This article is licensed under CC-BY 4.

pubs.acs.org/jced Article

Measurement of Physicochemical Properties and CO2, N2, Ar, O2, and

H2O Unary Adsorption Isotherms of Purolite A110 and Lewatit VP
OC 1065 for Application in Direct Air Capture
May-Yin Ashlyn Low, David Danaci, Hassan Azzan, Robert T. Woodward, and Camille Petit*
Cite This: J. Chem. Eng. Data 2023, 68, 3499−3511 Read Online

ACCESS
Downloaded via KOREA ADVANCED INST SCI AND TECHLGY on May 13, 2025 at 02:13:18 (UTC).

Metrics & More Article Recommendations *

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

ABSTRACT: Direct air capture (DAC) using solid adsorbents has gained significant
attention as a carbon dioxide removal (CDR) technology to help limit global
temperature rise to below 2 °C. One large area of focus is the development of new
adsorbent materials for DAC. However, the necessary data needed to employ these
materials in process models for adsorbent screening are rarely available. Here, we
showcase Purolite A110, a commercially available amine-functionalized polymeric resin,
as a new candidate adsorbent for DAC and compare its properties to a current
benchmark, Lewatit VP OC 1065. For both materials, we report their chemical features
and composition, skeletal, particle, and bed density, total pore volume, particle porosity,
BET area, thermal stability, and specific heat capacity. We determine their equilibrium sorption properties by measuring the
volumetric CO2 isotherms at 288, 298, 308, 333, 343, 353, and 393 K, N2 and H2O isotherms at 288, 298, and 308 K, and Ar and O2
isotherms at 298 K. For CO2, N2, and H2O, we also present the corresponding isotherm model fitting parameters and heats of
adsorption. These data can help facilitate process modeling and optimization studies to properly assess these adsorbents at scale.

1. INTRODUCTION those by Khurana and Farooq,4 Rajagopalan et al.,5 and

The 2015 Paris Agreement saw 196 countries commit to limiting Farmahini et al.6 have shown that simple metrics such as these
global temperature rise to below 2 °C, and preferably below 1.5 are not sufficient for adsorbent screening. Instead, performance
°C. According to recent reports by the Intergovernmental Panel indicators determined from process modeling and optimization,
on Climate Change (IPCC), this scenario can only be achieved such as purity, recovery, productivity, and energy consumption,
with carbon dioxide removal (CDR) technologies.1 The use of should be used. In the case of DAC, these are especially
solid adsorbents to remove CO2 directly from the atmosphere, important to determine the net CO2 removal achieved by the
i.e., adsorption-based direct air capture (DAC), is a CDR process. To conduct process-scale evaluation, various material
technology that is actively being investigated. It is a challenging properties are needed, as outlined in Table 1.
process due to the ultradilute concentration of CO2 in the air It is extremely rare to find adsorbents in the literature with all
(∼0.04%vol or 0.04 kPa) and varying ambient conditions (e.g., of the necessary material properties reported, even when
temperature and relative humidity), but it offers a route toward considering numerous sources. Of the more than 200 proposed
long-term carbon storage and allows the amount of CO2 DAC adsorbents in recent years, only two materials, zeolite 13X
captured to be quantitatively measured. and Lewatit VP OC 1065, had sufficient data required for
A number of commercial adsorption-based DAC plants are process modeling. These include: CO2, N2, and H2O isotherms
currently operational, amounting to a CO2 capture capacity on measured at three or more temperatures from which isotherm
the order of ca. 10 kt year−1.2 However, this only equates to model fitting parameters and heats of adsorption could be
0.0002% of the CO2 removal goal of 4.8 Gt year−1 by 2050, determined; bed density; particle density; skeletal density and/
stipulated in the “below 2 °C” scenario (of which 2.7 Gt year−1 is or total pore volume; heat capacity; thermal conductivity; and
thought to be removed by engineered CDR technologies such as mass transfer coefficients at low CO2 composition. While 13X is
DAC).3 Hence, much work is still needed for the scale-up of a well-known commercial benchmark for postcombustion
these DAC processes. One aspect actively being researched is
the development of new adsorbent materials. From our recent
work,2 we determined that over 200 adsorbents had been Received: June 27, 2023
proposed for DAC in the literature since 2016, including silicas, Accepted: September 18, 2023
metal oxides, metal−organic frameworks (MOFs), zeolites, and Published: October 11, 2023
polymers. Some of the most reported metrics were the
adsorbents’ CO2 uptake at 0.04 kPa, or a CO2 isotherm at a
single temperature (usually 298 K). However, works such as
© 2023 The Authors. Published by
American Chemical Society https://doi.org/10.1021/acs.jced.3c00401
3499 J. Chem. Eng. Data 2023, 68, 3499−3511
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

Table 1. Summary of Adsorbate and Adsorbent Properties resins, Shu et al. investigated a commercial resin (AmberLite
Required in the Modeling of an Adsorption-Based DAC HPR4800 OH) using pH swing regeneration in an electro-
Processa chemical cell,27 which had a CO2 adsorption capacity of 1.76
mmol g−1 under humid conditions at 0.04 kPa CO2 and 298 K.
material property required process modeling output
The particle and bed densities of the adsorbent are available
equilibrium isotherms considering coadsorption total amount adsorbed and from the manufacturer,28 but other properties needed for
for ALL adsorbing components in ambient air desorbed
at multiple temperatures process modeling (see Table 1) were not reported in this study.
overall mass transfer coefficient rate of mass transfer and Elfving et al. studied an undisclosed resin supplied by Oy
adsorption Hydrocell Ltd. under various process cycles,29 regeneration
density (skeletal, particle, and bed) mass of solid and volume of conditions,30 and dry versus humid conditions.31,32 The
process contactor required
porosity (particle and bed) derived from pore total volume of gas in vessel
adsorbent consists of a polystyrene matrix functionalized with
volume estimation space and mass of primary amines, like Lewatit, with a dry CO2 adsorption capacity
adsorbent of 0.8 mmol g−1 at 0.04 kPa CO2 and 298 K. Single-component
heat capacity (adsorbent and adsorbed phase) heat required to perform CO2 and H2O isotherms at multiple temperatures, as well as
desorption
CO2−H2O coadsorption isotherms, have been measured. The
heat of adsorption (of all adsorbing required energy input for
components) desorption and process corresponding isotherm equations and heats of adsorption were
temperature changes also reported along with the bed density, particle density, and
overall heat transfer coefficient (related to rate of heat transfer in specific heat capacity. Parvazinia et al. studied another amine-
thermal conductivity of the adsorbent) material
a
functionalized polymeric resin, Purolite A109, for CO2 capture
Adapted from Low et al. (open access).2 at 2 kPa and 298 K,33 which had a CO2 adsorption capacity of
0.2 mmol g−1 at these conditions. More recently, Purolite A110,
carbon capture, it has a relatively low CO2 adsorption capacity of the successor of Purolite A109, appeared in the literature for the
0.2−0.3 mmol g−1 at 0.04 kPa CO2 and 298 K.7,8 first time for DAC applications.34 However, the authors
Lewatit VP OC 1065 (referred to as Lewatit from here) is a modified the material with a CuCl2 solution to form their
commercially available polymeric resin functionalized with Polyam−N-Cu2+ adsorbent, which would capture CO2 only
primary amines and is believed to be very similar to the under humid conditions via a carbonate reaction. This material
adsorbents used by the company Climeworks,9 one of the was reported to have a CO2 adsorption capacity of 5.1 mmol g−1
current commercial operators of adsorption-based DAC. It has a at 0.04 kPa CO2, 298 K, and 50% RH. Under the same
comparably high CO2 capacity of 1.0 mmol g−1 at 0.04 kPa CO2 conditions, unmodified Purolite captured 1.8 mmol g−1 of CO2,
and 298 K under dry conditions.10 This is due to its which is comparable to Lewatit at similar conditions (∼0.035
chemisorbent nature, where two amine groups react with and kPa, 298 K, 55% RH).9 However, the necessary data needed for
capture CO2 to form ammonium carbamate11 (see eq 1). process-scale evaluation, such as equilibrium isotherms and
heats of adsorption, have not been measured.
CO2 + 2RNH 2 RNH+3 + RNHCOO (1) In this study, we propose unmodified Purolite A110 (referred
In recent years, Lewatit has gained attention for its application to as Purolite from here) as a new candidate benchmark
in DAC, and it is often regarded as a benchmark adsorbent. The adsorbent for DAC and compare its properties to those of the
current literature consists of investigating the performance of current benchmark, Lewatit. Our objective is to measure all the
Lewatit using different cycle processes9,10,12−14 and contactor material and equilibrium sorption properties needed to perform
designs,15−20 the impact of climate variations,21,22 a few pilot- process modeling and compare the results to that of Lewatit. For
scale investigations,23,24 and some kinetic studies at low CO2 both Purolite and Lewatit, we report their chemical features,
compositions.9,25,26 chemical composition, skeletal, particle, and bed densities, total
In addition to gathering the data to fully evaluate the potential pore volume, particle porosity, BET area, thermal stability, and
of adsorbents at scale, expanding the group of easily accessible specific heat capacity. To analyze their sorption properties, we
commercial materials for DAC to cover the wide range of measure CO2, N2, and H2O isotherms at multiple temperatures
climate conditions will help accelerate the deployment of and Ar and O2 isotherms at a single temperature. For CO2, N2,
adsorption-based DAC technologies. In the field of polymeric and H2O, we also fit the corresponding isotherm models and

Table 2. Details of the Commercial Resins and Reference Chemicals Used in This Study
Aluminum oxide (γ-phase,
Lewatit VP OC 1065 Purolite A110 Zeolite 13X Silica−alumina 99.97%)
source Sigma-Aldrich Purolite Micromeritics Micromeritics Alfa Aesar
product code 94136−100G-F SR60210479 004−16843−00 004−16821−00 1344−28−1
lot number BCCC7234 T014Z/21/6 9266−33 A-501−53 X30G032
matrix styrene-divinylbenzene styrene-divinylbenzene
functional group primary amine primary amine
exchange capacity (eq L−1)a 2.1 2.0
water retention (wt %)a 52−58 60−66
particle size (μm)a 300−1250 300−1200
fines, <300 μm (vol %)a 3 1
uniformity coefficienta 1.8 1.7

a
Information provided by the manufacturers on the technical data sheets.35,36

3500 https://doi.org/10.1021/acs.jced.3c00401
J. Chem. Eng. Data 2023, 68, 3499−3511
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

determine heats of adsorption. While a mass transfer coefficient sample, tapped 20 times, and the excess swept off using a
is also needed for process modeling, kinetic sorption properties straight-edged spatula. The total weight of the container and
are outside the scope of this study and are not reported. sample was then measured on an analytical balance (OHAUS
Adventurer Pro), from which the weight of the empty container
2. MATERIALS AND METHODS was subtracted to obtain the mass of the sample. This was then
2.1. Chemicals. Details of the commercial resins inves- divided by the known volume of the cylinder, determined by
tigated in this study are provided in Table 2, along with weighing the mass of water required to fill its cavity, to determine
chemicals used as reference materials throughout the study. the final bed density. This procedure was repeated three times
The purities of the gases used for the study are as follows. For for each sample.
gas sorption measurements: CO2 (BOC, N5.0 grade), N2 Once the skeletal and bed densities were determined, and
(BOC, N6.0 grade), Ar (BOC, N5.0 grade), and O2 (BOC, assuming a spherical bed voidage (εbed) of 0.36,39 the particle
N5.0 grade). For pycnometry measurements: He (BOC, N5.0 density (ρparticle), total pore volume (Vtotal), and adsorbent
grade). For elemental analysis: He (Messer, N5.0 grade). For voidage (εparticle) of each sample were calculated from eqs 2−4:
thermogravimetric measurements: N2 (BOC, N4.8 grade) and bed
compressed air (BOC, industrial grade). For differential particle
=
1 bed (2)
scanning calorimetry (DSC) measurements: N2 (BOC, N5.2
grade). 1 1
2.2. Characterization of Chemical Features. Fourier Vtotal =
transform infrared (FTIR) spectra of the resins were collected particle skeletal (3)
by using an Agilent Cary 630 FTIR spectrometer equipped with
an attenuated total reflectance (ATR) accessory. Samples were Vtotal
particle =
first degassed ex situ at 393 K and 0.002 kPa for at least 12 h and 1/ particle (4)
then ground using a pestle and mortar before analysis. Sixteen
spectra were measured per sample, and an average spectrum was 2.4. Characterization of Thermal Properties. Thermog-
obtained over the wavenumber range of 400−4000 cm−1 with a ravimetric analysis (TGA) of the resins was carried out using a
resolution of 2 cm−1. Netzsch TG 209 F1 Libra thermal analyzer. Approximately 20
Elemental analysis was performed using a Eurovector EA mg of the untreated sample was placed in an alumina crucible
3000 CHNS-O elemental analyzer on samples weighing and subjected to heat treatment from room temperature to 1173
between 0.75 and 3.0 mg, which were weighed into tin vials (4 K at a ramp rate of 10 K min−1. This was done under either a
× 6 mm) prior to measurement. Samples were degassed ex situ protective N2 flow (100 mL min−1) or a mixture of air (100 mL
at 393 K and 0.002 kPa for at least 12 h prior to measurement, min−1) and a protective N2 flow (20 mL min−1). A correction
and each sample was run in triplicate. Accurate sample weighing run with an empty alumina crucible and identical analysis
was achieved using a microbalance (Sartorius, ME 5 OCE). The conditions was performed prior to these measurements to
operating temperatures for combustion and reduction were account for buoyancy effects.
1273 and 1023 K, respectively. He was used as the carrier gas in Specific heat capacities were measured using a PerkinElmer
all cases. DSC 8000 differential scanning calorimeter following a
2.3. Characterization of Textural Properties and methodology adapted from Moosavi et al.40 Samples were first
Density. N2 sorption isotherms at 77 K were measured using degassed ex situ at 393 K and 0.002 kPa for at least 12 h and then
a Micromeritics 3Flex porosity analyzer. Samples were first ground into powder using a pestle and mortar before analysis.
degassed ex situ at 393 K and 0.002 kPa for at least 12 h and then Approximately 6 mg of powder was placed in an aluminum DSC
degassed in situ at 393 K and 0.000002 kPa for 24 h. pan and crimped with a corresponding aluminum lid with a
Approximately 100−200 mg of dried samples was used for the manually pierced pinhole. The sample pan was placed into the
measurements. The specific surface area was determined using sample chamber, while an identical empty aluminum pan and lid
the Brunauer−Emmett−Teller (BET) method,37 with the (uncrimped) were placed into the reference chamber. The
Rouquerol criteria applied.38 The cumulative pore size external chiller was set to 253 K, and the chambers were then
distribution was determined using the density functional theory subjected to the following temperature profile: isothermal at 293
(DFT) model available in the 3Flex software, the “N2-DFT K for 5 min, heating ramp to 393 K at a rate of 10 K min−1, and
Model”, with a slit-shaped pore geometry kernel, and 0.10000 isothermal at 393 K for 15 min, all under a N2 flow of 40 mL
regularization. min−1. This profile was repeated without opening the sample
Skeletal densities (ρskeletal) were determined via helium and reference chambers, until three overlapping heat flow curves
pycnometry using a Micromeritics AccuPyc II 1340 pycnom- were obtained. Prior to each series of measurements, an empty
eter. Samples were first degassed at 393 K and 0.002 kPa for at aluminum DSC pan and lid and a sapphire standard were used to
least 12 h, and approximately 1.7 g of dried sample was used to obtain the baseline and reference curves, respectively. The same
fill a 3.5 cm3 sample holder. The samples were then exposed to procedure as outlined above was used for these measurements,
20 purge cycles of He dosing (∼134 kPa), followed by 10 although the aluminum pans were not crimped. Heat flow curves
measurement cycles under the same conditions. The skeletal for all measurements are shown in Figure S2. The specific heat
density values presented in this study are the averages capacity curves were then derived using the PerkinElmer DSC
determined from 10 measurement cycles. 8000 software using the last three overlapping heat flow curves
Bed densities (ρbed) were measured by filling a 3D-printed for each component (baseline, reference, and sample). This
cylindrical container (2.0 cm × 2.0 cm cube with a 1.6 cm methodology was verified by measuring the heat capacities of γ-
diameter by 1.2 cm height cylindrical cavity; see Figure S1 for alumina powder and the α-alumina (sapphire) disc provided by
the image) with a dried sample (degassed at 393 K and 0.002 the DSC instrument manufacturer. Both experimental results
kPa for at least 12 h). The container was overfilled with the dried were comparable to the literature values provided by the
3501 https://doi.org/10.1021/acs.jced.3c00401
J. Chem. Eng. Data 2023, 68, 3499−3511
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

Table 3. Sources of Uncertainties for the Experimental Measurements of the Equilibrium Gas Sorption Isotherms
parameter equipment model tolerance
pressure (P) pressure transducer, 1000 Torr TE 85−015A 0.001 · P
pressure transducer, 10 Torr MKS 626B11T 0.0025 · P
pressure transducer, 1 Torr MKS 626B.1T 0.005 · P
mass analytical balance OHAUS Adventurer Pro 0.0001 g
temperature thermometer Brannan 44/010/7 1.0 K
recirculating cooler Julabo F250 0.1 K

National Institute of Standards and Technology (NIST)41 were determined using the smoothing spline interpolation. A
(Figure S3). The measured specific heat capacity data for linear regression was then applied to the ln(P) versus 1/T data
Lewatit and Purolite were fitted to a known specific heat capacity for each value of n using the LINEST function in Microsoft
equation using Microsoft Excel 2016 Solver. Excel. The heat of adsorption for each value of n (ΔHn) was then
2.5. Analyses of Gas and Vapor Sorption Properties. calculated by using eq 5, where the derivative corresponds to the
Gas and vapor sorption isotherms were measured volumetrically slope of the linear regression, and R is the universal gas constant.
using a Micromeritics 3Flex porosity analyzer for CO2 at 288, ÄÅ ÉÑ
ÅÅÅ lnP ÑÑÑ
298, 308, 333, 343, 353, and 393 K, for N2 and H2O at 288, 298, Hn = RÅÅÅ ÑÑ
and 308 K, and for Ar and O2 at 298 K. Samples were first ÅÅÇ (1/T ) ÑÑÑÖ (5)
n
degassed ex situ at 393 K and 0.002 kPa for at least 12 h and then
degassed in situ at 393 K and 0.000002 kPa for 24 h. For CO2 The limiting heats of adsorption (ΔH0) for CO2, N2, and H2O
measurements, approximately 100−200 mg of dried sample was were calculated using the Henry constants determined for each
used, and fresh sample was used for each temperature isotherm temperature. Virial plots (n vs ln(P/n)) were created,
measurement. For H2O measurements, approximately 100− and the low loading data at each temperature were fitted to a
200 mg of dried sample was used, and the same sample was used linear equation. The negative exponent of the intercept gives the
to measure all temperatures. For N2, Ar, and O2 measurements, Henry constant (K). A linear regression can then be applied to
approximately 1.0−1.5 g of dried sample was used, and the same lnK versus 1/T, from which the slope can be used to calculate
sample was used for all measurements, which were done in the ΔH0, as shown in eq 6.
ÅÄÅ ÑÉ
gas order listed. A glass stirring rod was also placed in each Å lnK ÑÑÑ
sample tube for these measurements to reduce the free space H0 = RÅÅÅÅ ÑÑ
ÅÅÇ (1/T ) ÑÑÑÖ (6)
volume. For measurements at and below 308 K, temperature 0
control was achieved by submerging the sample tubes in a dewar
containing a water−glycol-based bath fluid. The bath temper- 3. RESULTS AND DISCUSSION
ature was measured with a glass thermometer and maintained by
3.1. Material Properties. 3.1.1. Chemical and Textural
using a Julabo F250 recirculating cooler. For measurements at
Properties. Lewatit and Purolite present as spherical beads with
and above 333 K, temperature control was achieved using the
50−60% water content and particle sizes between 0.3−1.2
heating mantle provided with the 3Flex instrument.
mm.35,36 Images of the beads before and after drying are shown
The sources of uncertainties in the experimental method
in Figure S6. Both Lewatit and Purolite are reported by their
described above are summarized in Table 3. The method was
manufacturers to consist of styrene-divinylbenzene polymer
validated using the reference samples and measurement files backbones functionalized with primary amines, which should
provided by Micromeritics. CO2 adsorption and desorption drive the CO2 adsorption. We confirmed the presence of these
isotherms for a zeolite 13X reference material at 273 K (Figure functionalities for both resins using FTIR analysis (Figure 1).
S4) and N2 adsorption and desorption isotherms for a silica− The spectra exhibit bands between 1625 and 1500 cm−1
alumina reference material at 77 K (Figure S5) were measured
and compared with the reference data provided by Micro-
meritics. Independent aliquots of both materials were measured
on all three ports of our instrument to confirm the data
reproducibility.
Temperature-dependent Toth (td-Toth), chemi-physisorp-
tion (CP), and virial isotherm model fits for the CO2 adsorption
data, single-site Langmuir (SSL) isotherm model fits for the N2
data, and Guggenheim−Anderson−de Boer (GAB) isotherm
model fits for the H2O adsorption data were performed using the
procedure described in our previous works.42,43 This procedure
was carried out with MATLAB R2020a (The Mathworks Inc.)
using the in-house software package isothermFittingTool.44
The isosteric heats of adsorption for CO2, N2, and H2O were
calculated by using the van’t Hoff equation for specified loading
(n) and corresponding pressure (P) values. The isotherm data
(P vs n) at each temperature (T) were first interpolated using
smoothing spline interpolation in MATLAB. m points with 0.01
mmol g−1 intervals were generated over the loading range Figure 1. FTIR spectra for dried and ground Lewatit and Purolite
common to all temperatures, and the corresponding P values resins.

3502 https://doi.org/10.1021/acs.jced.3c00401
J. Chem. Eng. Data 2023, 68, 3499−3511
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

Table 4. Summary of CHN Elemental Analysis for Lewatit

and Purolite Resins, With Uncertainty Bounds for a 95%
Confidence Interval
sample C (%wt) H (%wt) N (%wt)
Lewatit 82.5 ± 0.1 8.1 ± 0.1 8.3 ± 0.1
Purolite 79.6 ± 0.4 8.2 ± 0.1 10.2 ± 0.3

corresponding to aromatic C�C bonds, bands between 3400

and 3300 cm−1 and 895 and 650 cm−1 characteristic of NH2
vibrations, and bands around 1200 cm−1 characteristic of C−N
bonds.45 We then used elemental analysis to quantify the
nitrogen (N) content in each resin, which is representative of the
level of amine functionalization. As seen in Table 4, Purolite has
a higher amine content (10.2 wt % of N) than Lewatit (8.3 wt %
of N). The N content measured for Lewatit is not far off from the
value of 9.5 wt % reported by Yu et al.46 Assuming that CO2
adsorption only occurs under dry conditions where two N atoms
are needed to capture one CO2 molecule (see eq 1), the
theoretical CO2 capacity for Lewatit and Purolite would be 2.96
and 3.64 mmol g−1, respectively. We refer to these values later in
our study as we quantify the CO2 sorption.
We used N2 sorption at 77 K to determine the BET area and
pore size distribution for Lewatit and Purolite, which are
reported as macroporous resins by the manufacturers. Images of
the porous surface of both resins are shown in Figure S7. Both
resins have similarly low BET areas of 31 and 28 m2 g−1,
respectively, indicative of macroporous materials. We can also
observe from the N2 sorption isotherms and pore size
distribution the absence of micropores in the materials (Figure Figure 2. Porosity analysis of Lewatit and Purolite resins. (a) N2
2). The total pore volumes of Lewatit and Purolite, obtained by a sorption isotherms at 77 K (filled symbols = adsorption; open symbols
direct reading of the N2 isotherms at 0.99 P/P°, are 0.34 and 0.23 = desorption). (b) Cumulative pore size distribution obtained from N2
sorption at 77 K with DFT analysis.
cm3 g−1, respectively. This, however, is an underestimation of
the actual total pore volume for these materials as N2 sorption at
Table 5. Summary of Density (Skeletal, Particle, and Bed)
77 K is ill-suited for macropore analysis. Instead, we estimated
and Textural Property (Total Pore Volume, Particle Voidage,
the total pore volumes of Lewatit and Purolite to be 0.43 and
and BET Area) Values for Dried Lewatit and Purolite Resinsa
0.64 cm3 g−1, respectively, by determining the bed and skeletal
densities of the materials. Using He pycnometry, we found that Lewatit Purolite
Purolite and Lewatit exhibit similar skeletal densities (1.15 vs ρskeletal(g cm−3) 1.1231 ± 0.0002 1.1471 ± 0.0002
1.12 g cm−3, respectively), but Purolite has a bed density lower ρparticle(g cm−3) 0.744 ± 0.001 0.663 ± 0.002
than that of Lewatit (0.42 vs 0.48 g cm−3, respectively). This ρbed(g cm−3) 0.476 ± 0.001 0.424 ± 0.002
result implies that a lower mass of Purolite can fill a specified bed Vtotal(cm3 g−1) 0.454 ± 0.001 0.637 ± 0.001
volume compared to that of Lewatit. Purolite also has a higher εparticle 0.338 ± 0.001 0.422 ± 0.001
total pore volume, higher porosity, and lower particle density SBET(m2 g−1) 31 28
than Lewatit. The values of bed density, particle density, and a
Uncertainty bounds for a 95% confidence interval are provided when
porosity are typically needed as inputs for process modeling and available.
are reported in Table 5, along with the total pore volumes and
BET areas. Literature values for these properties are provided in
Table S1 where available for comparison. conditions correspond to the degradation of the polymer
3.1.2. Thermal Properties. We used thermogravimetric backbone.
analysis (TGA) to determine the thermal stability of the resins, We used differential scanning calorimetry (DSC) to
which dictates the maximum temperature that can be used for determine the specific heat capacities of the resins in the
desorption in a DAC process without degrading the material. temperature range 298−393 K, which is a possible range of
The manufacturer of Lewatit recommends a maximum temperatures that the materials might experience in a DAC
operating temperature of 373 K,35 while none is recommended process. The heat capacity is a necessary input in process
for Purolite. As seen in Figure 3a, both resins contain adsorbed modeling as it relates to the energy input needed to heat an
water (weight loss <373 K), with Purolite having approximately adsorbent during the desorption process, where an adsorbent
20% higher water content than Lewatit. Small reductions in mass with a higher heat capacity would require more energy to be
are observed for both Lewatit and Purolite, starting at 398 K heated to a set temperature. As seen in Figure 3b, both Lewatit
under both N2 and air atmospheres (see Figure S8), likely and Purolite have similar heat capacities, as expected due to their
associated with amine degradation.47 The larger decreases in similar chemical structures. Both values are similar to the heat
mass between 573 to 673 K for both materials under both capacity of polystyrene in the same temperature range (1.2−1.7
3503 https://doi.org/10.1021/acs.jced.3c00401
J. Chem. Eng. Data 2023, 68, 3499−3511
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

Figure 4. Equilibrium adsorption (filled symbols) and desorption

Figure 3. Thermal analysis of the Lewatit and Purolite resins. (a) TG (hollow symbols) isotherms for CO2 measured at 288, 298, 308, 333,
curves under N2 (solid lines) and air (dotted lines) atmospheres up to 343, 353, and 393 K up to 100 kPa for (a) Lewatit and (b) Purolite.
1173 K. (b) Specific heat capacities of Lewatit and Purolite measured as
a function of temperature from 298 to 393 K compared to the reported reached by repeating the adsorption and desorption measure-
values of polystyrene, where the darker solid lines represent the fitted ments at 308 K, with the equilibrium interval increased from 60
results using eq 7. to 90 s, and observed no change in the results (Figure S9).
The results demonstrate that Purolite adsorbs more CO2 than
Table 6. Fitting Parameters for the Specific Heat Capacities Lewatit at 0.04 kPa CO2 pressure, which is representative of the
of Lewatit and Purolite CO2 feed stream for a DAC process. At an adsorption
parameter unit Lewatit Purolite temperature of 298 K and in the presence of 0.04 kPa CO2,
a J K g−1 −3.23 × 104 −5.86 × 104
the gravimetric CO2 adsorption capacity for Purolite versus
b J g−1 K−2 0.00227 0.000994
Lewatit is approximately 1.3 mmol g−1 versus 0.95 mmol g−1.
c J g−1 K−1 −0.994 −1.69
Considering the adsorbents’ bed densities (see Table 5),
Purolite also has a higher volumetric CO2 adsorption capacity
of 0.55 mmol cm−3bed compared to that of Lewatit (0.45 mmol
J g−1 K−1),48 which forms the backbone of both resins, and agree cm−3bed). This trend correlates with the higher N content
well with previously measured specific heat capacity values for observed for Purolite compared to Lewatit. CO2 isotherms for
Lewatit of 1.6 J g−1 K−1.49 Our experimental data can be Purolite have not yet been reported in the literature, while CO2
modeled using eq 7 shown below:48 isotherms for Lewatit have been measured before.9,25,51,52 We
cp = a· T 2
+ b· T c compared our adsorption data obtained at 298 K with that
(7)
obtained by Young et al.9 in Figure S10a. The data align very well
where cp is the specific heat capacity, T is the temperature, and a, at low CO2 pressures, but the CO2 loading reported by Young et
b, and c are constants to be fitted. The fitted parameters for al. at pressures closer to 100 kPa is higher than that of our
Lewatit and Purolite are listed in Table 6. experimental data. This deviation could be due to the different
3.2. Sorption Properties. 3.2.1. CO2 Sorption. We sorption measurement techniques used, i.e., gravimetrically
measured the equilibrium CO2 adsorption and desorption (Young et al.) versus volumetrically (our study).
isotherms up to 100 kPa for Lewatit and Purolite at 288, 298, Hysteresis is observed for both Lewatit and Purolite in Figure
308, 333, 343, 353, and 393 K (Figure 4), with more than 20 data 4. In both cases, the extent of hysteresis decreases as the
points measured below 1 kPa at most temperatures. The temperature increases, which is indicative of the chemisorbent
adsorption data points for each temperature for Lewatit and nature of the materials. At lower temperatures, there is
Purolite are provided in Table A.1, while all adsorption and insufficient energy available to remove the chemisorbed CO2
desorption data are provided as AIF files50 and CSV files in the at a given pressure, while at higher temperatures (e.g., 393 K),
Supporting Information. We ensured that equilibrium was the process is reversible. We attribute this behavior to a true
3504 https://doi.org/10.1021/acs.jced.3c00401
J. Chem. Eng. Data 2023, 68, 3499−3511
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

equilibrium effect rather than a kinetic effect. Indeed, we

repeated the measurements with a longer equilibration interval
(Figure S9) and did not observe any change in the degree of
hysteresis in the isotherms.
We also measured adsorption and desorption isotherms at
308 K for both materials approximately a year after our original
measurements on the same batch of resins to test if there was any
degradation of the adsorbents. As seen from Figure S11, there is
no discernible difference between the two measurements for
Lewatit; however, there is a slight decrease in loading in the later
Purolite measurement compared to the original, which could be
attributed to minor degradation of the material during storage. A
pilot-scale study by van Paasen et al.53 using Lewatit for
postcombustion CO2 capture showed that approximately 10% of
the adsorbent degradation after 33 days was due to shelf-life
instability, while the major contributor was oxidative degrada-
tion at high operating temperatures during regeneration. These
and other sources of oxidative degradation during a DAC
process cycle were recently described by Carneiro et al.54
Isotherm model equations that represent the experimental
data well are needed to implement the data in process modeling.
All previous studies with Lewatit have used the temperature-
dependent Toth (td-Toth) isotherm model to describe the
loading n of CO2, given by eqs 8−11:
q (T )· b(T ) ·P
n=
[1 + (b(T )·P) (T )]1/ (T )
(8)
T
q (T ) = q ,0
·exp[ (1 )]
T0 (9)
Figure 5. Equilibrium adsorption isotherms for CO2 measured at 288,
H 298, 308, 333, 343, 353, and 393 K up to 100 kPa for (a) Lewatit and
b(T ) = b0 ·exp( ) (b) Purolite with a log-scale of pressure. Solid lines represent the fitting
RT (10)
results from the CP isotherm model, while dashed lines represent the
T0 fitting results from the td-Toth isotherm model. The fitting parameters
(T ) = 0 + (1 ) of the isotherms are found in Table 7.
T (11)
where P is the pressure, T is the temperature, q∞(T) is the accounts for both physisorption and chemisorption was
maximum CO2 capacity, q∞,0 is the maximum CO2 capacity at previously proposed by Serna-Guerrero et al. to describe CO2
the reference temperature T0, b(T) is the CO2 adsorption adsorption on triamine-grafted pore-expanded MCM-41 from
coefficient defined by a constant b0 and the heat of adsorption 0.1 to 2000 kPa55 and used by Elfving et al. for CO2 adsorption
ΔH, R is the universal gas constant, τ(T) is the surface on a proprietary amine-functionalized resin from 0.01 to 0.5 kPa
heterogeneity parameter, τ0 is the surface heterogeneity for DAC applications.32 This model is given by the following
parameter at T0, and χ and α are the factors determining the equation:
temperature dependencies of q∞(T) and τ(T). As seen in Figure
5, this model does not fit our measured data very well for either n = n phys + nchem
ÄÅ ÉÑ ÄÅ ÉÑ
Lewatit or Purolite at the lower temperatures of 288, 298, and ÅÅ ÑÑ ÅÅ ÑÑ
ÅÅ ÑÑ ÅÅ ÑÑ
ÅÅ q (T )·b(T )· P ÑÑ ÅÅ q (T )·b(T )·P ÑÑ
308 K. The fitted parameters are found in Table 7, while the = ÅÅÅ Ä (T ) Ñ
ÑÑ + ÅÅÅ Ä (T ) Ñ
ÑÑ
ÅÅÅ ÅÅÅ ÑÉ1/ ÑÑ ÅÅÅ ÅÅÅ ÑÉ1/ ÑÑ
upper and lower bounds imposed on each parameter during the ÅÅ ÅÅ1 + (b(T )·P) (T ) ÑÑÑÑ ÑÑ
ÑÑÖ ÅÅ ÅÅ1 + (b(T )· P) (T )ÑÑÑÑ ÑÑ
ÑÑÖ
ÅÇ Ç Ö phys ÅÇ Ç Ö chem
fitting are summarized in Table S2. The fitting parameters
determined by Young et al.9 (Table S3) are also unable to (12)
represent our data well, as shown in Figure S10b. This may be where q∞(T), b(T), and τ(T) are given by eqs 9−11. However,
due to varying factors such as a different temperature range and a to implement this model, the contributions from nphys and nchem
greater number of data points, particularly in the low-pressure must be separately determined experimentally. Serna-Guerrero
region below 1 kPa. et al. did so by measuring adsorption isotherms for ungrafted
Given that td-Toth does not represent our data well, we MCM-41 and assumed that this was the contribution for nphys,
considered other approaches. While the dominant mechanism while Elfving et al. used a N2 purge with and without heat to
in amine-functionalized adsorbents is known to be chemisorp- separately desorb CO2 captured by chemisorption versus
tion via the reaction shown in eq 1, previous works have also physisorption, respectively. Other studies such as those by Pai
observed a non-negligible contribution from physisorption on et al.,57 Hefti et al.,58 and Hughes et al.59 have also proposed
the adsorbent surface, which would be more prevalent at lower solutions for modeling isotherms with both chemisorption and
temperatures.55,56 This seems to also apply to Lewatit and physisorption contributions, either by using a weighted isotherm
Purolite, as their sorption isotherms do not plateau in our model or modeling the chemisorption and physisorption
measured pressure range. As such, an isotherm model which separately, though the application here was for S-shaped
3505 https://doi.org/10.1021/acs.jced.3c00401
J. Chem. Eng. Data 2023, 68, 3499−3511
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

Table 7. Fitting Parameters With Uncertainty Bounds for a 95% Confidence Interval for the td-Toth and CP Isotherm Models for
CO2 Adsorption for Lewatit and Purolite
equation parameter unit Lewatit Purolite
−1
td-Toth q∞,0 mmol g 3.00 ± 0.05 3.74 ± 0.05
b0 (× 10−11) kPa−1 2.0 ± 0.2 3.4 ± 0.3
−ΔH kJ mol−1 70.1 ± 0.3 70.1 ± 0.2
T0 K 298.15 298.15
τ0 0.431 ± 0.009 0.401 ± 0.006
α 1.4 × 10−10 ± 1.1 × 10−2 1.43 × 10−4 ± 5.8 × 10−2
χ 0.7 ± 0.2 0.3 ± 0.1
CP q∞,0 mmol g−1 2.56 ± 0.01 2.91 ± 0.02
b0 (× 10−16) kPa−1 1.05 ± 0.03 0.96 ± 0.05
−ΔH kJ mol−1 105.51 ± 0.09 105.2 ± 0.1
T0 K 298.15 298.15
τ0 0.343 ± 0.02 0.401 ± 0.004
α 1.39 ± 0.04 1.99 ± 0.10
χ 1.63 ± 0.05 2.52 ± 0.09
q∞,c mmol g−1 1.05 ± 0.03 3.56 ± 0.09
b0,c (× 10−4) kPa−1 1.03 ± 0.05 21 ± 1
−ΔHc kJ mol−1 13.5 ± 0.1 7.0 ± 0.2
Ea kJ mol−1 0.60 ± 0.06 3.13 ± 0.07

isotherms from diamine-appended MOFs. For our purposes, we are summarized in Table S2. The virial equation provides a
propose a modification of eq 12 as follows, which we refer to reasonable fit of the data; however, it is unable to account for
here as the chemi-physisorption (CP) isotherm model. competitive adsorption when used in process modeling and is
often difficult to implement due to the need for the equation to
n = n phys + nchem ·
ÄÅ ÉÑ be solved numerically as it returns pressure as a function of
ÅÅ ÑÑ ÅÄÅ q · b(T )·P ÑÉÑ
ÅÅ ÑÑ ÅÅ ,c ÑÑ loading, rather than the conventional loading as a function of
ÅÅ q (T ) ·b(T )· P ÑÑ ÑÑ · expjij Ea zyz
= ÅÅÅ Ä É1/
ÑÑ
Ñ + ÅÅÅÅ Ñ j z pressure. As a result, the CP isotherm model we propose is likely
ÅÅ ÅÅ
ÅÅ ÅÅ1 + (b(T )·P) (T ) ÑÑÑÑ
(T ) Ñ ÑÑ ÅÅ 1 + b(T )· P ÑÑÑ k RT {
ÅÅÇ ÅÇ ÑÖ ÑÑ ÅÇ ÑÖchem the most appropriate method to use to represent our measured
ÑÖ
phys
data.
(13)
3.2.2. N2, Ar, and O2 Sorption. We measured N2 sorption
where q∞,c is the maximum CO2 capacity associated with isotherms up to 100 kPa for Lewatit and Purolite at 288, 298,
chemisorption, Ea is the activation energy, and again q∞(T), and 308 K (Figure 6, desorption shown in Figure S13a,b), as well
b(T), and τ(T) are given by eqs 9−11. Here, the contribution as Ar and O2 isotherms up to 100 kPa at 298 K (Figure 6,
from physisorption is given by the td-Toth isotherm model, desorption shown in Figure S13c,d). These are all gases present
while the contribution from chemisorption is given by the single- in air at concentrations higher than CO2. They may be
site Langmuir (SSL) model. We assume that the transition competitive for CO2 adsorption sites or adsorb on the backbone
between physisorption and chemisorption follows an Arrhenius and impact the achievable product purity. The adsorption data
behavior given by η (i.e., at a given temperature, some fraction of points for each adsorbate and temperature for Lewatit and
molecules will have sufficient activation energy to participate in Purolite are provided in Table A.2, while all experimental
chemisorption). Thus, the model favors physisorption at lower adsorption and desorption data are provided as AIF files50 and
temperatures and increases the contribution of chemisorption as CSV files in the Supporting Information. The desorption data
the temperature increases, while still allowing for the exothermic shown in Figure S13 exhibit an irregular pattern which we
nature of adsorption at even higher temperatures. As shown in attribute to the low loading levels, which are on the same order
Figure 5, the model provides an improved fit of the experimental of magnitude as the error of the porosity analyzer.
data. We performed an F-test between the CP and td-Toth Compared with CO2 adsorption, the N2 , Ar, and O2
isotherm model fitting results and confirmed that the improved adsorption levels are almost negligible. The SSL isotherm
results were indeed significant and not due to additional fitting model shown in eq 15 can be used to represent the N2 isotherm
parameters (ps S16−S17). The fitted parameters can be found experimental data, and the fitting parameters are reported in
in Table 7, while the upper and lower bounds imposed on each Table 8
parameter during the fitting are summarized in Table S2.
As the above CP isotherm model has not yet been thoroughly q ·b(T )·P
n=
validated, we also fitted our experimental data to the well-known 1 + b(T ) · P (15)
virial equation as an alternative (Figure S12), given by
m m2 where q∞ is the maximum CO2 capacity, b(T) is given by
1 1 equation 10, n is the loading, P is the pressure, and T is the
ln(P /n) = · aj ·n j + bj · n j
T j= 0 temperature. The upper and lower bounds imposed on each
j =0 (14)
parameter during the fitting are summarized in Table S2.
where a and b are the characteristic virial coefficients, n is the Interestingly, both Lewatit and Purolite are selective to O2
loading, P is the pressure, T is the temperature, m1 = 3, and m2 = compared to N2, where many traditional adsorbents such as
1. The fitted parameters can be found in Table S4, and the upper zeolites have the opposite performance. Although their O2
and lower bounds imposed on each parameter during the fitting adsorption capacities are low, these materials may have potential
3506 https://doi.org/10.1021/acs.jced.3c00401
J. Chem. Eng. Data 2023, 68, 3499−3511
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

Figure 6. Equilibrium adsorption isotherms for N2 (squares) measured

at 288, 298, and 308 K and for Ar (diamonds) and O2 (triangles) at 298
K up to 100 kPa for (a) Lewatit and (b) Purolite. Solid lines represent
the SSL isotherm fitting results (fitting parameters can be found in
Table 8). Figure 7. Equilibrium adsorption (filled symbols) and desorption
(hollow symbols) isotherms for H2O measured at 288 K up to 1.6 kPa
and 298 and 308 K up to 2.3 kPa for (a) Lewatit and (b) Purolite.
applications in air separation where it would be more energy-
and cost-effective to adsorb O2 (21%vol in air) versus N2 (78%vol
in air). To the best of our knowledge, O2 isotherms have not
been previously measured for Lewatit, and so this characteristic
While directly measuring CO2−H2O coadsorption isotherms is
was not observed. In addition, the O2 isotherm on Purolite at
ideal, it remains experimentally challenging; for this study, we
298 K does not appear to intersect with the origin. This may be
chose to only measure unary water sorption isotherms.
an indication for the degradation of, or reaction with, the amine
The water adsorption behavior of both resins follows a Type
functionalization of the sorbent. It also aligns with the impact of
III isotherm behavior according to IUPAC classification,38
oxidative degradation of the amines on the CO2 capacity, as
which represents unrestricted multilayer adsorption on macro-
discussed in Section 3.2.1.
porous materials. When the loading data are plotted against the
3.2.3. H2O Sorption. We measured the H2O adsorption and
adsorption potential, ϵ, given by
desorption isotherms for Lewatit and Purolite at 288 K up to 1.6
kPa and at 298 and 308 K up to 2.3 kPa (Figure 7). The = R ·T ·ln(x) (16)
adsorption data points for each temperature for Lewatit and
Purolite are provided in Table A.3, while all experimental where R is the universal gas constant, T is the temperature, and x
adsorption and desorption data are provided as AIF files50 and is the relative pressure, the results collapse on to a characteristic
CSV files in the Supporting Information. Water adsorption curve typical of nonmicroporous adsorbents (Figure S14).62
behavior is particularly important to understand for amine- We used the Guggenheim−Anderson−de Boer (GAB) model
functionalized adsorbents as the presence of H2O from the to fit the experimental data as a function of relative pressure for
relative humidity in the air is known to increase the CO2 uptake both materials, as used by Young et al. for Lewatit.9 The model is
of these materials as well as minimize amine degradation.11,60,61 as follows:

Table 8. Fitting Parameters With Uncertainty Bounds for a 95% Confidence Interval for the SSL Isotherm Model for N2
Adsorption for Lewatit and Purolite

equation parameter unit Lewatit Purolite

SSL q∞ mmol g−1 0.315 ± 0.004 0.0419 ± 0.0002
b0 (×10−8) kPa−1 5.06 ± 0.06 75.6 ± 0.05
−ΔH kJ mol−1 20.30 ± 0.03 18.65 ± 0.02

3507 https://doi.org/10.1021/acs.jced.3c00401
J. Chem. Eng. Data 2023, 68, 3499−3511
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

qm · k· c·x Their parameters reasonably fit our measured 308 K data but
n= deviate from our data at 288 and 298 K. This could be because
(1 k ·x)(1 + (c 1)·k ·x) (17)
Young et al. fitted their parameters to experimental data
iE E10 + yz collected at a wider and higher temperature range, from 298 to
c = expjjj 1 zz 373 K.
k RT { (18)
3.2.4. Isosteric Heats of Adsorption. We calculated the
iE E10 + yz limiting heat of adsorption (Table 10), as well as the heat of
k = expjjj 2 9 zz adsorption as a function of loading (i.e., isosteric heat of
k RT { (19) adsorption), of CO2, N2, and H2O for both Lewatit and Purolite
E1 = C exp(D·T ) (20) (Figure 9). The intermediate results obtained during the
calculation of the limiting heats of adsorption, as described in
E2 9 = F + G·T (21) Section 2.5, are presented in Figures S16−S21. Confidence
intervals associated with the heat of adsorption as a function of
E10 + = 44.38·T + 57, 220 (22) loading calculations are presented in Figures S22 and S23, along
where n is the water loading on the resins, x is the relative with the additional data calculated using alternative isotherm
pressure, qm is the water loading of the monolayer, and k and c temperatures for CO2. The smoothing parameter values applied
are affinity parameters governed by the heats of adsorption of the for each adsorbent−adsorbate combination, as described in
first (E1), second to ninth (E2−9), and tenth and higher (E10+) Section 2.5, are presented in Table S6. The heats of adsorption
layers of adsorbed water. E10+ is also equivalent to the heat of serve as inputs in process models and link to the energy
condensation of water, and C, D, F, and G are constants to be requirements for desorption in a DAC process, where a higher
fitted. The fitting results are shown in Figure 8, and the fitting heat of adsorption implies a higher energy needed to regenerate
the adsorbent. Both materials have high CO2 heats of adsorption
characteristic of chemisorbents. At the loading levels corre-
sponding to 0.04 kPaa CO2, Lewatit and Purolite have heats of
adsorption of ∼81 and 69 kJ mol−1, respectively. The data
presented in Figure 9a were obtained using the 333, 343, 353,
and 393 K CO2 isotherms. The isosteric heats of adsorption
calculated using the lower temperatures of 288, 298, and 308 K
are shown in Figure S22 and show an eventual transition to the
physisorption range, commonly said to be below 50 kJ mol−1.
Also characteristic of many chemisorbents, both materials
exhibit low N2 heats of adsorption of ∼20 kJ mol−1. The N2 heat
of adsorption stays relatively constant as a function of loading,
indicating homogeneous coverage of the adsorbent surface.
For H2O, the heat of adsorption steeply decreases from 60 kJ
mol−1 between 0 and 0.5 mmol g−1 loading, to a plateau of
approximately 40 kJ mol−1, which is close to the heat of
condensation of water of 44.0 kJ mol−1 at 298 K given by eq 22
and corroborated by NIST.63 This trend possibly reflects two
different sorption mechanisms consisting of a monolayer of
resin−H2O interactions, followed by multilayers of H2O−H2O
interaction.

4. CONCLUSIONS
In this study, we endeavored to experimentally determine as
many of the material and sorption properties needed for process-
scale evaluation of DAC adsorbents as possible for Purolite
A110, a commercially available amine-functionalized polymeric
resin, and compared its properties to those of Lewatit VP OC
1065, a current benchmark adsorbent for DAC. Lewatit and
Purolite have similar chemical features, specific heat capacities,
Figure 8. Equilibrium adsorption isotherms for H2O measured at 288, thermal stability, and skeletal density. Purolite has lower bed and
298, and 308 K as a function of relative humidity for (a) Lewatit and (b) particle densities than Lewatit and a higher total pore volume
Purolite. Solid lines represent the GAB isotherm fitting results (fitting and porosity. In terms of its gas sorption performance, Purolite
parameters are listed in Table 9). has a higher gravimetric and volumetric CO2 adsorption
capacity than Lewatit, attributed to its higher N content, as
parameters are reported in Table 9, while the upper and lower well as higher H2O adsorption, while both materials have low
bounds imposed on each parameter during the fitting are and comparable adsorption of N2, Ar, and O2. Purolite has a
summarized in Table S2. A comparison of our adsorption data lower CO2 heat of adsorption compared to Lewatit and a similar
obtained at 298 K with that obtained by Young et al.9 is shown in H2O heat of adsorption. This trend suggests potentially lower
Figure S15a, and we observe good agreement between the two energy requirements for desorption and thus lower process
data sets. An overlay of their fitted isotherm parameters (Table operating costs, but this will ultimately depend on actual
S5) with our experimental data is also shown in Figure S15b. loadings. In our work, we also proposed an empirical CO2
3508 https://doi.org/10.1021/acs.jced.3c00401
J. Chem. Eng. Data 2023, 68, 3499−3511
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

Table 9. Fitting Parameters With Uncertainty Bounds for a 95% Confidence Interval for the GAB Isotherm Model for H2O
Adsorption for Lewatit and Purolite
equation parameter unit Lewatit Purolite
−1
GAB qm mmol g 3.23 ± 0.02 7.75 ± 0.07
C kJ mol−1 46.83 ± 0.04 45.79 ± 0.04
D K−1 0.02446 ± 0.00008 0.02441 ± 0.00008
F kJ mol−1 53.31 ± 0.01 52.52 ± 0.01
G kJ mol−1 K−1 −0.03268 ± 0.00003 −0.03108 ± 0.00004

Table 10. Limiting Heats of Adsorption of CO2, N2, and H2O be used for the numerical expression of the sorption isotherms.
for Lewatit and Purolite with Uncertainty Bounds for a 95% Based on the data we have collected, we propose that Purolite
Confidence Interval A110 is indeed a suitable adsorbent for DAC. With this work, we
hope we have provided the adsorption community with the
Lewatit Purolite
necessary information to conduct process-scale modeling and
−ΔH0, CO2 (kJ mol−1) 107 ± 8 104 ± 9 optimization to assess the process performance of Purolite for
−ΔH0, N2 (kJ mol−1) 16 ± 8 13 ± 13 adsorption-based DAC and continue to advance the develop-
−ΔH0, H2O (kJ mol−1) 49 ± 29 50 ± 42 ment and scale-up of this technology.

■
*
ASSOCIATED CONTENT
sı Supporting Information
The Supporting Information is available free of charge at
https://pubs.acs.org/doi/10.1021/acs.jced.3c00401.
Additional information on bed density, specific heat
capacity, and thermal stability measurements; validation
of the porosity analyzer using reference materials; images
of Lewatit and Purolite resin beads; literature value
comparisons of textural properties, CO2 adsorption, and
H2O adsorption for Lewatit; checks for sample stability
and achievement of equilibrium; bounds for isotherm
fitting parameters; F-test procedure; fitting of the Virial
equation to CO2 isotherms; N2, Ar, and O2 desorption
isotherm data; H2O adsorption potentials; and additional
information on calculations of limiting and isosteric heats
of adsorption (PDF)
Isotherm data of Lewatit (ZIP)
Isotherm data of Purolite (ZIP)
Isotherm data in tabular format of Lewatit and Purolite
resin beads for CO2, H2O, N2, Ar and O2 (PDF)

■ AUTHOR INFORMATION
Corresponding Author
Camille Petit − Barrer Centre, Department of Chemical
Engineering, Imperial College London, London SW7 2AZ,
U.K.; orcid.org/0000-0002-3722-7984;
Email: [email protected]
Authors
May-Yin Ashlyn Low − Barrer Centre, Department of Chemical
Engineering, Imperial College London, London SW7 2AZ,
U.K.
David Danaci − Barrer Centre, Department of Chemical
Engineering, Imperial College London, London SW7 2AZ,
U.K.
Hassan Azzan − Barrer Centre, Department of Chemical
Engineering, Imperial College London, London SW7 2AZ,
U.K.; orcid.org/0000-0003-4985-419X
Figure 9. Isosteric heats of adsorption of Lewatit and Purolite for (a)
Robert T. Woodward − Institute of Materials Chemistry &
CO2, (b) N2, and (c) H2O. Research, University of Vienna, 1090 Vienna, Austria;
orcid.org/0000-0003-0834-5137
sorption isotherm model combining chemisorption and Complete contact information is available at:
physisorption terms. The model fitted the data well and could https://pubs.acs.org/10.1021/acs.jced.3c00401
3509 https://doi.org/10.1021/acs.jced.3c00401
J. Chem. Eng. Data 2023, 68, 3499−3511
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

Notes (13) Young, J.; Mcilwaine, F.; García-Díez, E.; Smit, B.; Garcia, S.; Van
The authors declare no competing financial interest. der Spek, M. Towards benchmarking and advancing solid-sorbent
The current version of the software package used for isotherm direct air capture. SSRN Electron. J. 2022, DOI: 10.2139/ssrn.4277724.
fitting and uncertainty calculation is available on the Imperial (14) Kim, S. m.; Leonard, G. Performance and Sensitivity Analysis of
Direct Air Capture (DAC) Model Using Solid Amine Sorbents for CO2
College London Github repository and can be accessed at
Capture. SSRN Electron. J. 2022, DOI: 10.2139/ssrn.4280585.
https://github.com/ImperialCollegeLondon/ (15) van Schagen, T. N.; van der Wal, P. J.; Brilman, D. W. F.
IsothermFittingTool. The version of the code used in this Development of a novel, through-flow microwave-based regenerator for
publication is referred to by the git commit ID “060ff65”. sorbent-based direct air capture. Chem. Eng. J. Adv. 2022, 9,

■ ACKNOWLEDGMENTS
M.Y.A.L. and C.P. would like to acknowledge Imperial College
No. 100187.
(16) Schellevis, H. M.; van Schagen, T. N.; Brilman, D. W. F. Process
optimization of a fixed bed reactor system for direct air capture. Int. J.
Greenh. Gas Control 2021, 110, No. 103431.
London for the President’s PhD Scholarship. D.D. and C.P.
(17) Yu, Q.; Brilman, D. W. F. Design Strategy for CO2 Adsorption
would like to acknowledge the funding provided by UK from Ambient Air Using a Supported Amine Based Sorbent in a Fixed
Research and Innovation (UKRI) under grants EP/P026214/ Bed Reactor. Energy Procedia 2017, 114, 6102−6114.
1 and EP/T033940/1. H.A. would like to acknowledge funding (18) Yu, Q.; Brilman, W. A Radial Flow Contactor for Ambient Air
through a donation to the Department of Chemical Engineering, CO2 Capture. Appl. Sci. 2020, 10 (3), 1080.
Imperial College London, by Mr. Mark Richardson. R.T.W. (19) Schellevis, M.; Jacobs, T.; Brilman, W. CO2 Capture From Air in
would like to acknowledge the funding support of the University a Radial Flow Contactor: Batch or Continuous Operation? Front. Chem.
of Vienna (Austria). All authors would like to thank Mr. James Eng. 2020, 2, No. 596555.
Rosie from Purolite Ltd., UK, for providing them with a free (20) Drechsler, C.; Agar, D. W. Simulation and optimization of a novel
sample of Purolite A110 to conduct their experiments. All moving belt adsorber concept for the direct air capture of carbon
authors would like to thank Mr. Chris Pilkington from dioxide. Comput. Chem. Eng. 2019, 126, 520−534.
(21) Shaik, A.; Benneker, A.; McCoy, S. Achieving net negative CO2
Micromeritics for providing them with the numerical isotherm
emissions with sorbent direct air capture under different energy inputs
data for the N2 and CO2 reference materials used in this work. and climate variations. SSRN Electron. J. 2022, DOI: 10.2139/
M.Y.L. would like to thank Mrs. Patricia Carry for her guidance ssrn.4285081.
and advice with the specific heat capacity measurements. (22) Sendi, M.; Bui, M.; Mac Dowell, N.; Fennell, P. Geospatial

■ REFERENCES
(1) Arias, P. A.; et al., ″Climate Change 2021: The Physical Science Basis.
analysis of regional climate impacts to accelerate cost-efficient direct air
capture deployment. One Earth 2022, 5 (10), 1153−1164.
(23) Klomp, J.; Srinivas, A.; Brilman, W. Design And Demonstration
Contribution of Working Group I to the Sixth Assessment Report of the Of A 10 kg CO2/day DAC Pilot Facility Using Sorbent Circulation.
Intergovernmental Panel on Climate Change″; IPCC, 2021. SSRN Electron. J. 2022, DOI: 10.2139/ssrn.4290510.
(2) Low, M.-Y.; Barton, L. V.; Pini, R.; Petit, C. Analytical review of the (24) Schellevis, M.; Brilman, W. Experimental Campaign with a 1 kg/
current state of knowledge of adsorption materials and processes for Day Fixed Bed Dac Unit Using Solid Amines. SSRN Electron. J. 2022,
direct air capture. Chem. Eng. Res. Des. 2023, 189, 745−767. DOI: 10.2139/ssrn.4290007.
(3) Smith, S. M. et al., ″The State of Carbon Dioxide Removal - 1st (25) Bos, M. J.; Kreuger, T.; Kersten, S. R. A.; Brilman, D. W. F. Study
ed.,″ The State of Carbon Dioxide Removal, 2023. on transport phenomena and intrinsic kinetics for CO2 adsorption in
(4) Khurana, M.; Farooq, S. Adsorbent Screening for Postcombustion solid amine sorbent. Chem. Eng. J. 2019, 377, 120374.
CO2 Capture: A Method Relating Equilibrium Isotherm Characteristics (26) Driessen, R. T.; Kersten, S. R. A.; Brilman, D. W. F. A Thiele
to an Optimum Vacuum Swing Adsorption Process Performance. Ind. Modulus Approach for Nonequilibrium Adsorption Processes and Its
Eng. Chem. Res. 2016, 55 (8), 2447−2460. Application to CO2 Capture. Ind. Eng. Chem. Res. 2020, 59 (15), 6874−
(5) Rajagopalan, A. K.; Avila, A. M.; Rajendran, A. Do adsorbent 6885.
screening metrics predict process performance? A process optimization (27) Shu, Q.; Haug, M.; Tedesco, M.; Kuntke, P.; Hamelers, H. V. M.
based study for post-combustion capture of CO2. Int. J. Greenh. Gas Direct Air Capture Using Electrochemically Regenerated Anion
Control 2016, 46, 76−85. Exchange Resins. Environ. Sci. Technol. 2022, 56 (16), 11559−11566.
(6) Farmahini, A. H.; Krishnamurthy, S.; Friedrich, D.; Brandani, S.; (28) DuPont. ″DuPontAmberLite HPR4800 OH Ion Exchange
Sarkisov, L. Performance-Based Screening of Porous Materials for Resin.″ https://www.dupont.com/content/dam/dupont/amer/us/
Carbon Capture. Chem. Rev. 2021, 121 (17), 10666−10741. en/water-solutions/public/documents/en/IER-AmberLite-
(7) Xiang, X.; et al. High Adsorption Capacity Fe@13X Zeolite for HPR4800-OH-PDS-45-D01234-en.pdf (accessed 2023).
Direct Air CO2 Capture. Ind. Eng. Chem. Res. 2023, 62 (12), 5420− (29) Elfving, J.; Bajamundi, C.; Kauppinen, J.; Sainio, T. Modelling of
5429. equilibrium working capacity of PSA, TSA and TVSA processes for
(8) Song, M.; et al. Cold-Temperature Capture of Carbon Dioxide CO2 adsorption under direct air capture conditions. J. CO2 Util. 2017,
with Water Coproduction from Air Using Commercial Zeolites. Ind. 22, 270−277.
Eng. Chem. Res. 2022, 61 (36), 13624−13634. (30) Elfving, J.; Kauppinen, J.; Jegoroff, M.; Ruuskanen, V.; Järvinen,
(9) Young, J.; García-Díez, E.; Garcia, S.; van der Spek, M. The impact L.; Sainio, T. Experimental comparison of regeneration methods for
of binary water−CO2 isotherm models on the optimal performance of CO2 concentration from air using amine-based adsorbent. Chem. Eng. J.
sorbent-based direct air capture processes. Energy Environ. Sci. 2021, 14 2021, 404, No. 126337.
(10), 5377−5394. (31) Elfving, J.; Sainio, T. Kinetic approach to modelling CO2
(10) Bos, M. J.; Pietersen, S.; Brilman, D. W. F. Production of high adsorption from humid air using amine-functionalized resin:
purity CO2 from air using solid amine sorbents. Chem. Eng. Sci.: X 2019, Equilibrium isotherms and column dynamics. Chem. Eng. Sci. 2021,
2, No. 100020. 246, No. 116885.
(11) Sanz-Perez, E. S.; Murdock, C. R.; Didas, S. A.; Jones, C. W. (32) Elfving, J.; Bajamundi, C.; Kauppinen, J. Characterization and
Direct Capture of CO2 from Ambient Air. Chem. Rev. 2016, 116 (19), Performance of Direct Air Capture Sorbent. Energy Procedia 2017, 114,
11840−11876. 6087−6101.
(12) Zerobin, F.; Pröll, T. Concentrated Carbon Dioxide (CO2) from (33) Parvazinia, M.; Garcia, S.; Maroto-Valer, M. CO2 capture by ion
Diluted Sources through Continuous Temperature Swing Adsorption exchange resins as amine functionalised adsorbents. Chem. Eng. J. 2018,
(TSA). Ind. Eng. Chem. Res. 2020, 59 (19), 9207−9214. 331, 335−342.

3510 https://doi.org/10.1021/acs.jced.3c00401
J. Chem. Eng. Data 2023, 68, 3499−3511
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

(34) Chen, H.; Dong, H.; Shi, Z.; SenGupta, A. K. ″Direct air capture (56) Serna-Guerrero, R.; Dana, E.; Sayari, A. New Insights into the
(DAC) and sequestration of CO2: Dramatic effect of coordinated Interactions of CO2 with Amine-Functionalized Silica. Ind. Eng. Chem.
Cu(II) onto a chelating weak base ion exchanger,″. Sci. Adv. 2023, 9 Res. 2008, 47 (23), 9406−9412.
(10), 1956. (57) Pai, K. N.; Baboolal, J. D.; Sharp, D. A.; Rajendran, A. Evaluation
(35) Lanxess. ″Product Information Lewatit VP OC 1065.″ https:// of diamine-appended metal-organic frameworks for post-combustion
lanxess.com/en/Products-and-Brands/Products/l/LEWATIT--VP- CO2 capture by vacuum swing adsorption. Sep. Purif. Technol. 2019,
OC-1065 (accessed 2023). 211, 540−550.
(36) Purolite. ″Purolite A110.″ https://www.purolite.com/product- (58) Hefti, M.; Joss, L.; Bjelobrk, Z.; Mazzotti, M. On the potential of
pdf/A110.pdf (accessed 2023). phase-change adsorbents for CO2 capture by temperature swing
(37) Brunauer, S.; Emmett, P. H.; Teller, E. Adsorption of Gases in adsorption. Faraday Discuss. 2016, 192, 153−179.
Multimolecular Layers. J. Am. Chem. Soc. 1938, 60 (2), 309−319. (59) Hughes, R.; et al. Isotherm, Kinetic, Process Modeling, and
(38) Thommes, M.; et al. Physisorption of gases, with special Techno-Economic Analysis of a Diamine-Appended Metal−Organic
reference to the evaluation of surface area and pore size distribution Framework for CO2 Capture Using Fixed Bed Contactors. Energy Fuels
(IUPAC Technical Report). Pure Appl. Chem. 2015, 87 (9−10), 1051− 2021, 35 (7), 6040−6055.
1069. (60) Kim, C.; Cho, H. S.; Chang, S.; Cho, S. J.; Choi, M. An
(39) Li, S.; Zhao, J.; Lu, P.; Xie, Y. Maximum packing densities of basic ethylenediamine-grafted Y zeolite: a highly regenerable carbon dioxide
3D objects. Chin. Sci. Bull. 2010, 55 (2), 114−119. adsorbent via temperature swing adsorption without urea formation.
(40) Moosavi, S. M.; et al. A data-science approach to predict the heat Energy Environ. Sci. 2016, 9 (5), 1803−1811.
capacity of nanoporous materials. Nat. Mater. 2022, 21 (12), 1419− (61) Kolle, J. M.; Fayaz, M.; Sayari, A. Understanding the Effect of
Water on CO2 Adsorption. Chem. Rev. 2021, 121 (13), 7280−7345.
1425.
(62) Shimizu, S.; Matubayasi, N. Temperature Dependence of
(41) NIST. ″Aluminium Oxide.″ https://webbook.nist.gov/cgi/
Sorption. Langmuir 2021, 37 (37), 11008−11017.
cbook.cgi?ID=C1344281&Type=JANAFS&Plot=on. accessed 2023
(63) NIST. ″Saturation Properties for Water � Temperature
(42) Hwang, J.; Azzan, H.; Pini, R.; Petit, C. H2, N2, CO2, and CH4
Increments.″ https://webbook.nist.gov/cgi/fluid.cgi?TLow=273.
Unary Adsorption Isotherm Measurements at Low and High Pressures
15&THigh=373.15&TInc=1&Digits=5&ID=C7732185&Action=
on Zeolitic Imidazolate Framework ZIF-8. J. Chem. Eng. Data 2022, 67 Load&Type=SatP&TUnit=K&PUnit=MPa&DUnit=
(7), 1674−1686. mol%2Fl&HUnit=kJ%2Fmol&WUnit=m%2Fs&VisUnit=
(43) Azzan, H.; Rajagopalan, A. K.; L’Hermitte, A.; Pini, R.; Petit, C. uPa*s&STUnit=N%2Fm&RefState=DEF (accessed 2023).
Simultaneous Estimation of Gas Adsorption Equilibria and Kinetics of
Individual Shaped Adsorbents. Chem. Mater. 2022, 34 (15), 6671−
6686.
(44) Azzan, H.. ″Adsorption Isotherm fitting tool.″ https://github.
com/ImperialCollegeLondon/IsothermFittingTool (accessed 2023).
(45) Socrates, G.; Infrared and Raman Characteristic Group
Frequencies: Tables and Charts, 3rd Edition; Wiley: Chichester. 2004.
(46) Yu, Q.; Delgado, J. P.; Veneman, R.; Brilman, D. W. Stability of a
Benzyl Amine Based CO2 Capture Adsorbent in View of Regeneration
Strategies. Ind. Eng. Chem. Res. 2017, 56 (12), 3259−3269.
(47) Rochelle, G. T. Thermal degradation of amines for CO2 capture.
Curr. Opin. Chem. Eng. 2012, 1 (2), 183−190.
(48) Gaur, U.; Wunderlich, B. Heat Capacity and Other
Thermodynamic Properties of Linear Macromolecules. V. Polystyrene.
J. Phys. Chem. Ref. Data 1982, 11 (2), 313−325.
(49) Sonnleitner, E.; Schöny, G.; Hofbauer, H. Assessment of zeolite
13X and Lewatit® VP OC 1065 for application in a continuous
temperature swing adsorption process for biogas upgrading. Biomass
Convers. Biorefin. 2018, 8 (2), 379−395.
(50) Evans, J. D.; Bon, V.; Senkovska, I.; Kaskel, S. A Universal
Standard Archive File for Adsorption Data. Langmuir 2021, 37 (14),
4222−4226.
(51) Veneman, R.; Frigka, N.; Zhao, W.; Li, Z.; Kersten, S.; Brilman,
W. Adsorption of H2O and CO2 on supported amine sorbents. Int. J.
Greenh. Gas Control 2015, 41, 268−275.
(52) Sutanto, S.; Dijkstra, J. W.; Pieterse, J. A. Z.; Boon, J.; Hauwert,
P.; Brilman, D. W. F. CO2 removal from biogas with supported amine
sorbents: First technical evaluation based on experimental data. Sep.
Purif. Technol. 2017, 184, 12−25.
(53) van Paasen, S.; et al. Development of the solid sorbent
technology for post combustion CO2 capture towards commercial
prototype. Int. J. Greenh. Gas Control 2021, 109, No. 103368.
(54) Carneiro, J. S. A.; Innocenti, G.; Moon, H. J.; Guta, Y.; Proaño,
L.; Sievers, C.; Sakwa-Novak, M. A.; Ping, E. W.; Jones, C. W.; et al.
Insights into the Oxidative Degradation Mechanism of Solid Amine
Sorbents for CO2 Capture from Air: Roles of Atmospheric Water.
Angew. Chem., Int. Ed. 2023, 62, No. e202302887, DOI: 10.1002/
anie.202302887.
(55) Serna-Guerrero, R.; Belmabkhout, Y.; Sayari, A. Modeling CO2
adsorption on amine-functionalized mesoporous silica: 1. A semi-
empirical equilibrium model. Chem. Eng. J. 2010, 161 (1−2), 173−181.

3511 https://doi.org/10.1021/acs.jced.3c00401
J. Chem. Eng. Data 2023, 68, 3499−3511