. Phys. Chem. C 2011, 115,
1540.
35) Bacsik, Z.; Ahlsten, N.; Ziadi, A.; Zhao, G.; Garcia-bennett, A.
E.; Martin-Matute, B.; Hedin, N. Langmuir 2011, 27, 11118.
́
́
́
(
58) Jah
Findenegg, G. H. J. J. Phys. Chem. C 2009, 113, 15201.
59) Muter, D.; Jahnert, S.; Dunlop, J. W. C.; Findenegg, G. H.; Paris,
O.; Mu, D.; Jahnert, S.; Dunlop, J. W. C.; Findenegg, G. H.; Paris, O. J.
J. Phys. Chem. C 2009, 113, 15211.
60) Findenegg, G. H.; Jahnert, S.; Mu
Chem. Chem. Phys. 2010, 12, 7211.
61) Vogt, B. D.; Pai, R. A.; Lee, H.; Hedden, R. C.; Soles, C. L.; Wu,
W.; Lin, E. K.; Bauer, B. J.; Watkins, J. J. Chem. Mater. 2005, 17, 1398.
62) Edler, K. J.; Reynolds, P. A.; White, W.; Cookson, D.; White, J.
W.; Cookson, D. J. J. Chem. Soc., Faraday Trans. 1997, 93, 199.
63) Hammond, W.; Prouzet, E.; Mahanti, S. D.; Pinnavaia, T. J.
Microporous Mesoporous Mater. 1999, 27, 19.
̈
̈
nert, S.; Muter, D.; Prass, J.; Zickler, G. A.; Paris, O.;
(
(
̈
̈
̈
̈
(
(
̈
̈
ter, D.; Prass, J.; Paris, O. Phys.
(
4
(
3
(
2
(
(
(
(
(64) Jaroniec, M.; Solovyov, L. A. Langmuir 2006, 22 (8), 6757.
(
(65) Muroyama, N.; Ohsuna, T.; Ryoo, R.; Kubota, Y.; Terasaki, O. J.
J. Phys. Chem. B 2006, 110, 10630.
(66) Pikus, S.; Solovyov, L. A.; Kozak, M.; Jaroniec, M. Appl. Surf. Sci.
2007, 253, 5682.
(67) Sauer, J.; Marlow, F.; Schu
(
2
(
1
(
2
(
1
̈
th, F. Phys. Chem. Chem. Phys. 2001,
3, 5579.
(68) Solovyov, L. A.; Kirik, S. D.; Shmakov, A. N.; Romannikov, V.
N. Microporous Mesoporous Mater. 2001, 44−45, 17.
(69) Albouy, P. A.; Aryal, A.; Ayral, A. Chem. Mater. 2002, 14, 3391.
(70) Hofmann, T.; Wallacher, D.; Huber, P.; Birringer, R.; Knorr, K.;
Schreiber, A.; Findenegg, G. H. Phys. Rev. B: Condens. Matter Mater.
Phys. 2005, 72, 1.
(
J
DOI: 10.1021/jacs.5b06823
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(
71) Schreiber, A.; Ketelsen, I.; Findenegg, G. H.; Hoinkis, E.;
Llewellyn, P. L.; Rodriquez-Reinoso, F.; Rouqerol, J.; Schreiber, A.;
Ketelsen, I.; Findenegg, G. H.; Hoinlds, E.; Hoinkis, E. Stud. Surf. Sci.
Catal. 2007, 160, 17.
(
72) Zickler, G. A.; Jah
Findenegg, G. H.; Paris, O. Phys. Rev. B: Condens. Matter Mater. Phys.
006, 73, 1.
73) Edler, K. J.; Reynolds, P. A.; White, J. W. J. J. Phys. Chem. B
998, 102 (97), 3676.
74) Mascotto, S.; Wallacher, D.; Brandt, A.; Hauss, T.; Thommes,
̈
nert, S.; Wagermaier, W.; Funari, S. S.;
2
(
1
(
M.; Zickler, G. A.; Funari, S. S.; Timmann, A.; Smarsly, B. M.
Langmuir 2009, 25 (13), 12670.
(
75) Smarsly, B.; Go
E.; Go, C.; Antonietti, M.; Ruland, W.; Hoinkis, E. J. J. Phys. Chem. B
001, 105, 831.
76) Erko, M.; Wallacher, D.; Brandt, A.; Paris, O. J. J. Appl.
Crystallogr. 2010, 43, 1.
77) Muter, D.; Shin, T.; Deme,
G. H. J. Phys. Chem. Lett. 2010, 1, 1442.
78) Shin, T. G.; Muter, D.; Meissner, J.; Paris, O.; Findenegg, G. H.
Langmuir 2011, 27, 5252.
̈
ltner, C.; Antonietti, M.; Ruland, W.; Hoinkis,
2
(
(
̈
́
B.; Fratzl, P.; Paris, O.; Findenegg,
(
̈
(
(
(
(
79) Sears, V. F. Neutron News 1992, 3 (3), 26.
80) Glatter, O. J. J. Appl. Crystallogr. 1979, 12, 166.
81) Debye, P.; Bueche, A. M. J. Appl. Phys. 1949, 20 (1949), 518.
82) Debye, P.; Anderson, H. R.; Brumberger, H. J. Appl. Phys. 1957,
28 (1957), 679.
(
83) Fan, Y.; Kalyanaraman, J.; Labreche, Y.; Rezaei, F.; Lively, R. P.;
Real, M. J.; Koros, W. J.; Jones, C. W.; Kawajiri, Y.; Realff, M. J.; Koros,
W. J.; Jones, C. W.; Kawajiri, Y. Ind. Eng. Chem. Res. 2015, 54, 1783.
K
DOI: 10.1021/jacs.5b06823
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX