CO₂ Adsorption in Packed-Bed Columns: Project Scope
What we are studying
Direct Air Capture (DAC) is the technology stack that removes CO₂ directly from the atmosphere — distinct from point-source capture at a power plant or cement kiln. The atmospheric concentration of CO₂ is around 420 ppm, roughly two orders of magnitude lower than a typical flue gas. That dilution sets the ceiling on every other parameter: sorbent capacity, regeneration energy, capital intensity per tonne removed.
Our project sits inside that constraint. We work on a small packed-bed column loaded with polymer-based sorbents, run a controlled CO₂ stream through it, and study how the sorbent saturates over time. The output is a breakthrough curve — concentration of CO₂ at the column outlet as a function of time. Everything we want to know about the sorbent and the process is encoded in the shape of that curve.
Why polymer sorbents
The dominant sorbents in commercial DAC today are amine-functionalised solid supports. They have high CO₂ affinity but degrade in the presence of trace oxygen and moisture, which limits cycle life. Polymer-based sorbents — particularly amine-grafted polystyrene resins and polyethyleneimine-impregnated polymer beads — are a complementary family with different trade-offs: somewhat lower equilibrium capacity, but better mechanical robustness, lower regeneration temperature, and a manufacturing pipeline that resembles ion-exchange resin production.
The specific question we want to answer: under realistic ambient conditions, what is the most useful operating envelope for a packed bed of these polymer sorbents, and how do parameter changes (flow rate, bed depth, particle size, humidity) propagate through the breakthrough curve?
What "useful" means
Useful to us means three things:
- A mass-transfer-coefficient estimate that we trust enough to feed into a sizing calculation.
- A breakthrough time prediction that lands within ±15% across the parameter range we tested.
- A capacity model that tells us how much of the sorbent capacity is usable per cycle.
These are modest goals at the scale of a Year-3 design project. They are also non-trivial — the literature is fragmented, much of it German and Dutch, and the modelling conventions differ across labs.
Who and when
This is ProjID3, my Ngee Ann Year-3 design project, supervised by Prof. Erik Birgersson (NUS) and Dr. Prapatsorn Borisut (SUTD). Interim report in Week 7, final report Week 17, final presentation Week 18. Those dates are the real constraint on everything below.
Where we are
The bench rig is built. Five real runs are recorded and they are the only measured data in the repo — everything else with a tidy filename is synthetic validation, and I label it that way so future-me does not mistake a placeholder for an experiment.
Fitting runs through a 24-model fitter against pseudo-first-order (PFO), linear-driving-force (LDF), and pseudo-second-order (PSO) families. The next post in this series goes deep on those models — what each one is actually saying about the underlying physics, and where each one breaks down. The post after that covers the day we changed the question.
Reading list
A small bibliography that has shaped our framing:
- Glueckauf, Theory of chromatography, Trans. Faraday Soc., 1955.
- Sircar & Hufton, Why does the linear driving force model for adsorption kinetics work?, Adsorption, 2000.
- Wurzbacher et al., Concurrent separation of CO₂ and H₂O from air by a temperature-vacuum swing adsorption process, ES&T, 2012.
- A 1997 dissertation from TU Berlin on amine-grafted polymer sorbents — German, hand-translated and re-typeset for our reference database.