Journal of Magnetic Resonance 346 (2023) 107343

Contents lists available at ScienceDirect

Journal of Magnetic Resonance

journal homepage: www.elsevier.com/locate/jmr

Nuclear magnetic resonance studies of carbon dioxide capture

Suzi M. Pugh, Alexander C. Forse ⇑
Yusuf Hamied Department of Chemistry, Lensfield Road, Cambridge CB21EW, UK

a r t i c l e i n f o a b s t r a c t

Article history: Carbon dioxide capture is an important greenhouse gas mitigation technology that can help limit climate
Received 3 July 2022 change. The design of improved capture materials requires a detailed understanding of the mechanisms
Revised 8 November 2022 by which carbon dioxide is bound. Nuclear magnetic resonance (NMR) spectroscopy methods have
Accepted 21 November 2022
emerged as a powerful probe of CO2 sorption and diffusion in carbon capture materials. In this article,
Available online 5 December 2022
we first review the practical considerations for carrying out NMR measurements on capture materials
dosed with CO2 and we then present three case studies that review our recent work on NMR studies
Keywords:
of CO2 binding in metal–organic framework materials. We show that simple 13C NMR experiments are
Carbon Capture
In situ NMR
often inadequate to determine CO2 binding modes, but that more advanced experiments such as multi-
Operando NMR dimensional NMR experiments and 17O NMR experiments can lead to more conclusive structural assign-
Carbon Dioxide Capture ments. We further discuss how pulsed field gradient (PFG) NMR can be used to explore diffusion of
CO2 capture adsorbed CO2 through the porous framework. Finally, we provide an outlook on the challenges and
Metal–organic framework opportunities for the further development of NMR methodologies that can improve our understanding
Nuclear magnetic resonance of carbon capture.
Diffusion NMR Ó 2022 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY license
CO2 adsorption
(http://creativecommons.org/licenses/by/4.0/).
CO2 abdsorption

1. Introduction to carbon dioxide capture (joules per kg CO2 captured), and (v) long term stability to repeated
cycling. The most established technology for point source carbon
The climate change crisis requires large-scale human behaviour capture employs aqueous amine solutions to capture CO2 and is
changes and the rapid deployment of a range of greenhouse gas used in a number of large-scale demonstration projects. However,
emissions mitigation technologies. Among these, carbon dioxide this technology has some limitations including relatively large
capture and storage is an important technology that can reduce energy consumption for regeneration, corrosion of the steel con-
emissions from point sources.[1] In this approach, CO2 is captured tainers in which the amines are housed, and amine degradation
from gas emissions at industrial sources that may include power (e.g. by oxidation) [3]. An active area of research is, therefore, to
stations, hydrogen production plants, cement factories and steel design new capture materials with improved performance [4–9].
factories. This is achieved using a capture material that selectively A large range of new carbon capture materials are under consid-
absorbs CO2 from the mixture of emitted gases (Fig. 1). The CO2 is eration, not limited to advanced amine solvents, ionic liquids and
then collected from the capture material via the application of heat solid sorbents [4–9]. In this article we focus on solid sorbents,
(and/or a vacuum) and is subsequently stored in the ground where which are generally porous materials that bind CO2 at their inter-
it can be permanently sequestered. A closely related technology is nal surfaces (inside their pores). A wide range of porous materials
direct air capture, where a capture material is instead used to cap- including metal–organic frameworks (MOFs), functionalised por-
ture and remove carbon dioxide directly from the atmosphere. If ous silicas, activated carbons and porous polymers have been
the carbon dioxide is then collected and stored securely, this tech- explored for CO2 capture. For example, a MOF was recently
nology can offer ‘‘negative emissions”, which are increasingly reported to have excellent performance for CO2 capture via the
thought to be important in climate change mitigation pathways physical adsorption of CO2 in the pores of the material [10].
[2]. Another promising strategy is to functionalise porous materials
A good carbon capture material for must generally have (i) large with reactive functional groups such as amines or hydroxides.
CO2 capture capacities, (ii) highly selective CO2 uptake, (iii) fast For example, amine-functionalised MOFs have shown promising
uptake kinetics, (iv) low energy consumption for CO2 desorption performance and have enabled new adsorption mechanisms with
unusual adsorption thermodynamics (see below) [11,12].
⇑ Corresponding author. The design of improved carbon dioxide capture materials
E-mail address: [email protected] (A.C. Forse). requires a detailed understanding of how carbon dioxide binds

https://doi.org/10.1016/j.jmr.2022.107343
1090-7807/Ó 2022 The Author(s). Published by Elsevier Inc.
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
S.M. Pugh and A.C. Forse Journal of Magnetic Resonance 346 (2023) 107343

Fig. 1. Schematic of CO2 capture from industrial emissions. A capture material selectively binds CO2, collecting it from the gas mixture. To collect the CO2 for storage and
regenerate the capture material, energy is normally supplied in the form of heat or a reduced pressure.

and moves within capture materials. Characterisation of carbon tions (Fig. 2). Experiments may then be further categorised into
capture chemistry is often challenging and requires the use of a ex situ and in situ experiments, where ex situ refers to experiments
wide range of experimental techniques as well as simulation meth- on samples that are dosed with CO2 and sealed in the laboratory
ods. Among these methods, nuclear magnetic resonance (NMR) prior to NMR studies, while in situ refers to dosing the sample with
spectroscopy has emerged as a powerful probe of binding chem- gas inside the magnet while simultaneously studying it with NMR
istry and molecular dynamics. NMR experiments can be performed methods (Fig. 2). Each of these approaches has advantages and dis-
on liquids, solids and gases, as well as samples containing a variety advantages, and the optimum method will depend on the exact
of phases. In the case of solid materials, there is no requirement for materials studies and the desired information. Importantly, safety
crystallinity and amorphous materials can readily be studied. In must always be the top priority. CO2 is a colourless and odourless
this article, we review the practical considerations for carrying gas at ambient conditions. Exposure to CO2 may cause headaches,
out NMR studies of carbon dioxide capture. We then show three dizziness, confusion and loss of consciousness. At high levels of
case studies of how NMR methods have been developed to under- exposure, death by asphyxiation could occur (note that CO2 is
stand CO2 adsorption and diffusion mechanisms in MOFs [13–15]. heavier than air, and can accumulate in the laboratory in some
scenarios).
2. Practical considerations for NMR studies of carbon dioxide Fig. 2a shows a common ex situ approach for studying solid-
capture adsorbents for CO2 capture, which we term ‘‘ex situ MAS”
[13,17,21]. In this approach, a MAS rotor is packed with the solid
The first practical consideration is which nucleus to study. adsorbent and is transferred to a gas-dosing manifold, with the
Many studies directly probe the bound CO2 molecules by 13C rotor left open. The sample is then place under vacuum before dos-
NMR spectroscopy. As a spin 1/2 nucleus 13C can readily be studied ing CO2 gas. Following a period of equilibration, the sample is
with a wide range of NMR techniques, though the low natural sealed, for example by using a mechanical plunger, and is then
abundance of 13C (1.11 %) may require the use of signal enhance- removed for MAS magnetic resonance experiments [13]. An advan-
ment techniques, especially since T1 relaxation times are often long tage of this approach is that standard MAS rotors and probes are
for 13C. The use of 13C-enriched CO2 is generally very helpful for used, which can greatly improve spectral resolution for solid sam-
increasing the signal to noise ratios and enabling more advanced ples. If the gas-dosing manifold is designed to have a low volume,
experiments. Beyond 13C NMR, 17O NMR may be used to directly isotopically enriched gases can also be readily used (e.g. 13CO2) to
study the CO2 molecules [14,16]. The very low natural abundance improve signal intensities. A clear disadvantage of this approach is
of this isotope (0.037 %) generally means that enrichment is likely that it is time-consuming to study multiple gas-dosing conditions
to be essential for most experiments. Since 17O is a spin 5/2 (e.g. different gas-dosing pressures, different gas-dosing tempera-
quadrupolar nucleus, solid-state experiments require the use of tures). A further disadvantage is that the dosed gas may escape
high magnetic fields and multi-dimensional experiments to from the MAS rotor over time (depending on how strong the
resolve closely related chemical environments (see later).[14] adsorption is, and how leak-tight the rotor is). In cases where
Beyond directly studying the captured CO2, insight into the capture bespoke rotor dosing equipment is not available, simple alterna-
behaviour can be gained by studying additional NMR-active nuclei tives such as a gas-filled glovebag or chamber may also be
present in the capture material. For example, [15] N NMR experi- employed, provided appropriate safety measures are taken to
ments have been used in amine-based materials, where reaction avoid CO2 exposure. The ex situ MAS approach has been applied
generally occurs at the nucleophilic nitrogen centres [13,17–20]. to a range of solid-adsorbent materials to study carbon capture
A critical further consideration for NMR spectroscopy studies of modes. A number of studies have explored carbon capture chem-
carbon dioxide capture is how to prepare and study the sample in istry in amine-functionalised silicas, [20–26] metal–organic frame-
the presence of CO2. NMR experiments of carbon dioxide capture works, [8,27–32] amine-functionalised metal–organic frameworks,
may be categorized according to whether they are performed with [11,13,14,33–36] zeolites,[37–40] polymers,[41] and porous car-
sample magic angle spinning (MAS), or with static sample condi- bons [42].

2
S.M. Pugh and A.C. Forse Journal of Magnetic Resonance 346 (2023) 107343

Fig. 2. Approaches for studying CO2 capture materials with NMR spectroscopy. a) and b) use MAS conditions for the NMR measurement, which c) and d) perform
measurements under static conditions. In a) and c) the sample is loaded with CO2 ex situ, and then taken to the magnet for NMR studies. In b) and d), the CO2 is dosed inside
the magnet, referred to as in situ.

The disadvantages of the ex situ MAS approach can be partly ular motion averages out anisotropic interactions), nor for many
resolved by the ‘‘in situ MAS” approach (Fig. 2b). Here MAS NMR solid adsorbents, where the considerable motion of CO2 (especially
experiments are performed on the solid adsorbent sample with for physisorbed CO2, with only non-covalent CO2-adsorbent
the simultaneous delivery of CO2 gas to the sample via a hole in interactions) can lead to relatively narrow peaks for adsorbed
one of the rotor caps [43–45]. This approach has the advantages CO2. This method has been used to study a wide range of materials
that the sample is studied directly under operating conditions, including ionic liquids [7,47,48] and metal–organic frameworks
and that the gas dosing conditions (sample temperature and partial [8,15].
pressure of CO2) can be varied continuously. Kinetic studies should Finally, the ‘‘in situ static” approach extends the above approach
also be accessible with this approach [43]. One potential disadvan- to directly dose CO2 gas into the static sample during the acquisi-
tage is the potentially high cost of flowing isotopically enriched tion of NMR data (Fig. 2d) [49,50]. This again has the advantage of
gases unless a gas recovery system can be employed. The flow of enabling the capture material to be studied during operation, while
gases also requires more careful planning with regard to safety also enabling a range of gas pressures and sample temperatures to
issues surrounding gas exposure. When setting up in situ experi- be studied efficiently. One excellent example of these measure-
ments, a further safety consideration is that one must avoid bring- ments explored the CO2 absorption pathways in a wide range of
ing any magnetic parts close to the magnet (e.g. for the gas delivery state-of-the-art aqueous amine CO2 absorbents [50]. CO2 gas was
system). Initial experiments with the in situ MAS approach on bubbled through the aqueous amine solutions inside the NMR
amine-functionalised silicas highlight the great promise of this magnet, and NMR measurements were recorded as a function of
approach [43]. time. These experiments revealed the time-dependent uptake
Two further approaches may be considered that employ NMR mechanisms, which varied significantly depending on the exact
measurements on static samples. These can be applied to both liq- choice of amine. For example, monoethanolamine initially captures
uid absorbents and solid adsorbents. In the ‘‘ex situ static” CO2 via the formation of ammonium carbamate, but at longer reac-
approach, CO2 is introduced to the sample in an NMR tube, either tion times ammonium bicarbonate products began to dominate at
by bubbling CO2 through a liquid or by dosing CO2 into a solid with the expense of the ammonium carbamates. A further excellent
a gas manifold (Fig. 2c) [13–15,46]. After the sample has equili- example used the ‘‘in situ static” approach to study mixed gas
brated, the tube is sealed and taken for NMR experiments. A key adsorption in a solid MOF adsorbent [49]. The role of framework
advantage of this approach is that only conventional solution- flexibility on the selectivity of gas adsorption was explored.
state NMR spectroscopy equipment is required and that both liquid Having reviewed the main practical details associated with
absorbents and solid adsorbents can be studied. A clear disadvan- NMR studies of carbon capture materials, we now give three case
tage is that MAS experiments are not possible. However, we do studies from our research on metal–organic frameworks to high-
note that MAS is not required for liquid absorbents (where molec- light the types of information that can be obtained.

3
S.M. Pugh and A.C. Forse Journal of Magnetic Resonance 346 (2023) 107343

3. Background on metal-organic frameworks into the Metal-N bond to form a negatively charged carbamate
(Fig. 3c) and a neighbouring diamine forms a positive ammonium.
MOFs are a class of microporous materials typically constructed Adjacent carbamates and ammoniums form ion pairing and hydro-
from metal nodes and organic linkers to give three-dimensional gen bonding interactions (indicated by dashed black lines in Fig. 3)
framework structures. Their use as CO2 sorbents is highly favour- to produce chains, explain the cooperative adsorption in these
able owing to their high surface areas and the possibility to pre- materials, and accounting for the step-shaped adsorption iso-
cisely control the pore chemistry. One emerging class of MOFs therms [13,35]. The ammonium carbamate chain product has been
that have shown promise for CO2 capture are amine-appended observed for many diamine-Zn2(dobpdc) analogues, and is thought
frameworks based on the MOF-74 structure, such as M2(dobpdc) to be the most prevalent product [35]. In (dmpn)2-Zn2(dobpdc)
where M = Mg, Mn, Fe, Co, Ni, Cu, Zn, Cd, and dobpdc = 4,40 -dioxi (dmpn = 2,2-dimethyl-1,3-diaminopropane), however, the forma-
dobiphenyl-3,30 -dicarboxylate (Fig. 3a) [51]. In this structure, the tion of carbamic acid pairs was reported, whereby the CO2 mole-
metal ion has a vacant coordination site along the microporous cule reacts with a ‘‘dangling” primary amine, and two carbamic
channels. The appendage of an organic amine compound to these acids interact via hydrogen bonding to form a pair (Fig. 3d) [33].
vacant metal sites (Fig. 3b) gives rise to large adsorption capacities Unfortunately, structure solution by X-ray crystallography gen-
for selective and reversible CO2 uptake [51]. As the nature of both erally requires large single crystals which are difficult to synthesise
the metal cation and the amine can be varied, the adsorption ther- for many analogues of M2(dobpdc). The case studies presented
modynamics can be tuned for a desired application here use multinuclear NMR spectroscopy to explore CO2 adsorp-
[12,33,35,42,46,52]. Amine-appended MOF-74 analogues may be tion on the molecular level without the need for large single crys-
advantageous over other sorbent materials, as they display step- tals and can, therefore, be applied to many different amine
shaped adsorption isotherms, suggesting that CO2 uptake/release analogues in these MOFs. Indeed, we will show how solid-state
is sudden upon exposure at a threshold pressure. As such, these NMR studies on (dmpn)2-Mg2(dobpdc) materials revealed that
materials have lower regeneration energies than traditional amine CO2 can adsorb by a third mechanism where both ammonium car-
scrubbing technologies [12]. bonate chains and carbamic acids are formed on adjacent metal
Basic characterisation methods such as powder X-ray diffrac- sites, stabilised by hydrogen bonding (Fig. 3e) [13].
tion and solution state 1H NMR of digested frameworks can be used
in order to ascertain the long-range ordering of the material and to 4. Case study 1: Elucidating chemisorption in MOFs with 13C and
assess the amine loading, respectively. In these materials a 1:1 1
H experiments
metal:amine ratio can be achieved, indicating that all initially
vacant metal sites are occupied by amines (Fig. 3b). X-ray crystal- As discussed previously, 13C NMR experiments can be imple-
lography studies on large single crystals of amine-appended- mented easily and provide a wealth of information about the num-
Zn2(dobpdc) materials have shown that at least two types of CO2 ber and type of carbon environments present in sorbent materials.
adsorption products can be formed in these materials. The first More detailed information about the binding mechanisms can be
product is ammonium carbamate chains, where CO2 inserts itself gained by exploring the local connectivity of the carbon atoms

Fig. 3. A) structure of theM2(dobpdc). Schematic representations of b) amine functionalisation yielding (amine)-M2(dobpdc) and c-e) CO2 adsorption products c) ammonium
carbamate chains, d) carbamic acid pairs and e) mixed ammonium carbamate – carbamic acid. The dashed black lines indicated hydrogen bonding.

4
S.M. Pugh and A.C. Forse Journal of Magnetic Resonance 346 (2023) 107343

through correlation experiments. However, this approach is typi- spectra of the final adsorption products under MAS conditions.
cally limited by the low natural abundance of 13C and as such, Here we present a case study showing how one-dimensional 13C
the use of isotopically enriched 13CO2 gas and/or more complex MAS and two-dimensional 1H–13C HETCOR experiments can be
sensitivity-enhanced experiments. In solids, the cross polarisation used to investigate binding mechanisms in MOF materials.
(CP) experiment can be used to transfer magnetisation via dipolar The 13C NMR spectrum of (e-2)2-Mg2(dobpdc) (Fig. 4a, e-2 = N-
couplings from a high abundance, high c nucleus, such as 1H, to a ethylethylenediamine) shows two peaks at 162.1 and 124.7 ppm,
low abundance, low c nucleus, such as 13C, thus increasing the sen- corresponding to chemisorbed and physisorbed CO2, respectively.
sitivity and decreasing spectral acquisition time. The build-up of [13] The chemisorbed CO2 signal is enhanced using a CP experi-
magnetisation on the low c nuclei is governed by the distance ment and this signal is assigned to the formation of an ammonium
between the coupled spins, as well as molecular dynamics. By carbamate chain upon CO2 adsorption. In contrast, the (dmpn)2-
using short transfer pulses, only spins in close proximity will be Zn2(dobpdc) shows two chemisorbed CO2 peaks with an intense
selectively enhanced in the NMR spectrum and therefore the CP signal at 161.2 ppm, assigned to carbamic acid pairs and a weaker
experiment can be used as a spectral editing technique. Longer- signal at d = 164 ppm, assigned to a small quantity of ammonium
range couplings can be explored by increasing the transfer time, carbamates (Fig. 4b). It is clear that the number and type of envi-
however, this build-up can be impeded by the T1q relaxation of ronments in (e-2)2-Mg2(dobpdc) and (dmpn)2-Zn2(dobpdc) are dif-
the high c nuclei and as such, the absolute intensities of signals ferent, a range of 13C chemical shift values are reported in the
in these experiments should be treated with care. Furthermore, literature for ammonium carbamate chains and carbamic acids.
cross polarisation can be used to transfer magnetisation in two- The overlapping chemical shifts ranges of the similar 13C environ-
dimensional HETCOR experiments, which allows for the observa- ments in these mechanisms, therefore, means that the confidence
tion of correlations between coupled spins. For these experiments, in the assignments of these binding modes by 13C MAS NMR spec-
an ex situ CO2 loading approach is favoured to gain high-resolution troscopy alone is low.

Fig. 4. 13C pulse-aquire MAS and CPMAS NMR (16.4 T) spectra (top) and 1H–13C HETCOR MAS (bottom) NMR (16.4 T) spectra for (e-2)Mg2(dobpdc) and (dmpn)Mg2(dobpdc)
dosed at  1 bar 13CO2. CPMAS and HETCOR NMR spectra were acquired using 1H decoupling, while the 13C pulse-acquire spectrum was recorded without decoupling. The
dashed black lines indicated hydrogen bonding.

5
S.M. Pugh and A.C. Forse Journal of Magnetic Resonance 346 (2023) 107343

To improve confidence in the peak assignments two- To ascertain the value of 17O NMR as a probe, the 17O NMR
dimensional 1H–13C HETCOR experiments at short contact times parameters were calculated using DFT calculations for amine
were performed, allowing for the investigation of 1H nuclei located appended Mg2(dobpdc) containing a wide range of organic
near the 13CO2 adsorption product [13]. The 1H–13C HETCOR exper- diamines.14 The use of DFT calculations provides an excellent
iment for (e-2)2-Mg2(dobpdc) (Fig. 4a) shows a strong correlation opportunity to explore and compare both known and novel
of the chemisorbed CO2 to a 1H environment at d = 4.3 ppm. This adsorption products. Of the three adsorption mechanisms
1
H signal can be assigned to HNCOO–, confirming the reaction of explored, ammonium carbamate chains, carbamic acid pairs and
the CO2 to the e-2 primary amine species. A weaker correlation the mixed adsorption mechanism, four unique O environments
can also be seen with a 1H signal at d = 13.2 ppm, arising from were identified: MOCO and MOCO–, resulting from ammonium
hydrogen bonding between the adsorption product and the sec- carbamate chains and COOH and COOH, resulting from the forma-
ondary ammonium on an adjacent metal site. As such, these corre- tion of carbamic acids. Each environment can be identified by their
17
lations provide greater evidence for ammonium carbamate chain O NMR parameters, as shown in Table 1. The key distinguishing
formation, where the ammonium cations interact with carbamate feature for carbamic acids is the lower chemical shift, large CQ
anions through coulombic ionic pairing and hydrogen bonding. and small gQ values for the OH. Furthermore, in general, the
Conversely, in the 1H–13C HETCOR experiment for (dmpn)2- metal-bound oxygen can be distinguished from the carbamate O
Zn2(dobpdc) (Fig. 4b) two strong correlations are observed for by a lower chemical shift. These calculations suggested that 17O
the 13C signal at d = 161.2 ppm to 1H signals at d = 5.7 and NMR spectroscopy is a powerful probe into these adsorption mech-
12.2 ppm, assigned to NHCOOH and NHCOOH, respectively. These anisms. Experiments were then performed to explore this further.
d values are in good agreement with DFT calculations for carbamic Firstly, ‘bare’ Mg2(dobpdc) (i.e., where no amine has been
acid pairs and these correlations are notably different from those appended to the vacant metal site) was dosed with C17O2. The
observed for ammonium carbamate chains, thus providing greater resulting 17O MAS NMR spectrum (Fig. 5a) shows a single peak at
confidence in the assignment of this adsorption mechanism for the d = 61.5 ppm, corresponding to physisorbed CO2 is present. This
(dmpn)2-Zn2(dobpdc). suggests that all chemisorption within these materials takes place
as a result of interaction between the CO2 and the amine.
The first amine-appended Mg2(dobpdc) analogue chosen for
study was (ee-2)2-Mg2(dobpdc); the binding mechanism in this
5. Case study 2: 17O NMR as a powerful new probe of CO2 material was confidently assigned as carbamate chain formation
adsorption chemistry using the techniques discussed in Case Study 1. The 17O MAS
NMR spectrum, given in Fig. 5b, shows three signals, corresponding
Whilst 1H and 13C NMR spectroscopy can be useful probes of to the two O environments in chemisorbed (d = 100–200 ppm) and
binding modes in amine-appended MOFs, the similar local envi- a signal at d = 70.5 ppm corresponding to physisorbed CO2. The
ronments surrounding the C atom makes unambiguous assign- extracted NMR parameters of these signals are consistent with
ment of adsorption products difficult. There is a need to develop the 17O NMR parameters calculated using DFT (Table 2) for the for-
new techniques to unequivocally characterise these materials. mation of ammonium carbamate chains and therefore the 17O
From the adsorption mechanisms shown in Fig. 3, it can be seen NMR spectroscopy confirms this binding mechanism for (ee-2)2-
that each binding mode has two unique oxygen sites, arising from Mg2(dobpdc).
a break in the chemical equivalence of the two O atoms in the CO2 The second amine-appended Mg2(dobpdc) material of interest
upon adsorption, and the type of oxygen site present varies drasti- is dmpn-Mg2(dobpdc). As discussed previously, dmpn-
cally depending on the binding mechanism. As such, 17O NMR Mg2(dobpdc) is thought to capture CO2 via a mixed adsorption
spectroscopy should be a highly diagnostic probe to study these mechanism, however, the assignment of this binding mechanism
materials. As discussed previously, the low natural abundance is not conclusive by 1H and 13C NMR spectroscopy owing to the
and quadrupolar nature of 17O mean that isotopic enrichment overlap of signals for similar 13C environments. Excitingly, the
17
and high magnetic field strengths are required to acquire NMR O MAS spectrum (Fig. 5c) for dmpn-Mg2(dobpdc) showed a
spectra. Furthermore, the quadrupolar interaction introduces a
second-order broadening which cannot be removed by MAS. This Table 1
coupling can be quantified by its magnitude (CQ), defined as CQ =- Summary of 17O NMR parameters calculated for diamine-appended Mg2(dobpdc) and
eQVzz/h, and the asymmetry (gQ), defined as gQ = (Vxx–Vyy)/Vzz, Zn2(dobpdc) structures for different adsorption modes. The first two columns show
parameters for ammonium carbamate capture products, and the third and forth
where e is the electronic charge, Q is the nuclear quadrupolar
columns show carbamic acid capture products. The oxygen site referred to in each
moment, h is Planck’s constant and Vxx, Vyy and Vzz are the princi- row is shown in red bold text in the Lewis structure.
ple components of the electric field gradient tensor. The 17O diso, CQ 17
Adsorption product Calculated O NMR parameters
and gQ can be determined by fitting the 17O MAS spectrum. For
spectra containing many overlapping 17O signals, fitting the MAS diso (ppm) CQ /MHz gQ
spectrum can be challenging. This can be overcome by using 155 – 180 7.2 – 8.0 0.5 – 1
two-dimensional MQMAS [53–55] or STMAS [56,57] experiments.
The use of 17O NMR spectroscopy, therefore, provides an additional
set of parameters compared to 13C NMR spectroscopy, which can
be used to classify CO2 adsorption products. One drawback of this 150 – 230 6.5 – 7.8 0.5 – 1
method is the relatively high cost of 17O-enriched CO2. This, there-
fore, precludes the use of in situ experiments where a flow system
is used. Using an ex situ MAS method, acquisition of a high- 150 – 230 6.6 – 7.8 0.5 – 1
resolution 17O MAS NMR spectrum for amine-functionalised MOF
materials can be achieved in approximately 30 min by dosing the
sample at  1 bar.14 This translates to a cost of £50 per sample, 110 – 150 8.0 – 8.4 0.3 – 0.5
although this value could be reduced by optimising the design
(i.e., reducing the volume of gas in the system) of the experimental
set up shown in Fig. 2a.
6
S.M. Pugh and A.C. Forse Journal of Magnetic Resonance 346 (2023) 107343

Fig. 5. (a-c) 17O MAS NMR spectra of (a) bare Mg2(dobpdc) (20.0 T, 12.5 kHz MAS), (b) C17O2-ee-2–Mg2(dobpdc) (20.0 T, 12.5 kHz MAS) and (c) C17O2-dmpn–Mg2(dobpdc) at
20.0 T (12.5 kHz MAS) (top) and 23.4 T (20 kHz MAS) (bottom) and (d) 17O MQMAS NMR (20.0 T, 12.5 kHz MAS) of C17O2–dmpn-Mg2(dobpdc). The NMR parameters for the
17
O MAS NMR spectra recoded at 20 T in (b) and (c) are given in Table 2. The blue lineshape in (c) recorded at 23.4 T is a simulation of the COOH resonance given in Table 2.
(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

broad overlapping of multiple signals corresponding to chemisorp- is now well-resolved from the other overlapping signals, allowing
tion of CO2 and two sharp resonances corresponding to physi- for increased confidence in the deconvolution of the spectrum
sorbed CO2. To aid in the deconvolution of this spectrum, an 17O acquired at 20.0 T. 17O NMR spectroscopy has, therefore, provided
MQMAS NMR spectrum was acquired (Fig. 5d). This spectrum conclusive evidence for the formation of the mixed adsorption
shows 3 signals and the extracted NMR parameters were consis- structure and, excitingly, has provided the first unambiguous evi-
tent with the calculated diso, CQ and gQ values for COO and dence of the formation of carbamic acid species in amine appended
MCOO, suggesting the formation of carbamate chains and a Mg2(dobpdc) analogues (Table 2). Overall, the work presented in
C@O in the mixed adsorption structure (Table 2). However, when case study 2 shows that 17O NMR spectroscopy is a powerful probe
fitting these signals to the 17O MAS spectrum, an additional signal to explore CO2 binding mechanisms in metal–organic frameworks.
at lower chemical shifts is required to fully represent the whole
spectral line shape. Deconvolution of the 17O MAS spectrum with 6. Case study 3: Measuring anisotropic in-pore CO2 diffusion in
4 signals shows that this additional peak can be assigned to car- MOFs
bamic acid OH resonance. This resonance is likely missing in the
17
O MQMAS NMR spectrum owing to inefficiencies in exciting CO2 diffusion is an important physical phenomenon that influ-
MQ transitions for nuclei with large quadrupolar couplings and/ ences the CO2 capture performance of porous solid adsorbents. Fol-
or relaxation effects. To confirm these results, a 17O MAS spectrum lowing on from earlier studies that measured CO2 diffusion in
was recorded at 23.4 T (Fig. 5c. The signal corresponding to COOH MOFs with pulsed-field gradient NMR (PFG NMR) methods,
7
S.M. Pugh and A.C. Forse Journal of Magnetic Resonance 346 (2023) 107343

Table 2
Experimental and DFT-calculated 17O NMR parameters for ee2-Mg2(dobpdc) and dmpn Mg2(dobpdc) compared to the ammonium carbamate pair structure and mixed adsorption
structure, respectively.
17
Compound Adsorption Product O NMR parametersExperiment
(DFT)
diso (ppm) CQ /MHz gQ
(ee2)-Mg2(dobpdc) MgOCO 177 (1 6 9) 6.8 (7.4) 1.0 (1.0)
MgOCO 91 (1 8 6) 6.4 (7.0) 0.8 (0.7)

(dmpn)-Mg2(dobpdc) MgOCO 168 (1 6 6) 6.9 (7.6) 0.8 (1.0)

MgOCO 194 (1 8 3) 7.2 (7.2) 0.7 (0.9)
COOH 217 (2 3 0) 7.6 (7.9) 0.6 (0.5)
COOH 137 (1 3 1) 8.0 (8.2) 0.3 (0.4)

[58,59] we explored CO2 diffusion in the material Zn2(dobdc) pores of the metal–organic framework. On the other hand, the
(dobdc4– = 2,5-dioxidobenzene-1,4-dicarboxylate). Similar to the much slower decay of signal intensity at the right hand edge of
M2(dobpdc) materials discussed above, this material features the spectrum reports on the much slower diffusion of CO2 perpen-
one-dimensional hexagonal channels in an ordered lattice dicular to the MOF pores. Inspection of the spectra in Fig. 6c there-
(Fig. 6a). The Zn(II) ions have a vacant coordination site, which is fore confirms the anticipate result that D//  D\. The above
the primary binding site for CO2. The one-dimensional pores of this arguments were also substantiated by additional experiments in
material, and the lack of connectivity between different pores in which the magnetic field gradients were instead applied along
the crystal structure, should result in a significant diffusion aniso- the laboratory x-axis.
tropy, i.e. CO2 self-diffusion along the pores (D//) is anticipated to Inspired by earlier work,[58] spectral simulations were then
be much faster than diffusion between pores D\. performed to enable a more quantitative analysis (Fig. 6d).
First, large rod-like crystals of Zn2(dobdc) with lengths of 100 s [60,64] The only free parameters in the simulations are D// and
of microns were synthesised via a solvothermal method (Fig. 6a). D\. The simulations shown in Fig. 6d suggested that CO2 diffusion
[60] In these crystals the hexagonal pores run along the length of along the framework pores takes a value of D// = D// = 1.5(4)  109
the rods. The availability of such large crystals ultimately simplifies m2s1. This value is comparable to liquid water diffusion at ambi-
the PFG NMR analysis, since exchange of CO2 between different ent conditions, and is 4 orders of magnitude slower than the diffu-
crystals and with free CO2 gas becomes negligible on the experi- sion of free CO2 gas. For diffusion of CO2 between the MOF pores,
mental timescales (typically tens of milliseconds for PFG NMR we were only able to conclude an upper limit of D\ < 1013
experiments). With large crystals in hand, the ‘‘ex situ static” m2s1, and it was not possible to conclude whether or not D\
approach was used whereby a large number of crystals were had non-zero value. These findings show that the CO2 diffusion
loaded into a valved NMR tube, which was dosed with 13CO2, and between the MOF channels is at least 10,000 times slower than dif-
then taken for static 13C NMR measurements (Fig. 6b). fusion along the MOF channels, i.e. the diffusion is very strongly
A static 13C NMR spectrum revealed a signal with a chemical anisotropic. This finding has implications for the practical applica-
shift consistent with physisorbed CO2 as expected, and with a line- tions of these materials, and suggests that rod-like crystals are in-
shape consistent with an anisotropic chemical shift observed for fact not desired for practical applications due to their long diffu-
the adsorbed CO2 (Fig. 6b). While the lineshape is much narrower sion pathways, and the possibility of poor blockage leading to inac-
than that reported for solid CO2,[61] the observed chemical shift cessible regions of the material.
anisotropy suggests that CO2 molecules have a preferred average Finally, we note that these measurements have also been
orientation relative to the framework pores, consistent with other applied to the larger-pore Zn2(dobpdc) framework.15 Qualitatively
studies of similar frameworks.[28,62] For the conditions studied in similar spectra were obtained, though the increase in pore size
Fig. 6b, signal intensity at the left hand edge of the spectrum with a from 15 to 22 Å led to an increase in D// by a factor of 4 to 6 across
chemical shift d// could be assigned to CO2 adsorbed in crystals ori- a range of CO2 dosing pressures. These measurements therefore
ented parallel to the applied magnetic field, while signal intensity highlighted a straightforward means to tune transport properties
at the right hand edge could be assigned to CO2 adsorbed in crys- through modification of the framework structure. Measurements
tals oriented perpendicular to the applied magnetic field d\. and analysis for the Zn2(dobpdc) sample interestingly also sup-
As was realised in earlier PFG NMR studies,[58,59,63] the obser- ported non-zero diffusion between the MOF pores, which was ten-
vation of a chemical shift anisotropy lineshape provides an excel- tatively attributed to defects in the MOF structure enabling
lent opportunity for the measurement of anisotropic diffusion. additional transport pathways.
Indeed, as the gradient strength was increased in a series of PFG
NMR experiments, clear lineshape changes are observed (Fig. 6c),
which arise from the anisotropic diffusion of CO2. In these experi- 7. Discussion
ments, the pulsed field gradient was applied along the laboratory
z-axis, i.e. co-linear with the main applied magnetic field, B0. The case studies above highlight the power of NMR spec-
Therefore, the rapid decay of signal intensity at the left hand edge troscopy in revealing the chemistry of carbon dioxide capture. A
of the spectrum corresponds to the relatively fast self-diffusion particular success of the approach has been in studying the adsorp-
along the length of the rod-like crystals, i.e. along the hexagonal tion mechanisms that operate and the speciation of CO2 in the cap-
8
S.M. Pugh and A.C. Forse Journal of Magnetic Resonance 346 (2023) 107343

Fig. 6. a) A portion of the crystal structure of the MOF Zn2(dobdc). Inset: microscope image showing the rod-like structure of the MOF crystals. b) static 13C NMR spectrum of
Zn2(dobdc) dosed with 13CO2 gas. Inset: photograph of the valved NMR tube containing the gas-dosed crystals. c) Experimental and d) simulated PFG NMR spectra.

tured state. In some cases NMR studies have revealed new adsorp- A further limitation is that most of the NMR work to date has
tion mechanisms in CO2 capture materials that were not readily been carried out under ex-situ rather than in situ conditions. Time
accessible with other techniques. Moreover, diffusion measure- consuming ex-situ experiments often lead to a limited number of
ments have been used to probe CO2 transport mechanisms in cap- experimental conditions being sampled, and may result in temper-
ture materials. ature and pressure dependent phenomena being missed. Further
A key limitation of much of the work to date is that it has largely method development will help to make in situ techniques more
been carried out under pure CO2 conditions. For real-world CO2 accessible and practical. Importantly, in situ measurements are
separations a mixture of gases is always present, not limited to well-suited to gas mixtures,[49] and should also enable kinetic
CO2, N2, O2, H2O, CH4, NOX and SOX. A number of initial efforts have studies, which have been rare to date [43]. We also note that low
been carried out to study mixed gas adsorption, including CO2/H2O field NMR methods are being developed for the study of adsorption
mixtures,[23,29,65] as well as CO2/CH4 mixtures.[49,66] The role of phenomena. The lower cost of these approaches are particularly
H2O should be studied in particular, since various studies have attractive, and may be well-suited for industrial applications
highlighted that H2O can have a significant impact on adsorption [69]. An important remaining challenge is to differentiate closely
thermodynamics.[65] We further note that few studies have inves- related adsorption products. As discussed above, the 13C isotropic
tigated CO2 capture materials in the presence of O2, despite O2 chemical shifts generally provide poor differentiation of ammo-
often playing an important role in the oxidative degradation of nium carbamate, carbamic acid and bicarbonate species.
many carbon capture materials.[67,68] Studies of adsorbent degra- Approaches that utilise additional NMR parameters, such as 13C
dation in general have also been lacking. chemical shift anisotropies,[25] as well as 17O NMR [14,70] may
9
S.M. Pugh and A.C. Forse Journal of Magnetic Resonance 346 (2023) 107343

help to resolve closely related adsorption products in the future. 3. What is the role of H2O in CO2 capture processes? Does H2O
The availability of increasingly high magnetic field strengths may need to be included or rigorously excluded in these NMR
make 17O measurements particularly attractive [71]. experiments?
Finally, signal enhancement methods based on dynamic nuclear This is a very important question, since water is present in
polarisation (DNP) may enable complex NMR measurements to be almost all target gas mixtures for carbon capture processes (the
carried out without the need for isotopic enrichment. The use of levels of humidity will depend on the specific application). H2O
exogenous radicals as polarising agents may perturb the CO2 cap- uptake often plays a significant role in the kinetics and mechanism
ture modes in some cases, though advances in the use of endoge- of CO2 uptake, especially when solid adsorbent materials are used.
neous radicals for DNP may be well suited for studies of CO2 In some classes of adsorbents the presence of H2O disfavours CO2
capture [72]. DNP methods may also allow for the enhancement adsorption (e.g. Environ. Sci. Technol. 2011, 45, 15, 6670–6675),
of signals from species adsorbed at adsorbent particle surfaces, and H2O adsorption is also hypothesised to impact the speciation
which may impact surface transport barriers. of the adsorbed CO2. In contrast, in other cases the presence of
Ultimately, further NMR method development may lead to H2O may enhance CO2 adsorption, for example due to stabilising
breakthroughs in our understanding of CO2 capture mechanisms, hydrogen bonds between H2O and the carbon capture products
and could enable the design improved carbon capture materials (e.g. J. Am. Chem. Soc. 2019, 141, 33, 13171–13186). In contrast
that can help tackle the climate crisis. to solid adsorbents, absorbent technologies based on aqueous ami-
nes already operate in an aqueous solvent, and water vapour in the
target gas stream will play a much less important role.
8. Questions and answers from reviewers In the NMR studies presented on solid adsorbents in this article,
care has been taken to exclude H2O from the samples, so that the
1. Is there a role for dynamic nuclear polarization (DNP) or other baseline carbon capture mechanisms can be understood under ide-
hyperpolarization methods in 13C, 15N, or 17O NMR studies of car- alised conditions. However, we believe that NMR spectroscopy is a
bon dioxide capture materials, to improve sensitivity and thereby very valuable tool to probe more realistic carbon capture condi-
expand the range of samples or properties that can be studied? tions and to explore the impact of H2O, though there has been rel-
Would significantly higher magnetic field strengths, such as 36 T, atively little work in this direction. Initial NMR studies under
be helpful? humid conditions have, for example, (i) shown how new species
While DNP is a very useful technique for overcoming the chal- may form in the presence of water (J. Am. Chem. Soc. 2018, 140,
lenges associated with low sensitivity nuclei and has been used 28, 8648–8651), and (ii) been used to probe the chemical environ-
to explore adsorbent structure in amine-appended materials ments of co-adsorbed water (e.g. J. Am. Chem. Soc. 2019, 141, 33,
(ChemSusChem. 2017, 10, 266-276) and covalent-organic frame- 13171–13186).
works (J. Am. Chem. Soc. 2018, 140, 22, 6969–6977), it is unclear 4. Is there a simple explanation for the greater spectral com-
how the introduction of an exogeneous radical species (as a polar- plexity in Fig. 5c compared with Fig. 5b? Can the spectra in
isation agent) would affect CO2 adsorption mechanisms. For exam- Fig. 5c at 20.0 T and 23.4 T be fit with the same quadrupole inter-
ple, the radical species may block the MOF channel system or action and chemical shift parameters? Do these parameters also
access to the CO2 binding sites, thus preventing or lowering CO2 explain the 2D MQMAS NMR spectrum in Fig. 5d? Do molecular
uptakes and preventing the study of adsorption mechanisms under motions affect the observed 17O lineshapes?
realistic operating conditions. The use of recently developed The greater spectral complexity in Fig. 5c compared to Fig. 5b
endogenous DNP methods, where a low concentration of radicals comes from the presence of more overlapping signals in the for-
is embedded in the adsorbent as the polarisation agent, may be less mer. In Fig. 5b, the 17O lineshape consists of 2 signals, arising from
disruptive and may offer a more realistic probe of CO2 adsorption one type of adsorbed CO2, assigned to ammonium carbamate
mechanisms. Finally, we anticipate that the use of significantly chains. In contrast, in Fig. 5c, the line shape consists on 4 signals,
higher magnetic fields (e.g. 36 T) would be very useful, especially 2 from ammonium carbamate chain species and 2 from carbamic
for the study of quadrupolar nuclei, including 17O. acid species. The data at the two different field strengths in
2. Can other spectroscopies (e.g., infrared) provide information Fig. 5c were indeed fit with a common set of chemical shift and
about chemical structures of chemisorbed CO2 species, similar to quadrupole parameters (see https://doi.org/10.26434/chemrxiv-
information provided by NMR? What are the relative advantages 2021-09vcw-v3 for the multi-field fit), and the NMR parameters
and disadvantages and complementarities among these were also reconciled with the MQMAS data (although one of the
techniques? resonances was not observable through MQMAS). Finally, we do
Other spectroscopies, in particular infrared, are indeed very anticipate that molecular motions will in general play an impor-
useful for determining the different chemisorption products pre- tant role in influencing 17O NMR lineshapes, though so far, we have
sent. For example, different characteristic wavenumbers have been obtained satisfactory spectral fits for the amine-functionalised
identified for ammonium carbamates, carbamic acids, and bicar- MOF materials system without considering dynamics. Variable
bonate species (e.g. J. Phys. Chem. Lett. 2014, 5, 4194–4200). A fur- temperature NMR studies will be an interesting direction in the
ther powerful technique is X-ray absorption spectroscopy, which future.
has been used to probe CO2 chemisorption mechanisms in
amine-functionalised metal–organic frameworks (e.g. Phys. Chem.
Data availability
Chem. Phys., 2015, 17, 21448–21457). Both techniques are, there-
fore, complementary to the NMR methods described here. Key
No data was used for the research described in the article.
advantages of the NMR methods, in our opinion, are that (i) more
definitive assignments of the sorbed CO2 species are often possible,
especially when DFT calculations on NMR parameters are available, Declaration of Competing Interest
(ii) the proportions of different sorbed CO2 species can be quanti-
fied quite straightforwardly, (iii) different sorbed species (CO2, The authors declare the following financial interests/personal
H2O etc.) can be studied separately by studying different NMR relationships which may be considered as potential competing
nuclei, and (iv) both structure and dynamics of the sorbed species interests: Dr Alexander Forse reports financial support was pro-
can be studied. vided by UK Research and Innovation.
10
S.M. Pugh and A.C. Forse Journal of Magnetic Resonance 346 (2023) 107343

Acknowledgements Navarro, T.K. Woo, S. Garcia, K.C. Stylianou, B. Smit, Nature 576 (2019) 253–
256.
[30] Y.T.A. Wong, V. Martins, B.E.G. Lucier, Y. Huang, Chem. - A Eur. J. 25 (2019)
This work was supported by a UKRI Future Leaders Fellowship 1848–1853.
to A.C.F. (MR/T043024/1). We further acknowledge the many col- [31] J.J. Gassensmith, H. Furukawa, R.A. Smaldone, R.S. Forgan, Y.Y. Botros, O.M.
Yaghi, J.F. Stoddart, J. Am. Chem. Soc. 133 (2011) 15312–15315.
leagues we have worked with on this topic for their contributions
[32] M. E. Zick, S. M. Pugh, J.-H. Lee, A. C. Forse and P. Milner, Angew. Chemie Int. Ed.,
to this work. , DOI:10.1002/anie.202206718.
[33] P.J. Milner, R.L. Siegelman, A.C. Forse, M.I. Gonzalez, T. Runčevski, J.D. Martell, J.
A. Reimer, J.R. Long, J. Am. Chem. Soc. 139 (2017) 13541–13553.
[34] J.D. Martell, P.J. Milner, R.L. Siegelman, J.R. Long, Chem. Sci. 11 (2020) 6457–
References
6471.
[35] R.L. Siegelman, T.M. McDonald, M.I. Gonzalez, J.D. Martell, P.J. Milner, J.A.
[1] M. Bui, C.S. Adjiman, A. Bardow, E.J. Anthony, A. Boston, S. Brown, P.S. Fennell, Mason, A.H. Berger, A.S. Bhown, J.R. Long, J. Am. Chem. Soc. 139 (2017) 10526–
S. Fuss, A. Galindo, L.A. Hackett, J.P. Hallett, H.J. Herzog, G. Jackson, J. Kemper, S. 10538.
Krevor, G.C. Maitland, M. Matuszewski, I.S. Metcalfe, C. Petit, G. Puxty, J. [36] J. Xu, Y.M. Liu, A.S. Lipton, J. Ye, G.L. Hoatson, P.J. Milner, T.M. McDonald, R.L.
Reimer, D.M. Reiner, E.S. Rubin, S.A. Scott, N. Shah, B. Smit, J.P.M. Trusler, P. Siegelman, A.C. Forse, B. Smit, J.R. Long, J.A. Reimer, J. Phys. Chem. Lett. 10
Webley, J. Wilcox, N. Mac Dowell, Energy Environ. Sci. 11 (2018) 1062–1176. (2019) 7044–7049.
[2] H.-O. Pörtner, D.C. Roberts, M. Tignor, E.S. Poloczanska, K. Mitenbeck, A. [37] P. Rzepka, Z. Bacsik, A.J. Pell, N. Hedin, A. Jaworski, J. Phys. Chem. C 123 (2019)
Alegría, M. Craig, S. Langsdorf, S. Löschke, V. Möller, A. Okem, B. Rama, IPCC, 21497–21503.
2022: Climate Change 2022: Impacts, Adaptation, and Vulnerability. [38] H. Thakkar, S. Eastman, A. Hajari, A.A. Rownaghi, J.C. Knox, F. Rezaei, A.C.S.
Contribution of Working Group II to the Sixth Assessment Report of the Appl, Mater. Interfaces 8 (2016) 27753–27761.
Intergovernmental Panel on Climate Change, Cambridge University Press. In [39] S. Chatterjee, S. Jeevanandham, M. Mukherjee, D.V.N. Vo, V. Mishra, J. Environ.
Press, 2022. Chem. Eng. 9 (2021).
[3] B. Dutcher, M. Fan, A.G. Russell, A.C.S. Appl, Mater. Interfaces 7 (2015) 2137– [40] T.H. Nguyen, S. Kim, M. Yoon, T.H. Bae, ChemSusChem 9 (2016) 455–461.
2148. [41] H. Yang, M. Singh, J. Schaefer, Chem. Commun. 54 (2018) 4915–4918.
[4] A.C. Forse, P.J. Milner, Chem. Sci. 12 (2021) 508–516. [42] H. Mao, J. Tang, J. Xu, Y. Peng, J. Chen, B. Wu, Y. Jiang, K. Hou, S. Chen, J. Wang,
[5] E.S. Sanz-Pérez, C.R. Murdock, S.A. Didas, C.W. Jones, Chem. Rev. 116 (2016) H.R. Lee, D.M. Halat, B. Zhang, W. Chen, A.Z. Plantz, Z. Lu, Y. Cui, J.A. Reimer,
11840–11876. Matter 3 (2020) 2093–2107.
[6] S.Y. Lee, S.J. Park, J. Ind. Eng. Chem. 23 (2015) 1–11. [43] M. Wenzel, M.A. Zaheer, D. Issayeva, D. Poppitz, J. Matysik, R. Gläser, M.
[7] W.L. Theo, J.S. Lim, H. Hashim, A.A. Mustaffa, W.S. Ho, Appl. Energy 183 (2016) Dvoyashkin, J. Phys. Chem. C 125 (2021) 10219–10225.
1633–1663. [44] M. Hunger, Prog. Nucl. Magn. Reson. Spectrosc. 53 (2008) 105–127.
[8] K. Sumida, D.L. Rogow, J.A. Mason, T.M. McDonald, E.D. Bloch, Z.R. Herm, T.H. [45] J.Z. Hu, J.A. Sears, H.S. Mehta, J.J. Ford, J.H. Kwak, K. Zhu, Y. Wang, J. Liu, D.W.
Bae, J.R. Long, Chem. Rev. 112 (2012) 724–781. Hoyt, C.H.F. Peden, Phys. Chem. Chem. Phys. 14 (2012) 2137–2143.
[9] D.M. D’Alessandro, B. Smit, J.R. Long, Angew. Chemie - Int. Ed. 49 (2010) 6058– [46] B. Dinakar, A.C. Forse, H.Z.H. Jiang, Z. Zhu, J.H. Lee, E.J. Kim, S.T. Parker, C.J.
6082. Pollak, R.L. Siegelman, P.J. Milner, J.A. Reimer, J.R. Long, J. Am. Chem. Soc. 143
[10] J. Bin Lin, T.T.T. Nguyen, R. Vaidhyanathan, J. Burner, J.M. Taylor, H. Durekova, (2021) 15258–15270.
F. Akhtar, R.K. Mah, O. Ghaffari-Nik, S. Marx, N. Fylstra, S.S. Iremonger, K.W. [47] M. Aghaie, N. Rezaei, S. Zendehboudi, Renew. Sustain. Energy Rev. 96 (2018)
Dawson, P. Sarkar, P. Hovington, A. Rajendran, T.K. Woo, G.K.H. Shimizu, 502–525.
Science (80-.) 374 (2021) 1464–1469. [48] M. Ramdin, T.W. De Loos, T.J.H. Vlugt, Ind. Eng. Chem. Res. 51 (2012) 8149–
[11] E.J. Kim, R.L. Siegelman, H.Z.H. Jiang, A.C. Forse, J.H. Lee, J.D. Martell, P.J. Milner, 8177.
J.M. Falkowski, J.B. Neaton, J.A. Reimer, S.C. Weston, J.R. Long, Science (80-) 369 [49] M. Sin, N. Kavoosi, M. Rauche, J. Pallmann, S. Paasch, I. Senkovska, S. Kaskel, E.
(2020) 392–396. Brunner, Langmuir 35 (2019) 3162–3170.
[12] T.M. McDonald, J.A. Mason, X. Kong, E.D. Bloch, D. Gygi, A. Dani, V. Crocellà, F. [50] P.V. Kortunov, M. Siskin, L.S. Baugh, D.C. Calabro, Energy Fuel 29 (2015) 5919–
Giordanino, S.O. Odoh, W.S. Drisdell, B. Vlaisavljevich, A.L. Dzubak, R. Poloni, S. 5939.
K. Schnell, N. Planas, K. Lee, T. Pascal, L.F. Wan, D. Prendergast, J.B. Neaton, B. [51] T.M. McDonald, W.R. Lee, J.A. Mason, B.M. Wiers, C.S. Hong, J.R. Long, J. Am.
Smit, J.B. Kortright, L. Gagliardi, S. Bordiga, J.A. Reimer, J.R. Long, Nature 519 Chem. Soc. 134 (2012) 7056–7065.
(2015) 303–308. [52] H. Jo, W.R. Lee, N.W. Kim, H. Jung, K.S. Lim, J.E. Kim, D.W. Kang, H. Lee, V.
[13] A.C. Forse, P.J. Milner, J.H. Lee, H.N. Redfearn, J. Oktawiec, R.L. Siegelman, J.D. Hiremath, J.G. Seo, H. Jin, D. Moon, S.S. Han, C.S. Hong, ChemSusChem 10
Martell, B. Dinakar, L.B. Porter-Zasada, M.I. Gonzalez, J.B. Neaton, J.R. Long, J.A. (2017) 541–550.
Reimer, J. Am. Chem. Soc. 140 (2018) 18016–18031. [53] L. Frydman, J.S. Harwood, J. Am. Chem. Soc. 117 (1995) 5367–5368.
[14] A. H. Berge, S. M. Pugh, M. I. M. Short, Z. Lu, J. Lee, C. J. Pickard and A. C. Forse, [54] A. Medek, L. Marinelli, L. Frydman, A.C.S. Symp, Ser. 717 (1999) 136–155.
DOI:10.26434/chemrxiv-2021-09vcw-v2. [55] R.E. Wasylishen, S.E. Ashbrook, S. Wimperis (Eds.), NMR of Quadrupolar Nuclei
[15] A.C. Forse, M.I. Gonzalez, R.L. Siegelman, V.J. Witherspoon, S. Jawahery, R. in Solid Materials, John Wiley & Sons Ltd, Chichester, 2012.
Mercado, P.J. Milner, J.D. Martell, B. Smit, B. Blümich, J.R. Long, J.A. Reimer, J. [56] S.E. Ashbrook, S. Wimperis, J. Magn. Reson. 156 (2002) 269–281.
Am. Chem. Soc. 140 (2018) 1663–1673. [57] Z. Gan, J. Am. Chem. Soc. 122 (2000) 3242–3243.
[16] W. Wang, W.D. Waang, B.E.G. Lucier, V.V. Terskikh, Y. Huang, J. Phys. Chem. [58] M. Peksa, J. Lang, F. Stallmach, Microporous Mesoporous Mater. 205 (2015)
Lett. 5 (2014) 3360–3365. 11–15.
[17] R.W. Flaig, T.M. Osborn Popp, A.M. Fracaroli, E.A. Kapustin, M.J. Kalmutzki, R.M. [59] M. Peksa, S. Burrekaew, R. Schmid, J. Lang, F. Stallmach, Microporous
Altamimi, F. Fathieh, J.A. Reimer, O.M. Yaghi, J. Am. Chem. Soc. 139 (2017) Mesoporous Mater. 216 (2015) 75–81.
12125–12128. [60] A.C. Forse, K.A. Colwell, M.I. Gonzalez, S. Benders, R.M. Torres-Gavosto, B.
[18] R. Afonso, M. Sardo, L. Mafra, J.R.B. Gomes, Environ. Sci. Technol. 53 (2019) Blümich, J.A. Reimer, J.R. Long, Chem. Mater. 32 (2020) 3570–3576.
2758–2767. [61] C.R. Bowers, H.W. Long, T. Pietrass, H.C. Gaede, A. Pines, Chem. Phys. Lett. 205
[19] D. Shimon, C.H. Chen, J.J. Lee, S.A. Didas, C. Sievers, C.W. Jones, S.E. Hayes, (1993) 168–170.
Environ. Sci. Technol. 52 (2018) 1488–1495. [62] L.C. Lin, J. Kim, X. Kong, E. Scott, T.M. McDonald, J.R. Long, J.A. Reimer, B. Smit,
[20] M.L. Pinto, L. Mafra, J.M. Guil, J. Pires, J. Rocha, Chem. Mater. 23 (2011) 1387– Angew. Chemie - Int. Ed. 52 (2013) 4410–4413.
1395. [63] P.T. Callaghan, Principles of Nuclear Magnetic Resonance Microscopy,
[21] L. Mafra, T. Čendak, S. Schneider, P.V. Wiper, J. Pires, J.R.B. Gomes, M.L. Pinto, J. Claredon Press, Oxford, 1992.
Am. Chem. Soc. 139 (2017) 389–408. [64] A.C. Forse, S.A. Altobelli, S. Benders, M.S. Conradi, J.A. Reimer, J. Phys. Chem. C
[22] A. Sayari, A. Heydari-Gorji, Y. Yang, J. Am. Chem. Soc. 134 (2012) 13834– 122 (2018) 15344–15351.
13842. [65] R.L. Siegelman, P.J. Milner, A.C. Forse, J.H. Lee, K.A. Colwell, J.B. Neaton, J.A.
[23] C.H. Chen, D. Shimon, J.J. Lee, F. Mentink-Vigier, I. Hung, C. Sievers, C.W. Jones, Reimer, S.C. Weston, J.R. Long, J. Am. Chem. Soc. 141 (2019) 13171–13186.
S.E. Hayes, J. Am. Chem. Soc. 140 (2018) 8648–8651. [66] K. Roztocki, M. Rauche, V. Bon, S. Kaskel, E. Brunner, D. Matoga, A.C.S. Appl,
[24] R. Vieira, I. Marin-Montesinos, J. Pereira, R. Fonseca, M. Ilkaeva, M. Sardo, L. Mater. Interfaces 13 (2021) 28503–28513.
Mafra, J. Phys. Chem. C 125 (2021) 14797–14806. [67] H. Liu, O.A. Namjoshi, G.T. Rochelle, Energy Procedia 63 (2014) 1546–1557.
[25] T. Čendak, L. Sequeira, M. Sardo, A. Valente, M.L. Pinto, L. Mafra, Chem. - A Eur. [68] C.S. Srikanth, S.S.C. Chuang, ChemSusChem 5 (2012) 1435–1442.
J. 24 (2018) 10136–10145. [69] J. Marreiros, R. De Oliveira-Silva, P. Iacomi, P.L. Llewellyn, R. Ameloot, D.
[26] C.H. Chen, E.L. Sesti, J.J. Lee, F. Mentink-Vigier, C. Sievers, C.W. Jones, S.E. Hayes, Sakellariou, J. Am. Chem. Soc. 143 (2021) 8249–8254.
J. Phys. Chem. C 125 (2021) 16759–16765. [70] J.H. Du, L. Chen, B. Zhang, K. Chen, M. Wang, Y. Wang, I. Hung, Z. Gan, X.P. Wu,
[27] X. Kong, E. Scott, W. Ding, J.A. Mason, J.R. Long, J.A. Reimer, J. Am. Chem. Soc. X.Q. Gong, L. Peng, Nat. Commun. 13 (2022) 2–7.
134 (2012) 14341–14344. [71] E.G. Keeler, V.K. Michaelis, M.T. Colvin, I. Hung, P.L. Gor’Kov, T.A. Cross, Z. Gan,
[28] R.M. Marti, J.D. Howe, C.R. Morelock, M.S. Conradi, K.S. Walton, D.S. Sholl, S.E. R.G. Griffin, J. Am. Chem. Soc. 139 (2017) 17953–17963.
Hayes, J. Phys. Chem. C 121 (2017) 25778–25787. [72] A. Harchol, G. Reuveni, V. Ri, B. Thomas, R. Carmieli, R.H. Herber, C. Kim, M.
[29] P.G. Boyd, A. Chidambaram, E. García-Díez, C.P. Ireland, T.D. Daff, R. Bounds, A. Leskes, J. Phys. Chem. C 124 (2020) 7082–7090.
Gładysiak, P. Schouwink, S.M. Moosavi, M.M. Maroto-Valer, J.A. Reimer, J.A.R.

11