Research Experiences via Integrating Simulations and Experiments (REVISE): A Model Collaborative Research Project for Undergraduate Students in CO2 Sorbent Design

Undergraduate research experiences are an instrumental component of student development, increasing conceptual understanding, promoting inquiry-based learning, and guiding potential career aspirations. Moving one step further, as research continues to become more interdisciplinary, there exists potential to accelerate student growth by granting additional perspectives through collaborative research. This study demonstrates the utilization of a model collaborative research project, specifically investigating the development of sorbent technologies for efficient CO2 capture, which is an important research area for improving environmental sustainability. A model CO2 sorbent system of heteroatom-doped porous carbon is utilized to enable students to gain knowledge of adsorption processes, through combined experimental and computational investigations and learnings. A particular emphasis is placed on creating interdisciplinary learning experiences, exemplified by using density functional theory (DFT) to understand molecular interactions between doped carbon surfaces and CO2 molecules as well as explain underlying physical mechanisms that govern experimental results. The experimental observations about CO2 sorption performance of doped ordered mesoporous carbons (OMCs) can be correlated with simulation results, which can explain how the presence of heteroatom functional groups impact the ability of porous carbon to selectively adsorb CO2 molecules. Through an inquiry-focused approach, students were observed to couple interdisciplinary results to construct holistic explanations, while developing skills in independent research and scientific communications. This collaborative research project allows students to obtain a deeper understanding of sustainability challenges, cultivate confidence in independent research, prepare for future career paths, and most importantly, be exposed to strategies employing interdisciplinary research approaches to address scientific challenges.


Notes for Instructors
Instructors should be aware of the general procedures for synthesis and doping of OMCs, as well as the characterization methods for physisorption and energy-dispersive X ray spectroscopy (EDX) which can be found in the Background and Introductory Materials section.It is recommended the instructor give recommended reading materials the week prior to an experiment, in addition to giving a brief introduction on experimental setup.This research project was designed to be successfully accomplished within an eight-week period with 15-20 hours of time committed per week, suitable for teams of undergraduate students including research experience for undergraduates (REU) students, but may also be extended to a project course for individual junior and senior undergraduates.Through this collaborative educationfocused research project, student understanding can be facilitated by experimental and computational research, though this requires consistent meetings (which could be virtual if needed) as well as regular updates between collaborators.It is recommended that these occur at least once per week to ensure a comprehensive understanding if fostered between students.The following section shows demonstrates outcome insights and future plans, showcasing the activities and learning outcomes expectations throughout this project as well as future plans.
Lesson Plan, Outcome Insights and Future Plans In the first cohort of students, three students were tasked with investigating the effect of loading level of one heteroatom dopant (nitrogen) on sorbent performance as well as comparing it with a second dopant (boron).These material systems were chosen to simplify behaviors and allow for clear relationships to be devised primarily by undergraduate student discussion between experimental and computation groups.Specifically, two undergraduate students were involved in the experimental group and one undergraduate student in the computation group.A room for improvement determined from this first cohort was that expanding the material system would allow for additional material behaviors that would further promote inquiry-driven learning.Based of this, the second cohort in the following summer probed several heteroatom dopants (adding phosphorus and sulfur) in addition to the initial experiments.This cohort had three undergraduate students in the experimental group and one undergraduate student in the computation group.This was found to convolute underlying mechanisms and require students to further spec ulate for hypotheses, though by coupling computational results to experimental findings as well as having weekly meetings between undergraduate students, these behaviors were clearly discussed and analyzed.

CO 2 Capture
The rise of anthropogenic greenhouse gases released into the atmosphere is a worldwide crisis, due to not only the environment impact, including global warming, extreme weather patterns, and the threat of mass extinction, but also in its complexity as mitigation requires addressing the combustion of fossil fuels, which are used in several business sectors, such as manufacturing, transportation, and electricity generation.Specifically, carbon dioxide (CO 2 ) emissions need to be diminished to address these impending challenges.The Paris Agreement in 2015, agreed upon by 194 nations, aimed at setting a goal to limit global warming at a maximum of 2 °C above preindustrial levels and pursue efforts to further limit it to 1.5 °C.This value was determined as a target that would potentially avoid the negative, irreversible impacts climate change would have on society.To achieve this, global CO 2 emission levels must be reduced by 45% by 2030 from levels in 2010 and a carbon neutral society must be achieved by 2050.Not only does this necessitate creative solutions to reduce the combustion of fossil fuels by industries and human society, but it also requires creative solutions to further minimize emission levels.Strategies to capture CO 2 focus on post combustion methods, where CO 2 is collected from the emission source and gas is cooled down and passed through a treatment process that reduces impurity concentrations.These include solvent adsorption, adsorption with solid sorbents, and membrane separation.Of these, solid sorbents are desirable due to their advantages of fast kinetics, long term stability, low energy demand for regeneration, high adsorption and selectivity at ambient conditions, and ease of handling.

Nanomaterials as sorbents
Several solid sorbent technologies have been developed over the past few decades, including zeolites, porous organic polymers, covalent organic frameworks, metal-organic frameowrks, and porous carbons.These rely on adsorption, where CO 2 is attached to the surface of the sorbent.This physical process benefits from the use of nanomaterials due to their physical and chemical properties.Their nanoscale size confers unique characteristics, with fairly high surface reactivity, stability, and the ability to be functionalized to tailor affinity towards specific target molecules.Of these technologies, ordered mesoporouc carbon (OMC) is particularly interesting due to its accessible, uniform pore channels, relatively high surface areas, and tunable carbon matrix.Conventional OMC production relies on a softtemplating strategy, where a carbon precursor (resol) is blended with an amphiphilic surfactant template to direct development of nanostructure.Following self-assembly, the material is crosslinked and carbonized.At these elevated temperatures the crosslinked carbon precursor is converted to carbon while the templating agent is degraded, resulting in a ordered porous carbon matrix.
Tailoring the sorbent design for CO 2 capture can be achieved by functionalization of the OMC matrix with heteroatom doping.By doping with boron or nitrogen, the sorbent capture performance can be modulated, depending on the chemical interaction with the target molecule.Doping of nitrogen into a carbon matrix, for example, can introduce Lewis basicity and form strong interactions with CO 2 molecules.Heteroatom doping of OMCs have been achieved by introducing ammonia gas or solid metal oxides during pyrolysis, which allows for the development of a sorbent design space with heteroatom identity and loading levels being controllable variables to explore how functionality can impact sorbent performance.

Density functional theory
Chemistry is mostly about electrons.The electronic structure of a molecule determines many of its chemical properties, such as reactivity.Getting the electronic structure of a molecular system requires quantum mechanical calculations such as solving the Schrödinger equation, as electrons are microscopic particles that obey quantum mechanics.Density functional theory (DFT) is an efficient way to solve the Schrödinger equation.The basic idea is that the (ground-state) properties of a molecular system depend and only depend on its electron density (first Hohenberg-Kohn theorem).The correct electron density is the one that makes the system reach its lowest possible electronic energy (second Hohenberg-Kohn theorem).Based on these two theorems, we can obtain the correct electron density of any molecular system by minimizing its electronic energy.With the correct electron density, we can then calculate any chemical properties we want.
The problem here is that, how can we calculate the electronic energy from the electronic density?
This requires a functional--a common mistake among students who are new to computational chemistry is to call DFT "density function theory" rather than "density functional theory".A function takes numbers as input and gives numbers as output, while a functional takes functions as input and still gives numbers as output.Here, a density functional E[ρ] takes the electron density as input, which is a 3-dimensional function ρ = ρ(x,y,z) that gives the probability to find an electron at a certain position (x,y,z), and gives the electronic energy as output, which is a number E. So far, we do not have a universal density functional that always gives the exactly correct result, but many good density functionals have been developed and can give reliable results for chemical systems.They rely on approximate theories and/or empirical parameter fitting.The density functional we used in this work, called M06-2X, is a widely used one that has excellent performance in predicting the structure and energy of organic molecules.

Safety considerations
Students must wear personal protective equipment at all times during experiments, including safety goggles, lab coats, gloves, and closed toed shoes.All reactions which evolve noxious and/or combustible chemicals must be performed in a fume hood.Special care must be taken when handling hydrochloric acid and potassium hydroxide due to their corrosiveness.Moreover, care should also be taken when handling tetraethyl orthosilicate (TEOS) as it is flammable as well as being a skin and eye irritant.It is strongly recommended that all instructors and students go through each reagent's MSDS (Material Safety Data Sheet) for more information prior to conducting experiments.During calcination and carbonization steps, the furnace and crucibles must be allowed to cool to avoid potential burn injuries.
Moreover, the tube furnace must be equipped with an exhaust vent or placed in a fume hood to ensure exit gas, which contains byproducts (e.g.CO 2 ) from thermal degradation of compounds.When handling liquid nitrogen for setting up physisorption characterization, cryo-gloves, face protection, and a lab coat must be worn to reduce the risk and severity of cryogenic burns.Moreover, liquid nitrogen should only be handled in rooms with good ventilation.

OMC fabrication
To a 50 mL round bottom flask, dissolve 2.4 g of Pluronic F127 in a mixture of 12 g of ethanol and 1.5 g of 0.2 M HCl.Heat the solution to 42°C and stir for 1 h until a homogenous mixture is obtained.
Following this, add 7.5 g of resol solution (20 wt% in ethanol) and 3.12 g of TEOS and stir at 42 °C for 2 h.
Cast the prepared solution onto petri dishes and dry overnight at room temperature.Crosslink the film at 100 °C for 24 h and calcinate at 350 °C for 2 h under a N 2 atmosphere using an MTI Corporation OTF-1200× tube furnace.After calcination, mix the calcinated sample with the desired dopant at the desired mass ratio by physically grinding with a mortar and pestle.Then, carbonize these mixtures at a rate of 1 °C/min to 600 °C followed by 5 °C/min to 800 °C under N 2 atmosphere.Etch the carbonized powders with a 2 M KOH solution and refresh daily for 3 days to remove silica and byproducts.Following this, wash several times with deionized water, centrifuge, and dry overnight at 105 °C.

Characterization
Gas physisorption behavior was determined here with a Micromeritics Tristar II 3020, though it can be adapted to alternative equipment.For CO 2 sorption, the experiment will be run at two temperatures, 25 °C and 0 °C.Room temperature physisorption requires no additional setup, while 0 °C requires the use of an ice bath.
For N 2 sorption, the experiment will be carried out at room temperature and 77 K by exposing samples to liquid nitrogen.The surface area of samples will be obtained from the 77K N 2 experiment through the Brunauer−Emmett−Teller (BET) equation: where p and p o is the equilibrium and saturation pressure of adsorbate, X m is the monolayer capacity (or the volume of gas adsorbed at standard temperature and pressure), and C is the BET constant.To determine the specific surface area, the BET equation is plotted at relative pressures (p/p o ) between 0.2-0.3 which by BET theory should be a straight line.The monolayer capacity is found from the gradient and intercept, which is then used to determine the total surface area by using the cross-sectional area of the molecule with the following equation: where V is the molar volume of the adsorbed gas and N s is Avogadro's number, or 6.02 x 10 23 molecules/mol.The specific surface area is then calculated from the total surface area over the mass of sample used.This theory relies on several assumptions, such as having a homogenous surface, infinite adsorption at saturation, and limited molecular interactions.Based off the representative experimental results that are consistent with previous reports, increasing doping content is observed to lead to a reduced surface area of OMC materials.The slight increase of surface area at the lowest nitrogen level can be attributed to NH 3 -induced carbon activation, which is a gaseous byproduct upon melamine decomposition.The control and all doped samples exhibited type IV nitrogen adsorption isotherms, which are consistent with mesopore formation, as well as displayed uniform, monomodal pore size distributions, indicating the nanostructures of all OMC samples are ordered.Following this, the pore size distribution will be determined through non-local density functional theory (NLDFT) theory, which interprets the adsorption isotherm in ideal pore geometries by employing classical fluid density functional theory.Using an appropriate model for describing cylindrical pores and carbon surface in our system, the resulting pore size distribution can be obtained.For simplicity, instrument analysis software with a cylindrical model is used for fitting of the model to experimental data as it is not a trivial mathematical process.In the representative experimental results, the averaged pore sizes of all the nitrogen-doped samples are slightly higher than the control, whereas the boron-doped sample is nearly identical to the control.This may be explained by nitrogen doping (which nitrogen atom has an increased atomic size than carbon) swelling the carbon matrix, resulting in slight pore expansion during carbonization.The selectivity of sorbents toward CO 2 over N 2 at room temperature can be investigated through the Henry's Law constant.This value can be calculated by placing the initial slope (<0.2 bar) of adsorption for CO 2 over N 2 , which demonstrates the amount of carbon dioxide adsorbed compared to nitrogen for a sorbent system.Finally, elemental composition of samples can be probed through energy-dispersive X ray spectroscopy (EDX) was conducted on a Zeiss Ultra 60 field-emission scanning electron microscope (SEM).Based off the representative experimental results, samples with increasing dopant mass ratios exhibited a clear trend of increasing doping content, indicating the facile and robust nature of this synthetic approach for preparing doped OMC.
Assessing representative experimental results, it was found that increasing nitrogen content from 7 wt% to 9 wt% and 11 wt%, CO 2 sorption capacity at 1 bar and 25 °C improves from 2.0 to 2.6 and 3.1 mmol/g, respectively.At 0 °C, performance is further enhanced to 3.3, 3.5, and 4.1 mmol/g, respectively.
The control sample, undoped OMC, exhibited a sorption capacity of 1.8 and 2.9 mmol/g at 25 °C and 0 °C, respectively.Notably, while the surface area of the control and lowest nitrogen level sample are higher, they do not exhibit a higher CO 2 sorption capacity than nitrogen-doped OMC samples containing greater nitrogen content.This result indicates that a strong influence of heteroatom functionality on CO 2 affinity and final sorbent performance which overrides the variations in surfaces areas.

Computational Methods
Atomistic molecular models for N-and B-doped OMCs were first constructed on computers with the coronene molecule (denoted as "Molecule C") as the starting point.Three N-doped molecules ("Molecule N1/N2/N3") and three B-doped molecules ("Molecule B1/B2/B3") were constructed to represent the structural motifs shown in Figure 1.N1/N2/N3 correspond to graphitic N/pyrrolic N/pyridinic N, while B1/B2/B3 correspond to BC 3 /BC 2 O/BCO 2 .When necessary, the amount of hydrogen atoms was adjusted (+1 or -1) to maintain the charge neutrality of the molecule.For P-doped OMCs, four structures ("Molecule P1/P2/P3/P4") were constructed by selecting four S-doped structures from previous works ("Molecule S1/S2/S3/S4") and replacing S/S=O with P/P-OH, followed by hydrogen adjustment when necessary.All structures were optimized using density functional theory (DFT) calculations with the M06-2X density functional and the def2-SVP basis set.
On each of the eleven optimized structures (C, N1/2/3, B1/2/3 and P1/2/3/4), one CO 2 molecule was added at different positions with different orientations.Each of these complexes was then optimized using the same density functional and basis set but also with the DFT-D3 dispersion correction to account for the van der Waals interaction between CO 2 and the adsorbent molecule.In some cases, different starting structures led to the same optimized structure.Finally, for each optimized complex, a counterpoise-corrected single-point energy calculation was performed using the same density functional and dispersion correction but with a larger basis set (def2-TZVP) to obtain the final CO 2 binding energy on each molecular model with higher accuracy.This computational protocol established in previous works showed good agreement with current experiments.All calculations were performed using Gaussian16 (revision A.03) on the Lonestar6 supercomputer at the Texas Advanced Computing Center.

Student Expectations for Oral Presentation
To assess student's grasp of important concepts, ability to perform research tasks and acquire important results related to the research project, and ability to analyze and correlate experimental/computational results, students are asked to perform an oral presentation at the end of the research project in week 8. Important background topics include environmental sustainability, CO 2 pollution and sorbent devices, and sorbent design principles, students At the end of 4 th week, students are expected to present a preliminary presentation with a background section that should be fairly close to being finalized, an ongoing work section that should show the progress of results/analysis within the first few weeks, and a future work section that outlines the plan for the rest of the research project.Following this preliminary presentation, students are given feedback from mentors to improve on prior to the final presentation.Additionally, after edits are made based off feedback, students have a discussion with mentors to ensure the development of good fundamentals in a presentation including a well-constructed PowerPoint slide deck, ability to present, and ability to answer questions.A full rubric for the presentation is available below, which all attendees are expected to fill out and students are also expected to self-assess themselves.
The preliminary presentation should be 7.

Table S1 .
Cartesian Coordinates of All Structures Shown in Figure 3.