Molecular simulation of carbon capture in MOFs: challenges and pitfalls
Tina Düren Institute for Materials and Processes
School of Engineering University of Edinburgh
Capture – Solid Adsorption @ UEd
http://www.eng.ed.ac.uk/carboncapture/
Hyungwoong Ahn
Process modelling for steady state and dynamic power production cycles
Daniel Friedrich Martin
Sweatman
Lev Sarkisov
Xianfeng Fan
Guilio Santori
Maria-Chiara Ferrari
Stefano Brandani
Tina Düren
Characterisation and development of
novel materials for adsorption &
membrane processes
Integration & optimisation of separation technologies (gas/liquid, membrane, adsorption, hybrid processes)
Process modelling for steady state and dynamic power production cycles
Molecular simulation of CC using MOFs, activated carbons,
mesoporous materials,
supported ionic liquids
Molecular simulation
Simulation methods based on statistical mechanics
e.g. Monte Carlo Molecular Dynamics
Output e.g. adsorption isotherms diffusion coefficients detailed picture on molecular level
Input e.g. model for fluids and solids force fields to describe interactions
adsorption isotherm
µ Tµ T
GCMC simulation
snapshot
CH4 CO2 IRMOF-1
Metal-organic frameworks
Self-assembly from different building blocks
metal organic linker
O O
OO
Br
O O
OO
NH2
Tailoring for specific applications by choosing building blocks
Different linker
molecules
Different pore shapes
Molecular simulation for screening
Screening MOFs for carbon capture with molecular simulations and quantitative structure-property relationship analysis
QSPR results in excellent prediction & ranking
A diverse range of MOFs outside the training set
MOF structure
Uptake Selectivity
Structural descriptors
GCMC ~ hours
< 5 min ~ seconds
Low Pressure CO2 Adsorption: Best MOFs have open-metal sites
• cus significantly enhance uptake of CO2 (and other molecules) at low pressures
• classical molecular simulations fail to predict the adsorption in MOFs with cus
* A. Yazaydın, et al., J. Am. Chem. Soc. 2009, 131 (51), 18198.
CO2 uptake @ 0.1 bar & RT
with cus’s
CO
2up
take
(mg/
g)
CPO-27-
Mg
CPO-27-
Ni
CPO-27-
Co
CPO-27-
ZnCuB
TCMIL
-47 ZIF-8IRM
OF-3IRM
OF-1MO
F-177
0
90
180
270 Experiment Simulation
*MOFs with cus
Challenge 1: Open metal sites
CO2 adsorption on open metal-sites
Linjiang Chen, Tina Düren School of Engineering, University of Edinburgh
Cus are challenging for classical simulations
Sum of vdW radii: 2.8 Å 2.3 Å • Interaction strong & localised
• Between physisorption & chemisorption
Combine classical molecular simulation with quantum mechanical methods to improve predictions
Combining molecular simulation with quantum mechanical methods to improve predictions of gas adsorption in MOFs with open metal sites
Information
Ab init io FFs o derived from first principles o convenient & fast computation o (potentially) transferable
QM calculations GCMC
Ab initio potential energy surface direct & minimal empirical ambiguities computationally expensive & system specific
L. Chen, et al., J. Phys. Chem. C 2012, 116(35), 18899. L. Chen, et al., J. Phys. Chem. C 2011, 115(46), 23074.
Very good predictions with ab initio derived FF: CPO-27-Mg / CO2
* P. Dietzel, et al., J. Mater. Chem. 2009, 19(39), 7362.
10-2 10-1 1000
100
200
300
400
CO2 u
ptak
e (m
g/g)
Pressure (bar)
our FF
exp.
UFF
@ 298 K *
0.2 0.4 0.6 0.8 1.0
20
30
40
Heat
of a
dsor
ptio
n (k
J/m
ol)
CO2 loading (mol/mol)
our FF
exp.
UFF
isosteric heat of adsorption
*
ab initio derived FF extendable to other metals: CPO-27-Co / CO2
Same topology with different open metal sites
Following the same route to FF development
Fewer QM calculations needed: 100 vs. 1000
our FF
exp.
UFF
10-2 10-1 1000
100
200
300
400
CO2 u
ptak
e (m
g/g)
Pressure (bar)
* @ 298 K
* A. Yazaydın, et al., J. Am. Chem. Soc. 2009, 131, 18198.
Ease of derivation: combine ab initio FF with classical FF: Cu-MOFs / CO2
ab initio FF
classical generic FFs
14
First signs that FF are transferable
CuBTC / HKUST-1 NOTT-140a
Challenge 2: Adsorption induced flexibility
CO2 adsorption in MIL-53(Sc) Linjiang Chen, Tina Düren
School of Engineering, University of Edinburgh
Carole Morrison School of Chemistry, University of Edinburgh
David Fairen-Jimenez Department of Chemical Engineering and Biotechnology, University of Cambridge
John Mowat, Paul Wright School of Chemistry, University of St Andrews
MIL-53 – Different structures depending on metal and stimulus
(c) M = Sc, Fe
(d) M = Fe, Cr
(e) M = Fe, Cr
(b) M = Sc, Fe
(a) M = Sc
Very narrow pore (vnp)
Intermediate (int)
Large pore (lp)
Fe
Fe, Sc
Sc Fe, Cr
Closed pore (cp) Narrow pore (np)
Fe, Cr, Sc
MIL-53(Sc) – breathing behaviour CO2 adsorption @ 196 K
Three phases observed (A): closed-pore form (C): large-pore form Both solved experimentally
(B1&2)
Different from known MIL-53 structures
anisotropically broadened diffraction peaks
ambiguity in indexing the cell
→ Ab inito MD
simulated
experimental
- DMF
100 K 293 K 623 K
AIMD simulations
Chen et al., J. Am. Chem. Soc., 135, 15763, 2013
empty structure
CO
2 uptake (mm
ol g-1)
Pressure (bar)
MIL-53(Sc): Response to CO2 adsorption @ 196 K
? Closed pore (cp)
Large pore (lp)
CO
2 uptake (mm
ol g-1)
Pressure (bar)
MIL-53(Sc): Response to CO2 adsorption @ 196 K + 4 CO2
+ 4 CO2
+ 2 CO2 + 2 CO2
+ 2 CO2 + 2 CO2
Narrow pore (np)
Intermediate (int)
or
?
Chen et al., J. Am. Chem. Soc., 135, 15763, 2013
MIL-53(Sc): Intermediate structure energetically favourable
np(2.2)
int(2.2)
Chen et al., J. Am. Chem. Soc., 135, 15763, 2013
int(2.2) np(2.2)
Opportunity: Integration into process modelling
Hydrogen purification using MOFs
Ana-Maria Banu, Tina Düren School of Engineering, University of Edinburgh
Daniel Friedrich, Stefano Brandani School of Engineering, University of Edinburgh
Assessing performance of MOFs
BPDC UiO-67(Zr)
+ 2 zeolites from literature
Cl2AzoBDC Zr-Cl2AzoBDC
BDC UiO-66(Zr)
BrBDC UiO-66(Zr)-Br
Assessing UiO MOFs for H2 purification from steam methane reformer offgas (SMROG)
H2
SMROG offgas
73 % H2 16 % CO2, 4 % CO, 4 % N2, 3 % CH4
Pressure swing adsorption Adsorption: 7 bar
Desorption: 1 bar
UiO-66(Zr) UiO-66(Zr)-Br UiO-67(Zr) Zr-Cl2AzoBDC zeolite
Working capacity: N(7 bar) – N(1 bar)
Impossible to tell from molecular simulation alone which material is best
SMROG: 73 % H2, 4 % N2, 3 % CH4, 4 % CO, 16 % CO2
Banu et al., Ind. Eng. Chem. Res., 52, 9946, 2013
Multiscale study to assess UiO MOFs for H2 purification
H2
SMROG offgas
CH4 CO2 CO N2 H2
3 % 16 % 4 % 4 %
73 %
Continuum model
Molecular simulation adsorption equilibrium micropore diffusion
Breakthrough curves
Ranking of materials
N2 breakthrough (1 ppm in outlet stream)
UiO-66(Zr)-Br could be promising material for H2 purification from SMROG streams
MO
F /
zeol
ite
Banu et al, Ind. Eng. Chem. Res., 52, 9946, 2013
Molecular simulation:
But be aware of the short comings Structural model of adsorbent must be known ab initio derived force fields required to describe adsorption
on open metal sites (relatively expensive) Description of flexibility requires advanced simulation
methods (expensive) Even rigid adsorbents might not be completely rigid
And know what you are doing!
A useful tool for Quantitative predictions Additional molecular-level insight Screening
Molecular simulation of �carbon capture in MOFs: challenges and pitfallsCapture – Solid Adsorption @ UEdMolecular simulationMetal-organic frameworksMolecular simulation for screeningSlide Number 6Slide Number 7Challenge 1: Open metal sitesSlide Number 9Slide Number 10Slide Number 11Slide Number 12Slide Number 13Slide Number 14Challenge 2: Adsorption induced flexibilityMIL-53 – Different structures �depending on metal and stimulusSlide Number 17Slide Number 18Slide Number 19Slide Number 20Slide Number 21Opportunity: Integration into process modellingAssessing performance of MOFsAssessing UiO MOFs for H2 purification from steam methane reformer offgas (SMROG)Working capacity: N(7 bar) – N(1 bar) Multiscale study to assess UiO MOFs for H2 purificationN2 breakthrough (1 ppm in outlet stream)Slide Number 28Slide Number 29