2371863

Final Technical Report / Bertole et al. / CORMETECH Inc.
/ DE-FE0032094
FINAL TECHNICAL REPORT
Bench-Scale Testing of Monolithic PPI Structured Contactors for Direct Air Capture of CO2
Reporting Period: September 15, 2021 – December 14, 2023
SUBMITTED UNDER AWARD NUMBER
DE-FE0032094
SUBMITTED BY
CORMETECH, Inc. (UEI = U2U7UKL8DJQ8)
11707 Steele Creek Road, Charlotte, NC 28273
PRINCIPAL INVESTIGATOR
Dr. Christopher J. Bertole
(919) 620-3524, [email protected]
TEAM MEMBERS
Georgia Tech Research Center
Dr. Christopher W. Jones, Dr. Ryan Lively, Dr. Athony Vallace
Global Thermostat Operations, LLC
Dr. Miles Sakwa-Novak, Dr. Stephanie Didas, Dr. Eric Ping
SUBMITTED TO
U.S. Department of Energy
National Energy Technology Laboratory
Page 1 of 61
Final Technical Report / Bertole et al. / CORMETECH Inc. / DE-FE0032094
CONTENTS
ACKNOWLEDGMENT ............................................................................................................................ 3
DISCLAIMER............................................................................................................................................. 4
EXECUTIVE SUMMARY ........................................................................................................................ 5
INTRODUCTION....................................................................................................................................... 7
SINGLE BRICK SORPTION TESTER (SBST) OVERVIEW AND VALIDATION ....................... 13
LINEAR-POLY(PROPYLENIMINE) SYNTHESIS AND CHARACTERIZATION....................... 15
SBST TESTING OF PPI STANDARD MONOLITH ........................................................................... 19
WICKE KALLENBACH (WK) DIFFUSION CELL TESTING ......................................................... 21
ADSORPTION ISOTHERM TESTING ................................................................................................ 33
SINGLE CHANNEL MONOLITH CONTACTOR MODEL.............................................................. 41
APPROACH FOR PPI MONOLITH OPTIMIZATION ..................................................................... 44
CORE TESTING OF OPTIMIZED PPI MONOLITH PROTOTYPES ............................................ 48
PPI OXIDATIVE STABILITY TESTING............................................................................................. 51
SBST TESTING OF OPTIMIZED PPI MONOLITH .......................................................................... 53
SINGLE CHANNEL MONOLITH CONTACTOR MODEL UPDATE ............................................ 56
TECHNO-ECONOMIC ANALYSIS (TEA) .......................................................................................... 58
TECHNOLOGY MATURATION PLAN .............................................................................................. 60
REFERENCES .......................................................................................................................................... 61
Page 2 of 61
ACKNOWLEDGMENT
This material is based upon work supported by the Department of Energy under Award Number(s) DE-
FE0032094.
Page 3 of 61
DISCLAIMER
This report was prepared as an account of work sponsored by an agency of the United States Government.
Neither the United States Government nor any agency thereof, nor any of their employees, makes any
warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness,
or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would
not infringe privately owned rights. Reference herein to any specific commercial product, process, or
service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its
endorsement, recommendation, or favoring by the United States Government or any agency thereof. The
views and opinions of authors expressed herein do not necessarily state or reflect those of the United States
Government or any agency thereof.
Page 4 of 61
EXECUTIVE SUMMARY
The overall project objective was to develop, optimize and bench-scale test the integrated embodiment
of a leading direct air capture (DAC) sorbent composition, linear-poly(propylenimine) (l-PPI), in a
structured material system, specifically a monolithic contactor for achieving low pressure drop, to justify
its further scale-up in a subsequent program. The project team was led by CORMETECH, a leading
monolithic gas/solid contactor manufacturer for environmental applications, and included Global
Thermostat, a well-known DAC start-up, and Georgia Tech, a prominent US academic institution in carbon
capture and DAC. The monolithic contactor for l-PPI in this project employed an advanced, cost-effective
fabrication technique from CORMETECH, namely a porous monolith substrate directly impregnated with
the amine, an approach contrasting to the traditional wash-coating methods used for activating monoliths.
The l-PPI monolithic contractor is targeted for use in DAC systems such as the Global Thermostat DAC
process, which involves the cyclic operation of ambient air flow over a monolithic amine contactor followed
by steam-mediated thermal desorption and CO2 collection, maximizing volumetric productivity while
reducing the auxiliary power required to capture CO2 from air. The Global Thermostat DAC process
currently utilizes poly(ethyleneimine) (PEI) materials as baseline sorbents, which are subject to oxidative
degradation and slow capacity fade at the elevated temperature required for efficient CO2 removal. Georgia
Tech, with Global Thermostat, had discovered in lab-scale efforts that l-PPI had superior resistance to
oxidative degradation compared to PEI, with similar CO2 adsorption capacity. While the novel l-PPI
sorbent offers significant advantages over PEI for DAC, including potential design simplification and
increased process efficiency, it is not commercially available, and prior to this project, had not yet been
evaluated on the bench-scale in a structured contactor.
Our main contributions from this bench-scale testing project were as follows:
1. Designed, built, and commissioned a single brick sorption tester (SBST) for measuring the adsorption
and steam desorption performance of a full-size monolithic contactor.
2. Synthesized over 1 kilogram of linear-PPI for the bench-scale evaluation.
3. Measured the effective diffusivity of CO2, under dry and humid conditions, through linear-PPI loaded
model and monolithic substrates.
4. Measured the adsorption isotherms of CO2, under dry and humid conditions, on linear-PPI loaded
model and monolithic substrates.
5. Measured the oxidative degradation kinetics of supported l-PPI, under both accelerated high
temperature conditions as well as more representative low temperature conditions, and incorporated it
into the modeling efforts, to understand the potential for reduced capacity fade.
Page 5 of 61
6. Developed and validated a COMSOL FEM model of the monolithic adsorber to simulate the adsorption
and steam-mediated desorption of CO2, to optimize the monolith substrate properties for l-PPI
(including impact of cell density, wall thickness, and meso- and macro-porosity of the porous material).
7. Extruded monolith substrates with modified textural properties (pore volume, and meso and macro
porosity) and cell density, for prototyping and bench-scale testing validations.
8. Completed the bench scale test of an optimized l-PPI monolith contactor, achieving over 2x the
adsorption capacity of the standard l-PPI monolith contactor, and exceeding the project’s success
criteria metrics.
9. Developed an overall process model for the novel PPI/monolithic contactor and applied it for the TEA
and LCA work.
Key lessons learned from this project included:
1. The structure of the l-PPI solid domains in the porous solid monolith structure impacts performance.
Humidity either promotes or inhibits CO2 adsorption, depending on the l-PPI loading and the pore size
distribution. Understanding this behavior was key to optimizing the monolith contactor for l-PPI and
maximizing performance.
2. The monolith synthesis process depends on the properties of the amine. PEI is a hydrophilic, liquid
amine, which lends itself well to an impregnation process for monolith synthesis. However, l-PPI is a
hydrophobic, solid amine (melting point ~ 110oC). The use of l-PPI instead of PEI was not a simple
substitution: the impregnation protocol for l-PPI required the use of non-aqueous solvents. This aspect
is an important consideration for scale-up and should be noted in work progressing from powders to
monolith cores to full-size monoliths.
In summary, it was determined that monolithic l-PPI is a promising Next Gen adsorber material for DAC,
a conclusion validated through fundamental measurements, modeling efforts, and prototyping and testing.
Page 6 of 61
INTRODUCTION
Direct air capture (DAC) is a carbon removal process that extracts CO2 directly from the atmosphere at
any location, converting a feedstock nominally containing 420ppm CO2 into a highly concentrated CO2
stream (>99%) that can be permanently sequestered in deep geological formations, or alternatively, used as
a commodity chemical in various applications (such as beverage carbonation and chemicals synthesis).
One example of a DAC process is the Global Thermostat (GT) DAC Process, shown in Figure 1. The
GT DAC process is a cyclic, batch process with two primary functional steps: (1) Adsorption, where CO2
is selectively bound to the contactor from the air, and (2) Desorption, where the adsorbed CO2 is then
unbound from the contactor using a combination of vacuum and direct steam injection (TVSA) and then
collected. The process leverages moving bed technology to transport panels of honeycomb monolith
contactors around stations of a carousel-based DAC module.
Figure 1. Schematic diagram of the Global Thermostat DAC process.
The monolithic, low pressure drop contactors are loaded with organic amine molecules or
macromolecules, such as PEI or l-PPI, to allow for high throughput gas/solid contacting with high
volumetric CO2 sorption productivity. Figure 2A illustrates that the amine sorbent is placed within the
meso-pore volume of the monolith contactor, leaving the macro-porosity largely open for efficient CO2
diffusive transport. Figure 2B shows an example of a 150mm x 150mm x 150mm porous monolith
contactor manufactured by CORMETECH.
Page 7 of 61
2A 2B
Figure 2. 2A: monolith contactor system used in Global Thermostat DAC process. 2B: 170cpsi porous
monolith contactor (dimensions=150mmx150mmx150mm) designed and manufactured by CORMETECH.
The baseline amine polymer, PEI [generally: – (NH – CH2 – CH2) n>1 – NH2], is typically deployed as
a branched oligomer of relatively low molecular weight (~1000 g/mol)1. In contrast, the selected next
generation polymer for this project, l-PPI [generally: – (NH – CH2 – CH2 – CH2) n>1 – NH2], is linear in
structure, with a comparable molecular weight. Given the different structure of the macromolecule (PEI:
branched with ethyl linkers, vs. l-PPI: linear with propyl linkers), the new polymer was expected to interface
with the porous monolith support differently from the baseline PEI case2. Powder tests of supported l-PPI
(data are shown in Figures 3A and 3B) highlight the potential of the l-PPI sorbent: it has similar CO2
sorption properties as PEI, yet superior oxidative stability and water sorption properties (more
hydrophobic), making it a promising next generation amine polymer for DAC3.
3B
3A
Figures 3A-B. 3A: Isotherms at varied temperatures for l-PPI / SBA-15 sorbents. 2B: Amine efficiency
vs. amine loading for l-PPI vs PEI (closed shapes), 400ppm CO2 at 35oC, showing loss after treatment in
oxidative environment (open shapes), 21%O2/N2, 110oC for 24h. Adapted from Reference 3.
Page 8 of 61
The Project Team was comprised of many members who contributed to the overall success of the
project. Team members from CORMETECH included: Christopher Bertole (project PI; VP, Product
and Applications Development), Gavin MacInnes (Lead Catalyst Development Engineer), Travis Jones
(Manager, Product and Applications Development Laboratory), Colby Burtt (Lead Process Development
Engineer), James Altizer (VP, Product and Manufacturing Development, Amanda Currie (Sr. R&D
Technician), Dilair Khara (Sr. R&D Technician), Joe Kliwinski (Sr. R&D Technician), Casey Huten
(P&AD Laboratory Engineer), Chris Lawson (Supervisor, P&AD Flow Laboratory), Emma Lavender
(Supervisor, P&AD Analytical and Synthesis Laboratory), John Powe (Development Technician), Chacko
John (Development Technician), and Kyle Kearney (Development Technician). Team members from
Georgia Tech Research Center included: Chris Jones (John F. Brock III School Chair & Professor), Ryan
Lively (Professor), and Anthony Vallace (Postdoctoral Research Fellow). Team members from Global
Thermostat Operations included: Miles Sakwa-Novak (VP, Materials), Eric Ping (VP, Process &
Operations), Stephanie Didas (Director, Special Projects), Cassandra Hertz (Formulations Development
Manager), and Abby Clabaugh (Development Engineer). Roles and responsibilities of the project team
organization members were as follows. CORMETECH: bench-scale test unit (SBST) design, build, and
commissioning; full-size l-PPI monolith impregnation and CO2 capacity testing, single channel monolith
modeling for optimization (COMSOL), and monolith porous substrate optimization and prototype
extrusion. Georgia Tech Research Center: l-PPI validation testing, CO2 diffusion rate measurements and
transport modeling. Global Thermostat Operations: l-PPI procurement and validation testing, CO2
adsorption/desorption measurements and adsorption modeling, l-PPI oxidative stability testing, and the
TEA and LCA.
The overall objective of this project was to develop and bench-test a novel CO2 sorbent-gas contactor
composition with low pressure drop optimized for CO2 removal from ambient air, to advance this direct air
capture (DAC) system to a sufficient maturity level that can justify its scale-up in a subsequent program.
The specific project objectives were to: (1) Develop a next generation oxide monolith + amine structured
contactor for DAC, based on linear-poly(propyleneimine) (l-PPI), (2) Refine, with inputs from experimental
kinetic, thermodynamic and transport measurements, the single channel monolith model and the DAC
process model, in order to simulate the impact of substrate textural properties (meso and macro porosity)
on DAC performance and enable contactor optimization prior to prototype synthesis & bench-scale testing,
and (3) Refine the DAC process techno-economic analysis (TEA) and life cycle analysis (LCA).
The project performance period consisted of two, one-year budget periods (BP1 = 09/15/2021–
09/14/2022, and BP2 = 09/15/2022–09/14/2023). The project work included experimental and simulation
studies of the novel CO2 sorbent/gas contactor system based on l-PPI sorbents for DAC to optimize its
Page 9 of 61
performance at environmentally relevant conditions. Budget Period 1 focused on: (1) Designing, building,
and commissioning of the bench-scale single brick sorption tester (SBST), (2) Production, validation and
bench-scale testing of l-PPI (at kg-scale) using a standard monolithic substrate and the SBST, (3)
Development of a single channel model to optimize the geometry and the meso/macro-porosity of
monoliths specifically tuned for l-PPI, for validation in year two, (4) Measurement of the isotherms, kinetic,
and transport behaviors of l-PPI supported on powder and slab materials to develop an adsorption model
and a transport model for inclusion in the single channel monolith model, (5) Updating of the process model
to incorporate the novel sorbent/contactor combination, and (6) Starting work on the TEA. Budget Period
2 focused on: (1) Continued refinement, as new data became available, of the single channel monolith
model, adsorption model, and transport model, (2) Measurement of the resistance of the sorbent/ contactor
composition to oxidation under relevant dry and humid conditions, and incorporation of the oxidation
kinetics into the single channel model, (3) Modeling simulations to generate the optimal contactor
formulation, followed by its synthesis and bench-scale validation testing on the SBST, and (4) Completion
of the TEA and the LCA. To accomplish these objectives, the project was organized into nine main Tasks,
outlined briefly below, with the project success criteria summarized in Tables 1 and 2.
Task 1 – Project Management and Planning
Subtask 1.1 – Project Management Plan; Subtask 1.2 – Technology Maturation Plan
Budget Period 1 Tasks
Task 2 – Bench-Scale SBST
Subtask 2.1 – Design and Build; Subtask 2.2 – Commissioning
Task 3 – Experimental Testing and Models Development
Subtask 3.1 – Single Channel Monolith Model Development
Subtask 3.2 – Adsorption / Desorption Testing and Model Development
Subtask 3.3 – Diffusion and Transport Testing and Model Development
Task 4 – l-PPI Production & Validation
Subtask 4.1 – Produce l-PPI at kg scale
Subtask 4.2 – Validate l-PPI Performance
Subtask 4.3 – Assessment of l-PPI Scalability to Commercial Scale
Page 10 of 61
Subtask 4.4 – Impregnate Standard Monolith with PPI for SBST
Subtask 4.5 – Testing of PPI Loaded Monolith on SBST
Task 5 – Techno-Economic Analysis (TEA) and Process Modeling
Subtask 5.1 – Process Modeling; Subtask 5.2 – Initial TEA and LCA
Budget Period 2 Tasks
Task 6 – Experimental Testing and Models Refinement
Subtask 6.1 – Single Channel Monolith Model Refinement
Subtask 6.2 – Adsorption / Desorption Testing and Model Refinement
Subtask 6.3 – Diffusion and Transport Testing and Model Refinement
Task 7 – Sorbent Oxidation Experiments to Validate l-PPI Stability
Subtask 7.1 – Accelerated Oxidation Testing under Dry & Humid Conditions
Subtask 7.2 – Incorporate Oxidation Kinetics into Single Channel Monolith Model
Subtask 7.3 – Quasi-Accelerated Humid Oxidation Testing
Task 8 – Development, Synthesis and Testing of Optimal Monolith Substrate Formulation for l-PPI
Subtask 8.1 – Single Channel Monolith Model Simulation Work to Develop Optimal Substrate for l-
PPI in DAC
Subtask 8.2 – Batching, Extrusion, and Impregnation of Prototype Substrates for Optimal l-PPI DAC
Performance
Subtask 8.3 – Testing of Optimized Monolith/l-PPI Compositions in SBST
Task 9 – TEA and Process Modeling
Subtask 9.1 – Process Cycle Opportunity Analysis
Subtask 9.2 – Process Model Updates
Subtask 9.3 – Final TEA and LCA
Page 11 of 61
Table 1. List of project success criteria for budget period 1.
Table 2. List of project success criteria for budget period 2.
Page 12 of 61
SINGLE BRICK SORPTION TESTER (SBST) OVERVIEW AND VALIDATION
The SBST, shown in Figure 4, is a fully automated DAC test unit designed, built, and commissioned
during this project. Located at CORMETECH in Durham NC, it can test a full-size monolith adsorber
contactor element (150mm wide x 150mm high x variable length) under a representative DAC test cycle of
air adsorption  vacuum purge  steam injection  vacuum purge / cooling. On the air flow stream side,
the unit has two low-range CO2 analyzers for ppm analysis, one each placed at the inlet and outlet of the
adsorber chamber, enabling the quantification of the CO2 adsorbed onto the part during the adsorption step.
At the start of the desorption step, the adsorber chamber’s inlet and outlet valves are closed; the air stream
continues to flow uninterrupted around the adsorber chamber through by-pass duct. Vacuum is pulled on
the adsorber chamber to purge it; subsequent direct steam injection provides the energy required to desorb
the CO2 from the adsorber. A high-range CO2 analyzer is located after the vacuum pump for product stream
analysis, which enables checking of the CO2 mass balance (desorbed product amount vs. adsorbed air feed
amount). A mass spectrometer (MS), specifically obtained for this project, was installed on the SBST, with
the sampling sniffer placed inside the adsorber chamber at the steam-outlet face of the adsorber part,
enabling fast transient analysis during the steaming / desorption step of the DAC test cycle.
Figure 4. Single Brick Sorption Tester (SBST) located at CORMETECH in Durham, NC.
Page 13 of 61
The SBST was commissioned using a standard PEI 170cpsi MO1 monolith contactor produced by
CORMETECH (MO = metal oxide, which is the porous material constituting the monolith’s structure).
The SBST data (Table 3) have excellent repeatability (RSD = 3%) and mass balance closure (~101%), as
well as excellent agreement with data for a core sample (of the same monolithic adsorber) tested on
CORMETECH’s core test unit. The SBST unit was ready for testing the l-PPI loaded monolith!
Table 3. SBST commissioning data for the standard PEI 170cpsi MO1 monolith contactor. Notes1,2: the
reported normalized adsorption and desorption CO2 capacity values have both been normalized using the
10-cycle average of the measured SBST adsorption capacity values; the normalized capacity value on this
same basis for the core sample tested on the core test unit was 1.01.
Adsorption Desorption
Cycle # 1
Capacity Capacity 2
1 1.06 1.00
2 0.99 1.01
3 0.98 1.04
4 0.98 0.97
5 0.99 1.02
6 0.97 0.97
7 1.04 1.08
8 0.98 0.98
9 0.99 1.00
10 1.01 1.00
avg 1.00 1.01
stdev 0.03 0.03
rsd 2.9% 3.2%
Page 14 of 61
LINEAR-POLY(PROPYLENIMINE) SYNTHESIS AND CHARACTERIZATION
Georgia Tech synthesized lab-scale quantities of linear-PPI to serve as reference for the project. The
synthesis of l-PPI (Figure 5) was performed through cationic ring-opening polymerization of 2-methyl-5,6-
dihydro-4H-1,3-oxazine (MeOZI)
with methyl p-toluenesulfonate
(MeOTs) as the initiator. The
resulting poly(MeOZI) was
hydrolyzed under acidic conditions
to remove the protecting groups, and
then carefully neutralized by the
addition of NaOH and copiously
washed with aqueous NH4OH to
remove any residual inorganic salt,
generating linear PPI. Synthesis of
all intermediates are discussed below
(Steps A, B, C, and D).
Figure 5. Synthesis of linear-PPI.
Synthesis of 2-methyl-5,6-dihydro-4H-1,3-oxazine (MeOZI) (A): Anhydrous ZnCl2 (5.5 g) was

dispersed in anhydrous and distilled MeCN (115 mL). To this solution, 3-aminopropanol (150.2 g) was
added dropwise at room temperature while stirring. The mixture was then purged with dry argon and heated
to 95 °C with stirring for 2 days. During the reaction NH3 gas evolves and vents out from the system. The
reaction mixture was cooled to room temperature and excess MeCN and NH3 were removed by rotavap at
30 °C. Finally, MeOZI was isolated by distillation under vacuum (≈400 mTorr, 65 °C) and after cooling the
distillate with a dry ice/isopropyl alcohol mixture a clear, colorless liquid was obtained. The product was
stored in ambient lab conditions. 1H NMR (400 MHz, CDCl3): δ=4.144 (t, J=5.4 Hz, 2 H), 3.338 (t, J=5.8
Hz, 2 H), 1.879+1.851 ppm (s+q, J=5.7 Hz, 5 H); 13C NMR (400 MHz, CDCl3): δ=157.85, 65.00, 42.41,
22.06, 21.78 ppm. Synthesis of Linear Poly (MeOZI) (C): Before the polymerization, MeOZI and MeCN
were dried over CaH2 by stirring at 65 °C. MeOZI and MeCN were stored under dry argon and transferred
into a Straus flask and under vacuumed Schlenk line. 10 g of dried MeOZI in the flask was transferred into
a round-bottom flask, to which 50 mL dried MeCN was added via canula. An appropriate amount of
methyl p-toluenesulfonate (MeOTs) was added, based on the desired final molecular weight of the polymer.
The round-bottom flask was sealed with a septum and heated to 90 °C with stirring to initiate
polymerization. The polymerization was allowed to proceed for 48 h, after which 2.0 equivalents of 1,3-
Page 15 of 61
diaminopropane (based on MeOTs) was added to terminate the reaction. MeCN and excess 1,3-
diaminopropane were evaporated by rotary evaporation at 50 °C to afford a viscous, yellow liquid as crude
poly(MeOZI). Synthesis of Linear Poly(Propylenimine) (D): Linear poly(propylenimine) (PPI) was
synthesized by acidic hydrolysis of poly(MeOZI), 200 mL of 5 M aqueous HCl was added to the crude
poly(MeOZI) residue and stirred at 100 °C for 48 h. After cooling to room temperature, the residual polymer
was concentrated by rotavap at 60 °C, which caused the polymer to precipitate off. The concentrated
polymer solution was cooled to 0 °C in an ice bath and carefully brought to pH 13 using concentrated
aqueous NaOH, which caused to polymer to first become fully dissolved, and then precipitate as an off-
white powder. Linear PPI was recovered by centrifugation (14 000 rpm, 10 min) and copiously washed
with 10 % NH4OH to remove any residual inorganic salt. The final polymer was dissolved in MeOH and
dried under high vacuum to afford an off-white solid (≈75 % yield based on 10 g monomer). The absence
of residual inorganic salt was verified by TGA. The number-averaged molecular weight was estimated
by 1H NMR by end group analysis and compared with Celares sample as well.1H NMR (400 MHz,
[D4]MeOH): δ=3.625 (t, −CH2CH2CH2NH2, 2 H), 2.612 (t, -CH2CH2CH2NH−, 4nH), 2.361 (s, CH3−, 3 H),
1.709 ppm (q, −CH2CH2CH2NH−, 2nH).
Global Thermostat performed preliminary experiments to develop an in-situ DSC oxidation method for
the degradation of l-PPI using previously synthesized material. Data for l-PPI and a reference PEI are
shown in Figure 6; these data confirmed the significant oxidative stability benefit of l-PPI over PEI.
Figure 6. Isothermal DSC oxidative stability test for l-PPI and PEI.
Page 16 of 61
In this project, Global Thermostat contracted Celares, an organization specialized in the development
and manufacturing of polymer-based pharmaceuticals, to produce 1.5-2 kg of l-PPI in a large batch run
with a mean MW of 1000 g/mol. Figure 7 shows the synthesis scheme. Global Thermostat outlined a series
of analytical tests to be performed by Celares for quality assurance upon completion of the polymer
synthesis to ensure the final product was of high quality and had the expected chemical and physical
properties of l-PPI; the tests are outlined in the COA (Figure 8) provided by Celares upon delivery.
Figure 7. Synthesis scheme for l-PPI. Figure 8. Celares COA for l-PPI produced.
Celares started their work with small scale-experiments to evaluate process scalability and to
improve/simplify the purification process. In this preliminary work, they improved a washing step, verified
through a qualitative chloride test which confirmed negligible amounts of residual chloride ions, allowing
them to reduce a step in the purification sequence. For the large batch run, Celares completed the monomer
synthesis, polymerization, and cleavage of the acetyl groups, and then worked on the purification of the
polymer, which is the most difficult step at larger scale. In summary, Celares delivered over 1kg of l-PPI
for the project, meeting all required specifications, as shown in the provided COA (Figure 8). Celares
Page 17 of 61
developed a new, scalable process for the manufacturing of l-PPI; further goals are to reduce the amount of
aqueous HCl needed to hydrolyze the amides and to further simplify the purification process ultimately
resulting in a more cost-effective and efficient synthesis.
After the l-PPI was delivered to the project team, the polymer (labeled as “large batch 3” l-PPI) was
validated at both Global Thermostat and GTRC via a variety of analytical tools, to assess performance and
compare it to the previously synthesized l-PPI. These tests provided further assurance of the high quality
of the large batch 3 l-PPI produced by Celares. Global Thermostat ran ATR, CO2 adsorption capacity, and
DSC oxidative stability tests on the l-PPI. The measured ATR spectrum and CO2 adsorption capacities
were consistent with expectation for l-PPI, and the DSC data, presented in Figure 9, show comparable
stability to the previous l-PPI material. GTRC performed the following validation tests: NMR, melting
point, GPC/MALDI, and CO2 adsorption in MO1 standards.
Figure 9. Left: picture of an aliquot of the large batch 3 l-PPI produced by Celares. Right: Isothermal DSC
oxidative stability test for l-PPI and PEI.
Page 18 of 61
SBST TESTING OF PPI STANDARD MONOLITH
Before activation commenced to make the full-size monolith for SBST testing, a core sample was
prepared with the standard 170cpsi-MO1 monolith substrate to confirm performance of the large batch 3 l-
PPI material and provide a comparison baseline for the subsequent SBST testing. The CO2 capacity
measurements were done on CORMETECH’s core test unit. The data are presented in Figure 10 as a
relative measure to the BP1 target. Contrary to the known promotive impact of humidity on CO2 adsorption
capacity for PEI, the data indicate a strong inhibtive of air feed humidity on the CO2 uptake.
Figure 10. CO2 adsorption capacity data for the l-PPI core sample, tested on the CORMETECH core tester.
CORMETECH synthesized the full-size l-PPI monolith in quarter sections, to conserve usage of the
batch 3 l-PPI, with the 170cpsi-MO1 substrate and the same PPI loading as the core sample. The synthesis
of the optimized monolith was completed in July 2022. Figure 11 illustrates the four quarters of the l-PPI
monolith after impregnation as a composite contactor before loading into the SBST; the overall dimensions
of the final assembled monolith face was ~ 150mm x 150mm square.
Page 19 of 61
Figure 11. l-PPI stamdard MO1 monolith contactor prior to loading into the SBST.
The SBST test campaign for the l-PPI standard monolith was conducted from 7/27/2022 to 8/5/2023.
The data are presented in Figure 12 and show attainment of the BP1 performance target goal. The data also
indicated a clear opportunity to optimize the monolith for l-PPI in BP2, by adjusting porosity and amine
loading, to increase the transport rate and the overall equilibrium capacity, respectively. The diffusion
transport testing, adsorption isotherm testing, and COMSOL modeling efforts reviewed in the next sections
helped guide this optimization, leading to validation through prototyping and final run SBST testing.
Figure 12. SBST data for the standard l-PPI 170cpsi MO1 monolith contactor; H2O content of the air feed
stream ranged 1.0-1.9v%. Note that the core sample test (purple triangle) is consistent with the SBST data.
Page 20 of 61
WICKE KALLENBACH (WK) DIFFUSION CELL TESTING
A Wicke-Kallenbach membrane diffusion cell was built in the ChBE machine shop at Georgia Tech.
The cell consists of two stainless steel disks, where the feed side has a 45mm slot to house the membrane
sample. Each side of the cell has two gas lines for the feed and sweep inlets and outlets, respectively. Initial
testing used Sterlitech Anodisk® Al2O3 ceramic membranes with 20nm, 100nm, and 200nm pore sizes.
Fabricating a membrane module is as follows: a 16mm hole is punched in the center of a 47mm piece of
aluminum tape. The membrane is placed between the aluminum tape and a piece of filter paper, then the
edges of the exposed membrane are sealed with J-B Weld 8272 MarineWeld Marine Epoxy, which is
allowed to set for 3 days. The surface area of the sample is determined using ImageJ software.
A B C
D E
Figure 13. Photographs of WK cell: A: Feed side, B: Sweep side, C: Top-down view (insulation removed),
D: Sample module, E: Overall apparatus.
Page 21 of 61
Loading a sample consists of five parts: the first O-ring, followed by a metal screen, a second O-ring,
the membrane sample, and a third O-ring, placed on the back of the sample. The cell is bolted down,
wrapped in heat tape, aluminum foil, and fiberglass insulation, and placed into an insulated stainless-steel
box. Running an experiment consists of degassing the sample at 80°C for 1 h under helium flow (50sccm
on both feed and sweep ends). The degas time and temperature were kept to 1 h at 80°C because higher
temperatures and longer degas times will compromise the epoxy seal in the membrane module. The sample
cell is then cooled to 30°C and the feed stream is sent to vent, using a 3-way valve. The desired CO2
concentration is introduced to the feed stream, with a total flowrate of 50sccm, the detector begins recording
and the 3-way valve is used to send the feed stream back to the sample. During the experiment, the flowrate
exiting the detector is recorded, using a bubble meter. The experiment is concluded once a constant CO2
concentration is measured for sufficient time. Photos of the Wicke-Kallenbach cell, a sample module and
apparatus are shown in Figure 13. Feed humidification was added later on in the project to enable transport
testing under various RH conditions, which was important due to the noted impact humidity on l-PPI.
For the transport modeling, a shrinking core model was written, which allows for decay of the diffusion
coefficient as a function of the CO2 loading in the amine polymer (PEI or PPI). From this model, three
parameters are extracted; the kinetic rate constant, the initial diffusivity, and the “hindering coefficient”
which describes the rate at which diffusion decreases. This model describes CO2 sorption into a single
particle and is used to
describe TGA experiments
on powdered samples. This
model will serve as a
comparison for the
parameters extracted from
membrane experiments.
Modeling results are shown
in Figure 14 for different
combinations of the three
parameters discussed above.
A similar model is being
developed that describes the
same phenomena as the
shrinking core model,
Figure 14. Conversion vs. time (left) and diffusivity vs. conversion (right)
applied to a membrane. This
plots for shrinking core model, showing the effect of reaction rate (kr).
Page 22 of 61
model is used to extract the same three parameters discussed above but is applied to the data set from the
WK experiments.
Figure 15. Conversion vs. time (left) and diffusivity vs. conversion (right) plots for shrinking
core model, showing the effect of hindering coefficient (K).
Figure 16. Conversion vs. time (left) and diffusivity vs. conversion (right) plots for shrinking
core model, showing the effect of initial diffusivity (D0).
Page 23 of 61
To measure the diffusive transport rates of CO2 through the amine polymer, two types of support
materials were considered: (1) model substrates with well-defined pore structure, and (2) porous substrates
with much more tortuous pore structures. The selected model substrates, commercially known as
Anodiscs®, are Al2O3 membrane filters, had three different reported pore sizes, 20nm, 100nm, and 200nm,
with a thickness of 60 μm. Hg porosimetry was used to characterize the pore size distribution and total
pore volume of the Anodiscs® and the results are shown in Figures 17 and 18. The pore sizes ranged from
170 to 310 nm for all three Anodiscs®, with the total pore volume increasing with increasing pore size.
SEM images of the three Anodiscs® shown below in Figure 19, confirm that the reported pore size applies,
at least, to the size of the pore mouth opening.
Figure 17. Incremental Hg porosimetry data for 20, 100, and 200nm Anodiscs®.
Figure 18. Cumulative Hg porosimetry results for 20, 100, and 200nm Anodiscs®.
Page 24 of 61
200n 100n
20n
Figure 19. SEM images of 20nm, 100nm, and 200nm Anodiscs®.
Figure 20. Side-on SEM images on “20nm” Anodisc® at a) low magnification, b) medium magnification.
Based on the pore size distribution results, from Hg porosimetry, the reported pore size is a very small
fraction of the average pore size, along the thickness of the membrane (the z-axis) and perhaps could even
only apply to the size of the pore mouth opening. Another note is that the reported pore sizes (as shown
Page 25 of 61
above in Figure 19), only applies to one side of the Anodisc®. Pore diameter estimations from side-on SEM
images shown Figure 20, of the “20nm” Anodisc®, are consistent with the Hg porosimetry data, of pore
sizes in the ~200nm range. More specifically, pore sizes were estimated using Image-J software on Figure
20b and showed pore sizes of ~200nm.
N2 physisorption was also performed on the three Anodiscs® and showed a BET surface area of 10-11
m2/g for all samples. The BET plot for the three Anodiscs® is shown below in Figure 3.3.1.5. Because of
the macroporous (pore size > 5nm) nature of the samples, reliable pore volume characterization, with gas
adsorption, was not reliable.
Figure 21. BET plot of “20nm”, “100nm”, and “200nm” Anodiscs®.
Post-characterization, all three bare Anodisc® samples were tested in the Wicke-Kallenbach cell at
30°C with 400ppm and 4% CO2. The baseline CO2 diffusivities without any amine polymer loaded were
determined using Fick’s first law and are shown below in Figure 22. The CO2 diffusivity remained
relatively constant with reported pore size, consistent with the reported pore size being a very small fraction
of an entire pore.
Page 26 of 61
Figure 22. CO2 diffusivities at 30°C for bare Anodiscs® under 400 ppm and 4% CO2.
After the blank measurements were completed, several synthesis trials were attempted to load PEI into
the Anodiscs® material for WK cell measurements, to validate the method and enable a comparison
between PEI and l-PPI. However, given the straight pore structures of the Anodiscs® and the fact that the
PEI sorbent is a liquid, it was very difficult to retain the PEI within the pore structure during testing, and
the testing proved unsuccessful. In contrast, l-PPI, which is a solid at DAC adsorption conditions, was
much more amenable to loading into the Anodiscs® and being retained during testing. Synthesis and WK
cell testing of l-PPI Anodiscs® samples was successful.
The CO2 transport parameters obtained from the WK cell testing for l-PPI/200nm Anodiscs® sample
at ~100% pore fill (20wt%) are shown below in Table 4. CO2 equilibrium adsorption capacities were also
obtained for this material at both 400ppm and 4% CO2 using the TGA; these data were necessary for the
shrinking core model calculations to resolve the parameters summarized in Table 4.
Table 4. WK results for l-PPI/ 200nm Anodiscs®.

Dpolymer (m2/s) Dgas (m2/s) Permeability Flux (mol/m2 s) Sorption
(mol/m s Pa) Coefficient
(mol/m3 Pa)
400 ppm CO2 1.1E-12 9.0E-9 3.6E-12 2.2E-6 3.3E0
4% CO2 9.2E-11 8.8E-9 3.5E-12 2.2E-4 3.8E-2
Page 27 of 61
The CO2 transport parameters obtained from the WK cell testing at 400ppm and 4% CO2 for samples
of l-PPI and PEI supported on MO1 monolith substrates at various pore fill levels are shown in Table 5.
The diffusivity values in the polymer domain are also plotted in Figure 23. Clearly an unexpected trend is
observed: the diffusivity increases as a function of PPI loading in the MO1 monolith; also, the diffusivity
values are higher than what was measured for the l-PPI loaded in the Anodiscs® material (Table 4). We
ascribe this trend to two possibilities: (1) the applicability of the solution diffusion model type model to this
porous system, or (2) the irregularity of the sample morphology, leading to voids in the PPI loaded monolith.
We hypothesize that because l-PPI is a solid at room temperature, the macroporous morphology could be
comprised of PPI particles coating the MO1 substrate particles, whereas the liquid PEI is a continuous
phase, in the MO1 substrate macropores, leading to the unexpected trend in transport behavior with
increases PPI loading. A reduction in the sorption coefficient with increased PPI loading with similar CO2
fluxes leads to significantly higher diffusivities.
Table 5. WK results for l-PPI and PEI supported on MO1 monolith substrates.
400 ppm CO2/He 4% CO2/He

30°C 30°C
Sample Pore Fill Dgas Dpolymer Dgas Dpolymer

(Meso, Macro, total) (vol%) (m2/s) (m2/s) (m2/s) (m2/s)
PEI-MO1-Lo MP 92, 0, 58 5.2E-07 6.0E-12 8.1E-07 7.4E-10
PEI-MO1-Hi MP 100, 4, 58 6.2E-07 8.1E-12 6.3E-07 6.8E-10
PEI-MO1-Lo MP 100, 58, 95 3.3E-08 4.0E-12 9.7E-08 9.9E-11
PEI-MO1-Hi MP 100, 40, 83 3.2E-07 3.9E-13 3.9E-07 4.6E-10
PPI-MO1-Lo MP 100, 1, 54 1.1E-06 1.8E-11 8.8E-07 8.8E-10
PPI-MO1-Hi MP 100, 46, 70 1.5E-06 2.5E-11 1.3E-06 1.3E-09
PPI-MO1-Lo MP 100, 100, 100 1.1E-06 8.9E-11 1.4E-06 4.3E-08
PPI-MO1-Hi MP 100, 100, 100 1.1E-06 1.1E-10 1.1E-06 1.5E-08
Page 28 of 61
Figure 23. WK results for l-PPI and PEI supported on MO1 monolith substrates.
Results from the WK cell testing, when comparing the PEI/monolith to the l-PPI/monolith, suggested
that l-PPI (which is a solid at RT), was not a continuous phase, whereas branched PEI (which is a viscous
liquid at RT) was a continuous phase. When l-PPI recrystallizes during solvent evaporation in synthesis,
relatively small crystallite domains residing in the monolith pores contain numerous micro-defects, which
allow fast CO2 transport across the sample, resulting in higher effective diffusivities (10-8 – 10-10 m2/s) than
expected. These diffusivity values were significantly higher than the PEI/monolith WK results, where
effective diffusivities ranged from (10-11-10-13 m2/s). The Anodiscs®, which has a relatively low BET
surface area to total pore volume ratio, little to no mesoporosity, and near unity tortuosity, is more conducive
for the solid polymer to form local thin films in the 20-200 nm macropores. When the solid polymers are
Page 29 of 61
impregnated on the monoliths, with much higher tortuosities and higher BET surface area to total pore
volume ratios, the polymer recrystallization during methanol evaporation (after impregnation) results in
small irregular crystallite domains with numerous voids for CO2 to permeate through the support during
the WK experiment. Liquid polymers have the opposite behavior. The liquid polymer (branched PEI) was
not retained in the Anodisc pores due to gravity and lack of surface area. On the other hand, the liquid
polymer was retained in the monoliths, and the films needed for accurate WK data were achieved.
The WK results for l-PPI/MO1 supported materials under dry sorption conditions suggest that the CO2
uptake rates for l-PPI/MO should be fast, due to the higher diffusivity, which is not inconsistent with the
core sample data in Figure 10 showing the highest capacity under dry conditions. Thus, it is important to
resolve the impact of humidification on the CO2 diffusivity in l-PPI.
Humidity is known to enhance the CO2 sorption capacity in supported PEI due to the higher theoretical
amine efficiency (AE, calculated as mmol CO2/mmol N of sorbent), arising from difference in sorption
mechanism, as compared to water-free conditions (CO2/N =1.0 – humid and 0.5 – dry). Another advantage
of humidity is the plasticizer effect that water can heave on the PEI chains, enhancing their mobility,
resulting in faster transport. These effects can outweigh the additional amine sites that water occupies,
resulting in higher sorption capacities (in some cases). In the earlier core sample and SBST testing of l-PPI
monolith samples, we observed that feed gas humidification had a significant inhibitory effect on the rate
of CO2 uptake in the monolith. The extra methyl group in l-PPI results in significant hydrophobicity, as
compared to the hydrophilic PEI. It is hypothesized that because water prefers to interact with itself rather
than l-PPI, the formation of water clusters hinders CO2 transport in the l-PPI monolith substrate, resulting
in lower sorption capacities relative to water-free conditions. Understanding the mechanism of the
inhibition is critical to finding a mitigation method. If the inhibition is due to a decrease in the CO2 diffusion
rate in the presence of H2O, adjusting the porosity of the monolith substrate could potentially reduce the
inhibition extent. Thus, we performed WK cell testing with feed humidification to yield further insight into
this behavior.
WK cell tests were run using the l-PPI/200nm Anodiscs® sample at ~100% pore fill (~23wt%) to
determine whether the presence of adsorbing gas moisture inhibits the CO2 diffusion rate through the
polymer. Figure 24 shows the average diffusivities for 100% pore filled l-PPI/ Anodiscs® as a function of
humidity level measured with the WK system. It is clear from the data that moisture does not impede the
diffusion rate of CO2 through the l-PPI polymer material.
Page 30 of 61
Figure 24. Polymer diffusivities of l-PPI/Anodiscs® as a function of %RH at 30°C.
In addition, l-PPI was impregnated on the 200nm pore width Anodiscs® membranes, to a targeted 90%
volumetric pore fill (18wt%) to determine if the CO2 transport, under humid conditions, was operating in
pore flow regime, or if the water vapor was blocking the pores, hindering CO2 transport. The hypothesis
was that the water would form a film in the remaining 10% free pore volume creating an additional mass
transfer resistance. The CO2 flux and gaseous diffusivity for the 90% pore filled AAO was the same
regardless of the presence of water vapor (8.5 x10-5 mol/m2 s and 3.4 x10-7 m2/s respectively), showing that
the water vapor in the large pores of the Anodiscs® material did not hinder CO2 transport in the remaining
10% free pore space in AAO.
To test the porous monolith substrate material and compare its behavior to the Anodiscs® material, a
sample of MO2-1-high macro PV monolith (note: MO2-1 has 2x higher meso-PV than MO1 and was part
of the monolith optimization work in BP2) was impregnated with ~15wt% l-PPI and then run in the WK
experiment; the obtained kinetic parameters are shown in Table 5. As observed previously, the presence
of moisture did not impact the diffusivity of CO2 through the l-PPI loaded monolith walls.
Table 5. WK results for 15wt% l-PPI MO2-1-high macro PV monolith at 30°C.
Page 31 of 61
The WK testing under dry and humid conditions conclusively shows that moisture does not inhibit the
diffusion rate of CO2 through l-PPI; thus, the inhibition effect of air feed humidification on CO2 adsorption
for supported l-PPI/MO1 at <100% pore fill (Figure 10) must result from a kinetic inhibition of the pooled
water surrounding the discrete l-PPI domains making it more difficult for CO2 to access the l-PPI during
the transient adsorption step of the DAC cycle. Monolith optimization then focused on trying to reduce the
impact of this inhibition on l-PPI performance.
One of the project’s success criteria (summarized in Table 2) was that transport data (in l-PPI) obtained
from at least two different methods show <50% variation. To validate this criterion, the data from the WK
cell testing (Figure 23) can be compared to transport data obtained by modeling TGA uptake data. For the
modeling of the TGA uptake data (l-PPI loaded on MO1 was used for this purpose), the effective diffusion
path length was estimated to be ~25 nm, by using the pore volume of the bare support and l-PPI impregnated
sample and assuming a cubic particle with a layer of l-PPI deposited on the support’s surface. The l-PPI
loading on the MO1 powder sample was 17wt%, which equates to a volumetric pore fill fraction of 70%.
Figure 24B shows the experimental and model CO2 uptake curves, as well as the average effective
diffusivity versus CO2 uptake for MO1-supported l-PPI at 30°C and 400ppm. The polymer diffusivity from
the WK and TGA agree, meeting the success criteria metric.
Figure 24B. TGA experimental and model uptake curves and diffusivity versus uptake, and the average
effective diffusivity versus CO2 uptake for l-PPI impregnated MO1 powder at 30°C and 400ppm (dry) CO2.
Page 32 of 61
ADSORPTION ISOTHERM TESTING
The TGA/DSC with humid components, used by Global Thermostat to collect the data for this subtask,
is shown below in Figure 25. Initial testing focused on validating CO2 adsorption capacity measurements
for l-PPI in several baseline porous metal oxides under dry adsorption conditions, to ensure that the system
was working correctly. The humid TGA method was then developed, with the main focus on characterizing
the range of temperature and relative humidity setpoints under which good stable data could be obtained.
Figure 25. Humid TGA/DSC used for the adsorption isotherm testing located at Global Thermostat.
A dry isotherm was performed at 30°C with varying CO2 concentrations between 150-980 ppm, on a l-
PPI sample impregnated onto MO1 substrate at ~50% pore fill to be fit to the appropriate model to extract
isotherm parameters. Figure 26 shows dynamic CO2 uptake curves, with CO2 capacity increasing as
concentration increases.
Figure 26. Dry CO2 adsorption isotherm testing for l-PPI on MO1 at 30oC.
Page 33 of 61
Humid CO2 experiments (400 ppm) were also performed at varying pore fills of PPI in MO1 at one
dewpoint (9°C) and one temperature (30°C). Figure 27 shows the dynamic CO2 uptake curves of both PPI
and PEI at a similar loading (~50% pf) both dry and humid. At this loading, PPI shows much slower uptake
under humid conditions than PEI. It also shows a loss in capacity in the presence of humidity whereas in
PEI an increase is seen.
Figure 27. Humid vs. dry CO2 adsorption isotherm testing for l-PPI and PEI on MO1 at 30oC.
At lower l-PPI loadings in the presence of humidity the opposite trend is observed, where humidity
provides an increase to capacity. Figure 28 shows the amine efficiencies at two loadings of l-PPI in MO1
(21 and 28% pf) both dry and humid. Figure 29 combines the varying loadings of l-PPI (21% to 50% pf),
expressed as amine loading, under both humid and dry conditions. It can be seen at the lowest loadings the
largest increase in amine efficiency occurs under the presence of humidity when compared to dry
conditions. However, as loading increases the presence of humidity decreases amine efficiency.
Figure 28. Amine efficiencies under humid and dry conditions at 30oC for two loadings of l-PPI/MO1.
Page 34 of 61
Figure 29. Amine efficiencies under humid and dry conditions at 30oC for varied loadings of l-PPI/MO1.
Dry isotherms were collected at a range of temperatures between 20oC-50oC with varying CO2
concentrations between 150-980 ppm, on a l-PPI sample impregnated into MO1 substrate at ~50% pore fill.
Figure 30 shows amine efficiency plotted vs. CO2 concentration, with 30℃ showing the overall highest
amine efficiencies, and 50℃ showing the overall lowest amine efficiencies. Whereas, both 20oC and 40oC
show similar amine efficiencies. At all temperatures amine efficiency increases as CO2 concentration
increases. Figure 31 shows the dynamic CO2 uptake curves at both the highest and lowest CO2
concentration to point out that a slightly slower uptake occurs at 20oC than with the other temperatures.
Figure 30. Amine efficiency vs CO2 concentration from dry isotherms at 20-50℃.
Page 35 of 61
Humid isotherms were collected to be used for the model at both 30oC and 40oC. Figure 32 shows the
dynamic CO2 uptake curves (left) and amine efficiencies vs CO2 concentration (right) at 30oC, at 9oC
dewpoint (27% RH). Just like under dry conditions amine efficiency increases as CO2 concentration
increases. The dynamic uptake curves are significantly slower and have not fully reached pseudo-
equilibrium conditions, a phenomenon also observed when analyzing powders at 400ppm CO2
concentrations. Figure 33 shows the differences between humid and dry conditions at 30oC, with dry
conditions resulting in slightly higher amine efficiencies and faster uptakes than humid conditions.
Figure 31. Dynamic CO2 uptake curves from dry isotherms at 20-50oC.
Figure 32. Dynamic CO2 uptake curves for humid isotherms at 30oC (left); amine efficiency vs CO2
concentration (right).
Page 36 of 61
Figure 33. Humid vs. dry 30oC isotherm data amine efficiency vs CO2 concentration (left) and dynamic
CO2 uptake curves (right).
Figure 34 shows the dynamic CO2 uptake curves (left) and amine efficiencies vs CO2 concentration
(right) at 40oC, at 9oC dewpoint (27% RH). Just like under dry conditions, the amine efficiency increases
as CO2 concentration increases. The dynamic uptake curves are significantly slower and have not fully
reached pseudo-equilibrium conditions, a phenomenon also observed when analyzing powders at 400 ppm
CO2 concentrations. Figure 35 shows the differences between humid and dry conditions at 40oC. There is
an inflection in the trend at 330 ppm, with humid conditions having a much higher amine efficiency than
dry conditions. That data point was re-run, and the amine efficiency stayed the same. There was no
significant trend that showed amine efficiency was higher in dry samples at 40oC than in humid samples at
40oC. The dry conditions: however, showed faster CO2 uptake than humid conditions, which trended the
same for 30℃. Finally, Figure 36 shows both humid/dry amine efficiencies plotted together, where the
most conclusive trend is that 30oC under both humid and dry conditions showed higher amine efficiencies.
Figure 34. Dynamic CO2 uptake curves for humid isotherms at 40oC (left); amine efficiency vs CO2
concentration (right).
Page 37 of 61
Figure 35. Humid vs. dry 40oC isotherm data amine efficiency vs CO2 concentration (left) and dynamic
CO2 uptake curves (right).
Two different dewpoints of 3oC and 9oC (18% RH and 27% RH respectively) were evaluated at 30oC
for the model shown in Figure 36. Overall, the higher dewpoint resulted in slightly higher amine
efficiencies, with an increase in efficiency seen as CO2 concentration increases. The uptake dynamics were
very similar amongst the two dewpoints, with the 9oC appearing slightly faster.
Figure 36. Amine efficiency vs. CO2 concentration at 3℃ / 9℃ dewpoints (left) and dynamic CO2 uptake
curves (right).
Samples of the higher meso PV MO2 monolith substrates, specifically MO2-1-Hi MP and MO2-3-Hi
MP, were tested for CO2 adsorption under humid conditions. The test samples were prepared by cutting
one channel of each monolith sample then crushing it to be analyzed as a powder in the TGA. A 50% pore
fill with linear PPI (l-PPI) was targeted for both substrates. The appropriate amount of l-PPI was measured
and dissolved in methanol. Once the l-PPI solution was homogeneous, the substrate was added and stirred
for 6 hours. Excess solvent was then removed via rotary evaporation, and the samples were dried under
vacuum. Loadings were confirmed via TGA burnoff method. Humid CO2 adsorption studies were
conducted using TGA on the l-PPI-impregnated MO2 substrates, as well as l-PPI-impregnated MO1 to
Page 38 of 61
compare the substrate effects on CO2 capacity. Each sample is first pre-humidified at 9°C dewpoint
(27%RH) at 30°C under N2 gas, then CO2 adsorption begins. These experiments use 400 ppm CO2 and take
roughly 18 hours per sample to complete. Figure 37 shows both the CO2 capacity and amine efficiency of
l-PPI run in the baseline MO1 as well as the two MO2 substrates. On the left, CO2 capacity (calculated as
mmol CO2/g support) is plotted as a function of time. On the right, CO2 capacity normalized to the polymer
sorbent amine loading of l-PPI, known as the amine efficiency (calculated as mmol CO2/mmol N of
sorbent) is plotted as a function of time. Each plot compares the CO2 adsorption of l-PPI in MO2-1, MO2-
3 and MO1, respectively. While all substrates show slow uptake under humid conditions, both l-PPI-
impregnated MO2 samples, MO2-1 and MO2-3, demonstrate a higher amine efficiency than MO1. MO2-
1, which was ultimately selected as the optimal MO substrate for humid CO2 adsorption on l-PPI, showed
the highest amine efficiency compared to other substrates (Figure 37).
Figure 37. Humid CO2 adsorption (left) and amine efficiency curves (right) of l-PPI in various substrates.
Figure 38 shows the humid capacity/amine efficiency compared to the previously collected MO1
baseline under dry conditions. On the left, CO2 capacity is plotted as a function of time; on the right, CO2
capacity normalized to the polymer sorbent amine loading of l-PPI, known as the amine efficiency, is
plotted as a function of time. These data show what has been previously observed: that dry uptake conditions
are faster and result in overall higher amine efficiencies than under humid conditions during the same
duration. Both of the MO2 substrates resulted in higher amine efficiencies under humid conditions than the
original MO1 substrate.
Page 39 of 61
Figure 38. Humid CO2 adsorption curves of various substrates compared to dry CO2 adsorption of l-PPI-
impregnated MO1.
Page 40 of 61
SINGLE CHANNEL MONOLITH CONTACTOR MODEL
The single channel monolith contactor model is built on the COMSOL simulation platform, and
incorporates flow, pressure, temperature, reaction/equilibria, and transport, with steam as the desorption
heating source, across a 3D domain of a monolith cell channel and porous wall. A stepwise approach was
taken for the model development. The first two steps in the development of the COMSOL model
framework were related to air flow and pressure drop simulation; the objective of this initial work was to
confirm the pressure loss prediction vs. expectation to assure that the air fluid flow model was set up
correctly. Step 1: established the flow pressure loss model (see Figure 39; 3D single channel, entrance and
exit space on either side of channel with uniform velocity at inlet). Step 2: added “filter” pressure loss
model (see Figure 40; 3D two channels, alternate entrance/exit plugged, and porous media flow in substrate
and free flow in the channels).
Figure 39. Single channel model. 5 m/s approach velocity. Flow bottom to top with entrance and exit
zones. Velocity profile shown in Figure. Symmetry to show full cell and scaled axially for visual clarity.
Expected pressure loss for monolith is < 200Pa, which agrees well with the simulated value.
Figure 40. Two channel model of porous flow through wall. Model fits the measured data well.
Page 41 of 61
Next, the steam condensation model framework in COMSOL was established, which is important for
simulating the adsorber regeneration process which utilizes steam as the desorption medium. The structure
of the steam condensation test model was as follows: 1D porous wall, Darcy flow with gas
generation/reaction rate, transport of concentrated species: multicomponent diffusion with inert, H2O and
CO2 species, heat transfer in solids, and the COMSOL ODE module to handle the condensed water loading.
For the reaction rate of H2O with the amine, the following rate law was used: (actual – equilibrium
concentration) x k1, with the equilibrium calculated by Raoult’s law using water/amine group concentration
with activity factor, and the k1 rate coefficient chosen for solution stability (note: the mesh size kept coarse
for stability). Example simulation results from the steam condensation model are presented in Figure 41.
Figure 41. Simulation results from the steam condensation model.
We then incorporated the adsorption isotherms and diffusivities for PEI to establish a fully working
COMSOL single channel contactor model and validated it using the SBST commissioning data for the
standard PEI MO1 monolith contactor. Model simulation results are shown in Figure 42 for both the
adsorption and desorption steps of a typical DAC cycle. The model yields a good fit of the SBST
commissioning data (Figure 43), in both the adsorption and desorption steps of the overall DAC cycle.
Note that the elevated signal noise in the mass spec (MS) data during the steam desorption is due to
condensation and evaporation of water in the MS sampling line; this condensation effect is difficult to
eliminate during desorption due to the high concentration of water vapor resulting from the use of direct
steam injection.
Page 42 of 61
Figure 42. COMSOL simulation showing adsorption / desorption velocities and CO2 partial pressures.
Figure 43. COMSOL model simulation vs. actual SBST data (PEI standard monolith contactor).
The COMSOL model parameters were then updated for l-PPI using a combination of the project’s data
outputs from the fundamentals data (CO2 and H2O adsorption isotherm tests from Subtask 3.2 and diffusive
transport rate tests from Subtask 3.3) and the SBST testing of the l-PPI standard MO1 monolith contactor.
The model fit of the SBST data for the l-PPI monolith is shown in Figure 44. The model simulation of the
l-PPI standard MO1 monolith yields a good fit of the SBST data run, as shown in Figure X, in both the
adsorption and desorption steps of the overall DAC cycle. The CO2 MS desorption data again exhibit
elevated signal noise due to water condensation in the MS sampling line during steam desorption.
Figure 44. COMSOL model simulation vs. actual SBST data (l-PPI standard monolith contactor).
Page 43 of 61
APPROACH FOR PPI MONOLITH OPTIMIZATION
In the core sample tests for l-PPI supported on MO1 monolith substrate, we observed that the presence
of moisture in the air feed during adsorption can decrease l-PPI’s CO2 adsorption capacity (recapped in
Figure 45). The WK cell measurements, for the 90-100% pore fill of l-PPI in Anodiscs® samples, revealed
that water does not fundamentally impact the CO2 transport rate through the polymer. And from the TGA
data (also recapped in Figure 45), we determined that the impact of humidity depends on the l-PPI loading:
water promotes adsorption at low l-PPI loadings but inhibits adsorption at higher l-PPI loadings. These
findings provided a basis for the approach to optimize the monolith substrate properties for l-PPI.
Figure 45. Impact of adsorption air humidity. Humid TGA data by Global Thermostat. Core sample data
by CORMETECH.
In general, the focus areas for improvement were to (1) increase the CO2 uptake rate, by modifying the
size and distribution of meso / macro pores in the substrate, to promote faster CO2 diffusion, and the cell
density, for faster CO2 transport (bulk channel to monolith wall), (2) increase the CO2 uptake capacity, by
increasing the total meso pore volume to enable higher l-PPI loadings, by lowering the cell density and
increasing the wall thickness to enable higher l-PPI loadings, and by increasing the amine efficiency to
better utilize the l-PPI, and (3) reduce the H2O inhibition effect on CO2 uptake capacity, by modifying the
mean size / distribution of the meso / macro pores in the substrate, and by optimizing the l-PPI loading.
The l-PPI validated single channel monolith contactor model was used to run simulations to determine
the optimal monolith structure for l-PPI to maximize DAC performance. The outcome of this modeling
optimization work guided the subsequent extrusion of the next generation substrates for l-PPI. The
simulation work focused on three areas to improve performance: (1) increasing the porous support’s
mesopore volume to increase the l-PPI loading in the monolith and the CO2 uptake capacity, (2) increasing
the macropore volume of the monolith’s porous wall to increase the rate of diffusive transport of CO2 into
Page 44 of 61
the walls and the CO2 uptake rate, and (3) increasing the cell density of the monolith to increase the transport
rate of CO2 from the bulk open channel to the external surface of the monolith’s porous wall. The modeling
simulation constraints and results were as follows; note that simulations were run at both 30oC and 10oC
inlet air temperatures for adsorption to evaluate the temperature impact on the optimal monolith selection:
1. Increasing pore volume (PV) of the mesopores:

 MO1 same meso-PV as BP1 contactor substrate. MO2 represents 100% increase in meso-PV.
 Results (presented in Figure 46):
o Increasing the meso-PV increased both the uptake rate and total capacity.
o Desorption takes longer due to higher CO2 loading, similar shape.
2. Increasing pore volume (PV) of the macropores:
 Lo represents same as BP1 part. Mid and Hi represent ~50% and 100% increases, respectively.
o Increasing the macro-PV slightly increases the uptake rate, while the total capacity
remains the same.
o Desorption rate is slightly faster with increasing macro-PV.
3. Increasing cell density of the monolith:
o No change in total capacity.
o Slower uptake rate for the thick wall 86 cpsi, especially at 35°C.
o Similar adsorption rates for the 140, 170, 325 cpsi.
o Similar adsorption rates for all geometries at 10°C.
o Similar desorption rates for all geometries.
Based on the model simulation results, we selected three extrusion die options for the monolith substrate
cell density and wall thickness to span the tradeoff between bulk adsorber mass and geometric surface area;
the order for the three dies was placed in December 2022 and were ready for extrusions in April 2023: (1)
Control (170cpsi), (2) Lower cpsi (86cpsi) w/thicker wall (same DP 170cpsi), and (3) Higher cpsi (300cpsi).
We then proceeded to batch and extrude a suite of monolith samples to experimentally validate the
simulation results at the core testing scale as well as to gain additional insights for optimization. This
experimental work included:
1. Batching and extruding monoliths with several different MO2 powders with higher meso PV and
different average pore diameters.
2. Batching and extruding monoliths with a range of macroporosity levels.
3. Impregnating monoliths with a range of l-PPI loadings.
Page 45 of 61
Figure 46. Simulation results from the l-PPI single channel monolith model.
Page 46 of 61
Figure 47. Simulation results from the l-PPI single channel monolith model.
Page 47 of 61
CORE TESTING OF OPTIMIZED PPI MONOLITH PROTOTYPES
In the interim period before the new extrusion dies were delivered, CORMETECH batched and
extruded monolith samples using available lab-scale extrusion dies (140cpsi and 86cpsi) to generate
performance and characterization data for l-PPI supported on higher meso PV substrates. These “Set 1”
monoliths utilized three higher meso PV powders (all based on MO2) with a range of meso-pore diameters:
MO2-1 (smallest diameter meso pores), MO2-2 (middle diameter meso pores), and MO2-3 (largest
diameter meso pores), with each powder batched and extruded at three levels of macro-porosity achieved
by addition of a removable pore former: Low MP (no pore former addition), Mid MP (mid-level pore former
addition), and High MP (high-level pore former addition). Thus, the “Set 1” monoliths comprised a total of
nine monolith sample types; note that, with regards to cell density, MO2-1 is 140 cpsi, MO2-2 is 86 cpsi,
and MO2-3 is 140 cpsi. Samples of these monoliths were shipped to Global Thermostat and Georgia Tech
for adsorption isotherm, oxidative degradation, and diffusional transport testing. CORMETECH also
prepared PPI-monolith core samples from the “Set 1” monoliths, to evaluate DAC performance on the core
test scale.
The core test data for these “Set 1” samples (relative to the BP2 target) are shown in Figure 48. It is
clear from the data that MO2-1 with the high MP level is the best performing substrate, as it has the highest
CO2 adsorption capacity (left-side graphs; performance meets the BP2 target) and the lowest extent of
moisture inhibition (right-side graphs). In contrast, the MO2-2 and MO2-3 samples had CO2 adsorption
capacity values that were significantly lower that the BP2 target. These data helped to narrow down the
optimal powder selection towards MO2-1 as the primary high meso PV material focus for the “Set 2”
monoliths using the new extrusion dies.
Figure 48. Core sample test data for the “Set 1” l-PPI monoliths. Conditions: Adsorption = 35oC, 450ppm
CO2, 1.3% H2O, 20% O2; Desorption = 105oC, 100% N2.
Page 48 of 61
The “Set 2” monolith samples made using the three new extrusion dies were as follows: (1) control
(170cpsi), (2) lower cpsi (86cpsi) w/ thicker wall (to yield the same DP as 170cpsi), and (3) higher cpsi
(300cpsi). Twenty-one substrates were extruded in total for “Set 2”, while seven samples were activated
with l-PPI. The primary focus of the “Set 2” work was the MO2-1-high MP substrate, based on the “Set
1” results, but it included several tests to confirm selection of the optimal MO2 powder type and MP level.
The seven l-PPI activated samples were then tested for CO2 capacity in the core test unit, the data for which
are presented in Figures 49, 50, and 51.
Tests to confirm MP level selection (Figure 49):
 MO2-1-low MP, MO2-1-mid MP, MO2-1-high MP; all with 86 cpsi / thick wall.
 Similar to the “Set 1” samples, these data confirm that the high MP level is the optimal selection
for maximizing both the CO2 adsorption capacity and the uptake rate.
Tests to confirm MO2 powder type selection (Figure 50):
 MO2-1, MO2-2, MO2-3: all with 86 cpsi / thick wall and high MP.
 Similar to the “Set 1” samples, these data confirm that MO2-1 is the optimal selection for
maximizing the CO2 adsorption capacity and minimizing the water inhibition effect.
 Note that the water inhibition effect on adsorption capacity exhibits an inverse relationship
with the pore size of the MO2 material.
Test to confirm cell density selection (Figure 51):
 86 cpsi / thick wall, 170 cpsi reference, 325 cpsi: all with MO2-1-high MP.
 The data confirm that 170cpsi is the optimal selection maximizing both the CO2 adsorption
capacity and the uptake rate, balancing low pressure drop as compared to the 300cpsi sample.
From the “Set 2” data, we conclude that the optimal substrate for l-PPI is 170cpsi MO2-1 powder with the
high level of macro porosity addition; this formulation was selected for additional extrusion and synthesis
of the “Set 3” monolith contactors.
Page 49 of 61
Figure 49. Core sample test data for the “Set 2” monoliths (86cpsi) showing the impact of MP level for
MO2-1 on CO2 adsorption capacity (left-side graph) and the normalized CO2 uptake rate (right-side graph).
Test conditions: Adsorption = 35oC, 450ppm CO2, 1.3% H2O, 20% O2; Desorption = 105oC, 100% N2.
Figure 50. Core sample test data for the “Set 2” monoliths (86cpsi) showing the impact of MO2 powder
type on CO2 adsorption capacity (left-side graph) and the extent of water inhibition (right-side graph). Test
conditions: Adsorption = 35oC, 450ppm CO2, 1.3% H2O, 20% O2; Desorption = 105oC, 100% N2.
Figure 51. Core sample test data for the “Set 2” monoliths (MO2-1 powder, high MP level) showing the
impact of monolith cell density on CO2 adsorption capacity (left-side graph) and the normalized CO2
uptake rate (right-side graph). Test conditions: Adsorption = 35oC, 450ppm CO2, 1.3% H2O, 20% O2;
Desorption = 105oC, 100% N2.
Page 50 of 61
PPI OXIDATIVE STABILITY TESTING
Accelerated oxidation experiments to collect oxidation/degradation curves for l-PPI at elevated

temperatures were completed for both humid and dry conditions using differential scanning calorimetry
(DSC). A total of four test temperatures were tested: 150℃, 162.5℃, 167.5℃, and 175℃. The dewpoint
of the TGA feed gas was 69℃ for all four test temperatures and the O2 concentrations were 17%. Batch
#3 l-PPI (large batch) was chosen for these sets of experiments. The same sample type (MO1 support) was
used at all temperatures with 17wt% PPI loading, which corresponds to a pore fill of 54%. Figure 52 (left
graph) shows the oxidation/degradation curves for three temperatures performed under humid conditions.
Overall, there was no difference in oxidation extents between 162.5°C and 167.5°C; however, as expected,
the 175°C test had a faster oxidation rate than the lower temperatures. The DSC signal under humid
conditions was noisy, due to the water vapor generator pulsing in pure water vapor, thus the curve shapes
were not as smooth under humid conditions as they were under dry conditions. Figure 52 (center graph)
shows the four temperatures run under dry conditions. As expected, higher temperatures correspond to a
faster oxidation rate; the largest difference was seen between 150°C and 162.5°C. Figure 52 (right graph)
presents both the dry and humid oxidation curves plotted together for three temperatures; it is evident that
humid conditions slightly accelerated the oxidation rate, however, it the curve shapes at 162.5°C and
167.5°C are different than their dry counterparts, with a faster rise to ~50% oxidation and then a slower
approach to 100% oxidation.
Figure 52. Data for the extent of oxidation of l-PPI at three test temperatures under humid adsorption
conditions (left graph) and dry adsorption conditions (center graph); the humid and dry tests are also
overlaid for easier comparison (right graph).
The quasi-accelerated humid oxidation testing worked to bridge the gap between the purely accelerated
tests presented above with truer lifetime measurements. Since MO2-1-High MP (high macropore volume)
was the optimal monolith core composition, it was selected for the quasi-accelerated humid oxidation
testing which involved the DAC cyclic experiments at elevated air temperatures to evaluate the loss in
Page 51 of 61
capacity of l-PPI. In the testing, samples of l-PPI on MO1 and the base PEI on MO1 were also included
for comparative purposes.
The monolithic core samples were first baselined to measure their initial humid CO2 capacity; all test
results were as expected and previously confirmed and reported in the project. After the initial baselines
were completed, the l-PPI monolith samples were exposed to a series of cyclic adsorption/regeneration
exposure cycles representing the quasi-accelerated aging test. Two data points for CO2 adsorption capacity
were taken at 191 hours and 389 hours of aging to chart the extent of oxidative degradation and capacity
loss. The data are represented as the relative change in CO2 capacity compared to the baseline measurement
over the total accelerated testing hours. The data validated that the l-PPI polymer was indeed more resistant
to oxidation degradation than the base PEI.
Page 52 of 61
SBST TESTING OF OPTIMIZED PPI MONOLITH
The data for the “Set 1” and “Set 2” monoliths confirmed that the optimal substrate for l-PPI is 170cpsi-
MO2-1, with the high level of macro porosity (MP) addition; this formulation was selected for the extrusion
and synthesis of the “Set 3” monolith contactor used for SBST testing (i.e., the segmented full-size part).
Before activation commenced, however, additonal core samples were prepared with the 170cpsi-MO2
substrate to confirm the optimal PPI loading to maximize DAC performance (i.e., to minimize the water
inhibition effect and meet the BP2 succes criteria capacity target). The data for these variable PPI loadng
core samples tests are shown in Figure 53. The humid CO2 adsorption capacity increases with higher PPI
loading (top-left). At lower PPI loading, humidity does not inhibit the CO2 adsorption capacity (bottom-
left); a series of humidication experiments with differing sequences of moisture introduction and
concentrations (top-right) confirms this result. At higher PPI loadings, humidity inhibits the CO2 adsorption
capacity (bottom-left) which is confirmed with the additional humidification experiments (bottom-right).
Figure 53. Right graphs: Adsorption = 35oC, 450ppm CO2, 0% (dry) or 1.3% H2O (humid), 20% O2;
Desorption = 105oC, 100% N2. Left graphs: Adsorption = 35oC, 450ppm CO2, 0% to 4.5% H2O, 20% O2;
Desorption = 105oC, 100% N2; method of humid treatment is specified in the data labels (H2O Pre 
CO2/H2O = H2O + N2 pre-treatment followed by CO2 + H2O + N2 adsorption; CO2/H2O = immediate CO2
+ H2O + N2 adsorption; H2O Pre  CO2 Dry = H2O + N2 pre-treatment followed by CO2 + N2 adsorption.
Page 53 of 61
CORMETECH synthesized the full-size l-PPI monolith in quarters, to conserve usage of the batch 3 l-PPI,
using the 170cpsi-MO2-1 substrate and the selected PPI loading (green dashed vertical line in Figure 53)
that yielded maximum adsorption capacity and minimal water inhibition. The synthesis of the optimized
monolith was completed in July 2023. Figure 54 shows the four quarters of the l-PPI monolith after
impregnation (left-side picture) and as a composite contactor (right-side picture) before loading into the
SBST; the overall dimensions of the final assembled monolith face was ~ 130mm x 130mm square.
Figure 54. Left: optimized PPI monoliths post activation; Right: composite prior to loading into the SBST.
The SBST test campaign for the l-PPI 170cpsi MO2-1 monolith contactor ran from 7/31/2023 to
8/14/2023. The data are presented in Figure 55, clearly showing attainment of the BP2 performance target
across the entire range of tested conditions, with an average ~2x capacity increase vs. the BP1 result on the
MO1 monolith.
Page 54 of 61
Figure 55. SBST data for the optimized l-PPI 170cpsi MO2-1 monolith contactor; the H2O content of the
air feed stream ranged 0.8-2.2v%. Note that the final core sample data point (purple triangle) is consistent
with the SBST data.
Page 55 of 61
SINGLE CHANNEL MONOLITH CONTACTOR MODEL UPDATE
Post-SBST testing of the l-PPI optimized monolith, we checked the predictive capability of the
developed single channel monolith contactor model (COMSOL). Figures 56 and 57 show reasonably good
agreement between the COMSOL model and the experimental data for the air adsorption step and the steam
desorption step; note that the “trough” in the SBST CO2 after steam breakthrough is driven by condensation
in the MS.
Figure 56. COMSOL model simulation vs. actual SBST data (l-PPI optimized monolith contactor).
Figure 57. COMSOL model simulation vs. actual SBST data (l-PPI optimized monolith contactor).
Page 56 of 61
In addition, the data from the accelerated aging testing of the PEI-MO1 and l-PPI MO2 samples were
subsequently incorporated into the single channel monolith model, with the assumption of a first order
oxidative degradation kinetic rate law with a linear autocatalytic factor applied to it. Figure 58 shows good
agreement between the model predictions and the accelerated oxidation test results for both the PEI and
PPI at <50% extent of oxidation.
Figure 58. COMSOL model simulation vs. actual accelerated PEI and l-PPI oxidation test data.
Once the oxidation kinetic rate law was incorporated into the COMSOL model,
temperature/composition profiles were then generated for a few adsorption/desorption cycles and then
applied to oxidation model on a rate averaged basis. Figure 59 presents the simulation results: PPI has over
2x greater stability than PEI for the same DAC cycling. Note that the predicted oxidation axial extent is
elevated at the adsorption outlet position (steam inlet position) due to the longer thermal exposure.
Figure 59. COMSOL model simulation of the oxidation extent during DAC cycling of l-PPI and PEI
monolith contactors.
Page 57 of 61
TECHNO-ECONOMIC ANALYSIS (TEA)
A techno-economic analysis provides a levelized baseline to compare differences in the removal cost
of CO2 between technology choices. For the scope of this project, the technology choices are (i) the amine
structure and (ii) the contactor skeleton used to form a DAC sorbent. Specifically, the project was oriented
around evaluating linear poly(propylenimine), (PPI), as a replacement for poly(ethylenimine) (PEI). PEI
is a benchmark amine that has been evaluated in numerous support materials and has well known
advantages and disadvantages. PPI overcomes one of the disadvantages of PEI, offering higher oxidative
stability. The project delivered a modified contactor skeleton into which PPI could be incorporated to form
an improved sorbent contactor for DAC processes.
The project results demonstrated that the PPI-based contactor maintained an enhanced oxidative
stability relative to a baseline PEI-based contactor when exposed to quasi-accelerated DAC cycle
conditions, providing a real improvement to the material lifetime. However, PPI has different adsorption
dynamics than PEI and additionally has a lower overall amine content (mol N / kg polymer). Further, PPI
is not a commercial product, so manufacturing and purchase cost is not established at a commercial scale.
Thus, the objective of the techno-economic and lifecycle analysis presented here was to evaluate the
primary differences in technology performance between a baseline PEI-based contactor and the PPI-based
contactor developed through the course of the project to understand where to focus future R&D and TRL
advancement. The TEA is intended to help the project stakeholders identify where to put resources moving
forward. To support this objective, the analysis is oriented towards changes in the mass and energy balance
around the contactor within Global Thermostat’s process design platform to elucidate differences in the
anticipated OPEX and productivity of the DAC facility. Further, the analysis is done independent of a
specific embodiment scale or site location. Both of these considerations impact the overall techno-
economics and are beyond the scope of the work performed in this project. Being at a low-mid TRL level,
this kind of scoping analysis is appropriate for the level of development as the cost uncertainties of new
sorbents outweigh the level of effort and detail needed to perform cost analysis for a specific project (i.e.
scale, site location, use case).
The scope of the analysis focused on addressing the key uncertainties between the sorbent technologies,
namely: (1) Can the PPI contactor achieve similar productivity and OPEX despite having a lower amine
content per unit of polymer, and different CO2 adsorption dynamics compared to PEI? And (2) What is the
sensitivity of the DAC cost to the ultimate purchase cost of the PPI, given that the PPI is not yet produced
at a commercial scale? This scope was adopted due, in part, to the contactor-replacement feature of Global
Thermostat’s DAC technology. In brief, Global Thermostat DAC plants are designed such that contactors
can easily be changed out in the field, allowing for improvements to be made on existing capital. To support
Page 58 of 61
this analysis, a semi-empirical process model was developed to simulate the dynamic mass and energy
balance around the contactors during DAC cycles. The model uses experimental data as inputs and
generates process flows as outputs. In this way, plant productivity and energy use information are generated
at a level that is appropriate for low-mid TRL technology evaluation and screening. These process data are
then used in conjunction with some basic utility cost information to generate OPEX data. Each scenario
(i.e., trial) from the model generates a projection of annualized CO2 productivity, and heat and electricity
demand. Note that the Global Thermostat DAC technology is transient, and so process flows change based
on a variety of factors including the ambient conditions. In the analysis that is presented, the values of
productivity and energy use represent annual averages for Brighton, Colorado, where Global Thermostat is
located.
Model simulations combined with the economic assumptions above were used in a variety of model
scenarios to identify the relationship between contactor productivity and required operational cost for the
baseline PEI-based contactor, and the project developed PPI-based contactor. Scenarios were run on a
single material and process basis for the PEI-based contactor to form a baseline, and then on several material
and process-based permutations for the PPI-based contactor. Importantly, the process-based permutations
assess for the PPI-based contactor are enabled by the enhanced oxidative stability of PPI compared to PEI.
The results of these scenario simulations reveal an optimal frontier in the Productivity-OPEX relationship.
This frontier characterizes the maximum economic potential of the technology within the constraints of the
evaluation. The process-based permutations enable the PPI-based contactor to obtain a projected
performance frontier that is improved over the PEI baseline; that is that at similar OPEX. The analysis did
not consider the cost of replacing the sorbent material, nor the changeout frequency of the amine. As noted
above, PPI is not a commercial product so its manufacturing and purchasing costs are not known or
established. In this project, PPI was produced and procured at a bench scale, resulting in a high unit cost of
production. However, these costs are not representative of manufacturing costs due to a variety of factors.
While the scalability of the polymer synthesis was explored during the project, finite cost projection for the
polymer remains uncertain. For this reason, we assessed the economic impact of the cost of the PPI by
leveling it on a $/CO2 basis and comparing it to market purchase price for PEI (as of 2023). Including in
this analysis are ranges estimated for the changeout frequency of the PPI and the productivity of the PPI,
each of which has a strong impact on the contribution of the amine cost to the total capture cost. The
analysis shows a wide range of potential impact of the amine cost to the total capture cost based on the
purchase cost of the PPI, as one would sensibly expect. Importantly though, there is some room for cost
increases in the PPI production relative to PEI for the total capture cost to remain beneficial for the use of
PPI, which means that reasonable targets for production costs can be set in the medium term (2-5 years) as
the industry grows.
Page 59 of 61
TECHNOLOGY MATURATION PLAN
The technology under development in this project was at TRL 3 prior to the start of this project: our
leading composition for DAC applications, l-PPI, had been synthesized and validated on the lab-scale.
Now, as a result of this project’s efforts, the adsorber technology has attained TRL 5 (i.e., the “component”
validated in a relevant laboratory environment). l-PPI has been synthesized on the kilogram scale (Subtasks
4.1 and 4.2). Potential improvements have been identified in the synthesis process to enable cost reduction
for further scale up efforts (Subtask 4.3). The produced l-PPI has been incorporated into standard and
optimized monoliths (in Subtasks 4.4 and 8.2, respectively), and then evaluated for DAC performance on
the bench-scale using temperature vacuum swing adsorption (TVSA) with steam as the heating medium,
under process conditions similar to the Global Thermostat DAC process system (in Subtasks 4.5 and 8.3).
The CO2 adsorption capacity data obtained from the single brick sorption tester (SBST) testing campaign
for the l-PPI optimized monolith contactor (Subtask 8.3) exceeded the budget period 2 capacity targets for
the optimized l-PPI monolith. We also confirmed the significant oxidative stability benefit of l-PPI over
PEI, as shown in the isothermal DSC testing (Subtask 4.2) and the quasi-accelerated humid oxidation testing
(Subtask 7.2).
Given the favorable performance results demonstrated in the project on the bench-scale, in particular
with respect to the oxidative stability of l-PPI, post-project, the next steps for the technology development
would be to work towards attainment of either TRL 6 or 7.
The Global Thermostat DAC process design is capable of receiving new and future generations of
adsorbents materials, particularly in monolithic form, and both engineering-scale and full-scale systems
would be available for testing.
CORMETECH already has the production facility available for large-scale monolith manufacture (of
SCR catalyst, oxidation catalyst, and CO2 adsorbers), and the optimized l-PPI monolith substrate
formulation and monolith cell type are within CORMETECH’s capability to scale-up.
It is noted that engineering-scale or commercial-scale DAC unit will require substantially more l-PPI
than was needed for the bench-scale work. l-PPI is not a commercially available product, so the
manufacturing and purchase cost is not yet established at a commercial scale. A quote for ton-level
quantities of l-PPI has not yet obtained, as a suitable manufacturer for that scale will need to be identified
(that work was outside the scope of the current project, which focused on bench-scale testing validation,
and the production of sufficient l-PPI to produce a single monolith brick for testing).
Page 60 of 61
REFERENCES
1. Sujan, A. R.; Pang, S. H.; Zhu, G.; Jones, C. W.; Lively, R. P., Direct CO2 Capture from Air using
Poly(ethylenimine)-Loaded Polymer/Silica Fiber Sorbents. ACS Sus. Chem. Eng. 2019, 7 (5), 5264-
5273.
2. Holewinski, A.; Sakwa-Novak, M. A.; Jones, C. W., Linking CO2 Sorption Performance to Polymer
Morphology in Aminopolymer/Silica Composites through Neutron Scattering. J. Am. Chem. Soc. 2015,
137 (36), 11749-59.
3. Pang, S. H.; Lively, R. P.; Jones, C. W., Oxidatively-Stable Linear Poly(propylenimine)-Containing

Adsorbents for CO2 Capture from Ultradilute Streams. ChemSusChem 2018, 11 (15), 2628-2637.
Page 61 of 61

2371863

Uploaded by

Copyright:

Available Formats

2371863

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

2371863

Uploaded by

Copyright:

Available Formats

Final Technical Report / Bertole et al. / CORMETECH Inc.

FINAL TECHNICAL REPORT

Reporting Period: September 15, 2021 – December 14, 2023

SUBMITTED UNDER AWARD NUMBER

CORMETECH, Inc. (UEI = U2U7UKL8DJQ8)

11707 Steele Creek Road, Charlotte, NC 28273

Dr. Christopher J. Bertole

(919) 620-3524, [email protected]

Georgia Tech Research Center

Dr. Christopher W. Jones, Dr. Ryan Lively, Dr. Athony Vallace

Global Thermostat Operations, LLC

Dr. Miles Sakwa-Novak, Dr. Stephanie Didas, Dr. Eric Ping

U.S. Department of Energy

National Energy Technology Laboratory

Key lessons learned from this project included:

Figure 1. Schematic diagram of the Global Thermostat DAC process.

Task 1 – Project Management and Planning

Budget Period 1 Tasks

Task 2 – Bench-Scale SBST

Subtask 2.1 – Design and Build; Subtask 2.2 – Commissioning

Task 3 – Experimental Testing and Models Development

Subtask 3.1 – Single Channel Monolith Model Development

Subtask 3.2 – Adsorption / Desorption Testing and Model Development

Subtask 3.3 – Diffusion and Transport Testing and Model Development

Task 4 – l-PPI Production & Validation

Subtask 4.1 – Produce l-PPI at kg scale

Subtask 4.2 – Validate l-PPI Performance

Subtask 4.3 – Assessment of l-PPI Scalability to Commercial Scale

Subtask 4.4 – Impregnate Standard Monolith with PPI for SBST

Subtask 4.5 – Testing of PPI Loaded Monolith on SBST

Task 5 – Techno-Economic Analysis (TEA) and Process Modeling

Budget Period 2 Tasks

Task 6 – Experimental Testing and Models Refinement

Subtask 6.1 – Single Channel Monolith Model Refinement

Subtask 6.2 – Adsorption / Desorption Testing and Model Refinement

Subtask 6.3 – Diffusion and Transport Testing and Model Refinement

Task 7 – Sorbent Oxidation Experiments to Validate l-PPI Stability

Subtask 7.3 – Quasi-Accelerated Humid Oxidation Testing

Subtask 8.3 – Testing of Optimized Monolith/l-PPI Compositions in SBST

Task 9 – TEA and Process Modeling

Subtask 9.1 – Process Cycle Opportunity Analysis

Subtask 9.2 – Process Model Updates

Subtask 9.3 – Final TEA and LCA

Table 1. List of project success criteria for budget period 1.

Table 2. List of project success criteria for budget period 2.

SINGLE BRICK SORPTION TESTER (SBST) OVERVIEW AND VALIDATION

LINEAR-POLY(PROPYLENIMINE) SYNTHESIS AND CHARACTERIZATION

Synthesis of 2-methyl-5,6-dihydro-4H-1,3-oxazine (MeOZI) (A): Anhydrous ZnCl2 (5.5 g) was

SBST TESTING OF PPI STANDARD MONOLITH

WICKE KALLENBACH (WK) DIFFUSION CELL TESTING

Figure 19. SEM images of 20nm, 100nm, and 200nm Anodiscs®.

Figure 21. BET plot of “20nm”, “100nm”, and “200nm” Anodiscs®.

Table 4. WK results for l-PPI/ 200nm Anodiscs®.

400 ppm CO2/He 4% CO2/He

Sample Pore Fill Dgas Dpolymer Dgas Dpolymer

PEI-MO1-Lo MP 92, 0, 58 5.2E-07 6.0E-12 8.1E-07 7.4E-10

PEI-MO1-Hi MP 100, 4, 58 6.2E-07 8.1E-12 6.3E-07 6.8E-10

PEI-MO1-Lo MP 100, 58, 95 3.3E-08 4.0E-12 9.7E-08 9.9E-11

PEI-MO1-Hi MP 100, 40, 83 3.2E-07 3.9E-13 3.9E-07 4.6E-10

PPI-MO1-Lo MP 100, 1, 54 1.1E-06 1.8E-11 8.8E-07 8.8E-10