Identification of the strong Br?nsted acid site in a metal–organic framework solid acid catalyst

Article abstract of DOI:10.1038/s41557-018-0171-z

It remains difficult to understand the surface of solid acid catalysts at the molecular level, despite their importance for industrial catalytic applications. A sulfated zirconium-based metal–organic framework, MOF-808-SO₄, was previously shown to be a strong solid Br?nsted acid material. In this report, we probe the origin of its acidity through an array of spectroscopic, crystallographic and computational characterization techniques. The strongest Br?nsted acid site is shown to consist of a specific arrangement of adsorbed water and sulfate moieties on the zirconium clusters. When a water molecule adsorbs to one zirconium atom, it participates in a hydrogen bond with a sulfate moiety that is chelated to a neighbouring zirconium atom; this motif, in turn, results in the presence of a strongly acidic proton. On dehydration, the material loses its acidity. The hydrated sulfated MOF exhibits a good catalytic performance for the dimerization of isobutene (2-methyl-1-propene), and achieves a 100% selectivity for C8 products with a good conversion efficiency.

Full text of DOI:10.1038/s41557-018-0171-z

Articles
https://doi.org/10.1038/s41557-018-0171-z
Identification of the strong Brønsted acid site in a
metal–organic framework solid acid catalyst
ChristopherA.Trickett1,2,6, ThomasM.Osborn Popp1,2,3,6, JiSu1,2, ChangYanꢀ ꢀ4, JonathanWeisberg1,
AshfiaHuq5, PhilippUrban1,2, JuncongJiang1,2, MarkusJ.Kalmutzki1,2, QingniLiu1,2, JayeonBaek1,2,
MartinP.Head-Gordon1, GaborA.Somorjaiꢀ ꢀ1,2, JeffreyA.Reimer2,3 and OmarM.Yaghiꢀ ꢀ1,2*
It remains difficult to understand the surface of solid acid catalysts at the molecular level, despite their importance for indus-
trial catalytic applications. A sulfated zirconium-based metal–organic framework, MOF-808-SO4, was previously shown to be
a strong solid Brønsted acid material. In this report, we probe the origin of its acidity through an array of spectroscopic, crys-
tallographic and computational characterization techniques. The strongest Brønsted acid site is shown to consist of a specific
arrangement of adsorbed water and sulfate moieties on the zirconium clusters. When a water molecule adsorbs to one zirco-
nium atom, it participates in a hydrogen bond with a sulfate moiety that is chelated to a neighbouring zirconium atom; this
motif, in turn, results in the presence of a strongly acidic proton. On dehydration, the material loses its acidity. The hydrated
sulfated MOF exhibits a good catalytic performance for the dimerization of isobutene (2-methyl-1-propene), and achieves a
100% selectivity for C8 products with a good conversion efficiency.
he chemistry at the surface of solid acid catalysts is of vital (BTC, benzenetricarboxylate), with a subsequent exchange of the
importance for industrial catalytic applications, yet a precise formate ions on the zirconium clusters for sulfate ions simply by
T
molecular picture of these surfaces remains elusive. Attempts washing the MOF in dilute sulfuric acid12,14. The MOF-808 back-
to obtain a clear view of the Brønsted acid sites in solid acids such as bone comprises an octahedron of zirconium atoms that are triply
3
3
sulfated zirconia have resulted in multiple proposed models, in part bridged by μ -O and μ -OH groups. The formate groups in the pris-
due to the difficulty in characterizing the surface structures of these tine structure each bridge two zirconium atoms to form a six-mem-
materials, but also because of wildly variable properties that depend bered belt around the cluster13. One cluster is connected to six other
on the preparation conditions1–11. Discerning the molecular struc- clusters through BTC linkers, three above and three below the belt
tures responsible for the activity of solid acid catalysts provides a of formates, which results in a framework with spn topology. Once
richer perspective on the functional properties and catalytic mecha- the formate ions are exchanged for sulfate to yield MOF-808-SO4
nisms of these materials, and illuminates the fundamental surface (Fig. 1b), these sulfates may take on multiple binding modes and
chemistry that relates the molecular structures and their functions. can take one of several positions along the belt interspersed between
Recently, the synthesis of a metal–organic framework (MOF) solid additional ligated water molecules, which results in long-range dis-
acid catalyst was reported, achieved by treating a Zr-based MOF, order from one cluster to the next. As this disorder is confined to
MOF-808, with sulfuric acid to yield the solid acid MOF, MOF- the surface species on the zirconium clusters, the surface of each
808-SO4, which was shown to be capable of performing several cluster has a slightly different local molecular ‘decoration’ (Fig. 1c),
acid-catalysed reactions12,13. In this report, we conclusively identify whereas the structural backbone of MOF-808 is still conserved
the structure of the strong Brønsted acid site in MOF-808-SO4 as throughout15–18. Our challenge is to understand the molecular deco-
a hydrogen bond pair of two species, water and chelating sulfate, ration of the zirconium clusters in MOF-808-SO4 by first identify-
adsorbed on the surface of its zirconium clusters, where the acidic ing the structures that decorate the cluster surface, and from there,
proton arises as a result of the hydrogen bond. We achieved this discerning which arrangement of decorating structures results in a
through a union of crystallographic, spectroscopic and computa- strong Brønsted acid site.
tional studies. We also show that MOF-808-SO4 exhibits good activ-
ity and selectivity for the dimerization of isobutene to isooctene, Understanding the molecular decoration of the zirconium
and that dehydration of the material significantly reduces the cata- clusters. To elucidate the coordination mode of sulfate is essen-
lytic activity, which confirms the role of water as necessary to the tial to discern the local structures that exist on the surface of the
strong acidity of the site.
clusters. Single-crystal X-ray diffraction analysis of a crystal in
aqueous solution showed that the sulfate groups are coordinated
in both a bridging and chelating mode (Supplementary Fig. 1),
Results and discussion
The preparation of MOF-808-SO4 was performed by first synthesiz- with the bridging mode dominating in a 4:1 ratio over the chelat-
ing pristine MOF-808 (Fig. 1a), Zr6O5(OH)3(BTC)2(HCOO)5(OH2)2 ing one (Supplementary Section 3). To obtain further insight into
1Department of Chemistry, Kavli Energy NanoSciences Institute at Berkeley, and Berkeley Global Science Institute, University of California-Berkeley,
Berkeley, California, USA. 2Materials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA. 3Department of Chemical and
Biomolecular Engineering, University of California, Berkeley, California, USA. 4Department of Chemistry, Stanford University, Stanford, CA, USA.
5Neutron Scattering Division, Oak Ridge National Laboratory, Oak Ridge, TN, USA. 6These authors contributed equally: Christopher A. Trickett,
Thomas M. Osborn Popp *e-mail: yaghi@berkeley.edu
NATURe CHeMISTRY | www.nature.com/naturechemistry

NATure CheMISTry
Articles
cise information on the occupancies and thermal ellipsoids of light
elements within the framework (Fig. 2a). A sample of MOF-808-SO4
with a deuterated BTC linker was measured at 10K and 300K and
refined simultaneously against a structure model, which revealed
a
b
3
3
a 1:1 ratio of μ -O to μ -OH in both independent crystallographic
positions within two standard deviations (Supplementary Section 2).
3
An excess of μ -O is therefore not what balances the excess positive
charge. Substantially more electron density is located around the
position of the oxygen that connects zirconium to sulfur, O6, which
is the same location as the coordinated water molecules bound to
the cluster in the as-synthesized MOF-808-SO4. Note that the sulfate
position could not be located by PND due to the low occupancy and
extremely weak neutron scattering factor of sulfur, and thus infor-
mation from PXRD was used in combination with elemental analysis
to confirm its presence in this sample.
c
As the only electron density unaccounted for in this model is
located at position O6, where water is present in the structure prior
to activation, we can infer that the balance of the excess positive
charge is achieved here by a terminal hydroxide, produced by the
deprotonation of water molecules. This assumption is plausible
given that terminal water molecules bound to zirconium hydroxide
clusters have been found to be acidic19,20. The position thus accounts
for the crystallographically superimposed oxygen from sulfate
groups, hydroxide and water molecules that were not removed dur-
ing the activation process. This overlap excludes the possibility of
determining the precise coordinates of the hydroxide, water and
sulfate oxygen, but the total occupancy of these species was refined
freely, and converged to 78.4 1.1%. This corresponds to 9.4 oxygen
atoms per cluster, out of a possible 12. As there must be 4.6 oxy-
gen atoms from 2.3 bidentate sulfate groups, as found by elemental
analysis, and 1.4 hydroxide groups for charge balancing, this leaves
3.4 0.1 oxygen atoms unaccounted for, and so are assigned to
ligated water. This was confirmed by thermogravimetric analysis–
mass spectrometry on the activated sample, which demonstrated
the loss of 3.1 water molecules per cluster prior to structure decom-
position (Supplementary Fig. 16). The first water signal observed
from the mass spectrometer peaked at 143°C, which indicates water
is still present after evacuation and heating. After a small, second
water-loss event at 236°C, the structure decomposes at around
350°C. This trend can be explained by considering that the loss of
neutral terminal water ligands would not collapse the structure, but
Fig. 1 | MOF-808, MOF-808-SO4 and visualization of differences in
molecular decoration. a, Pristine MOF-808 comprises six connected
zirconium-based metal clusters that contain five formate groups and are
linked by BTC into the depicted spn topology framework. b, These formates
may be substituted with sulfate anions, which coordinate in a bidentate
fashion to zirconium, either in a chelating mode to a single zirconium atom,
or in a bridging mode to two zirconium atoms. Sulfate is predominantly
in the bridging mode in the solvated MOF, and converts exclusively to
the chelating mode after activation by heating under a dynamic vacuum.
c, Two representations of the modelled zirconium clusters, with BTC
linkers omitted beyond the coordinating carboxylate group, highlight the
differences in molecular decoration between different clusters in the overall
structure. A similar stoichiometry of hydroxide, water and sulfate groups
are present on each cluster, but the local arrangement and apportionment
of these groups differ. Zr-based clusters (Zr6O5(OH)3), blue polyhedra;
O, red; C, grey; S, yellow; H, white; pores, large yellow spheres. In a and b,
hydrogen atoms are omitted for clarity. a.u., arbitrary units.
what factors control the coordination mode of these ions, selenated once the framework is completely dehydrated any further mass loss
MOF-808 (MOF-808-SeO4) was synthesized in a similar manner leads to structure decomposition, as this involves the loss of charged
to sulfated MOF-808. The MOF-808-SeO4 framework in aqueous species. Evidence from elemental analysis, PND, 1H NMR spectros-
solution was found to possess only one coordination mode for sel- copy of the digested MOF and thermogravimetric analysis–mass
enate, in which selenate bridges two zirconium atoms, which sug- spectrometry, led to an average molecular formula of Zr6O4(OH)4
gests that perhaps the increased atomic radius of selenium enforces (BTC)2(SO4)2.3(OH)1.4(OH2)3.1(DMF)0.4 (DMF, dimethylformamide)
the bridging coordination mode. However, on activation of these for the activated form of MOF-808-SO4.
two MOFs under a dynamic vacuum and heating at 120°C, both
sulfate and selenate were found to have shifted exclusively into the Identifying the strong Brønsted acid site. With the average chemi-
chelating mode. This was confirmed using a Rietveld refinement cal formula now known, the possible species that decorate each
of the samples measured by powder X-ray diffraction (PXRD) in zirconium cluster are constrained, which simplifies the task of
an argon atmosphere. The solid acid nature of MOF-808-SO4 is identifying the Brønsted acid site in MOF-808-SO4. The potential
only observed after activation at 120°C, which suggests that the acidic sources are discussed in turn. First, terminal hydroxide may
chelating coordination mode of sulfate is a key contributor to its be eliminated simply because terminal water is present and bound
catalytic activity.
to the cluster in the same manner as the hydroxide, with terminal
Quantifying the average molecular formula for MOF-808-SO4 water being known to be more acidic18,19. Protons on the sulfate can
constrains further the possibilities for ligand disorder on the surface also be ruled out because the pH of the solution when the MOF is
of the zirconium clusters. Here, balancing the charge on the zirco- washed with water after incorporation of the sulfate is 3.5, whereas
nium clusters guides our stoichiometric analysis. Using inductively the pKa2 value of sulfuric acid is 1.92 (ref. 21). Therefore, sulfate must
3
coupled plasma–optical emission spectroscopy for elemental analy- be fully deprotonated at this stage. A direct comparison between μ -
sis, 2.3 sulfur atoms per 6 zirconium atoms were found, so an average OH and terminal water is not as straightforward; however, we found
of 2.3 sulfate groups per zirconium cluster. As each zirconium atom that the water molecules bound to the framework could be success-
is in the +4 oxidation state, there is an excess of positive charge that fully removed by holding the temperature at 220°C overnight with
is not properly accounted for within the model so far. To probe this, the crystallinity and porosity maintained. Hereafter, this sample is
we turned to powder neutron diffraction (PND) to obtain more pre- referred to as dehydrated MOF-808-SO4. If the water molecules are,
NATURe CHeMISTRY | www.nature.com/naturechemistry

NATure CheMISTry
Articles
a
b
69 ppm
Observed
Calculated
Difference
(i)
Bragg position
(ii)
0.5
1.0
1.5
2.0
2.5
3.0
3.5
80
60
31P chemical shift (ppm)
40
20
Q space (Å–1
)
Fig. 2 | Structural characterization of MOF-808-SO4 by a Rietveld refinement of powder neutron data, and NMR evidence for the presence of water
being central to the strong acid site. a, The data obtained from PND is compared against the calculated pattern from the structural model and their
difference. R factors in the Rietveld refinement: Rwpꢀ=ꢀ2.91%, Rpꢀ=ꢀ9.59%. b, 31P MAS solid-state NMR spectra of TMPO adsorbed into MOF-808-SO4 (blue)
(i) and dehydrated MOF-808-SO4 (red) (ii). The peak at 69ꢀppm, assigned to TMPO interacting with the strong Brønsted acid site, is lost on dehydration.
The peak centred at 42ꢀppm is due to excess TMPO that does not interact with acid sites directly. Other peaks in the spectra belong to TMPO adsorbed at
1
3
various µ -OH, µ -OH and terminal water sites.
indeed, the most acidic species present, the material should lose its on the terminal water molecules that have no significant interac-
strong acid properties on dehydration.
tions with neighbouring adsorbed molecules. However, when the
To determine if water molecules are the source of the most acidic terminal water molecule was adjacent to a chelating sulfate, there
protons, we adsorbed trimethylphosphine oxide (TMPO) into was a strong hydrogen bond interaction, with an O–H bond length
MOF-808-SO4 as a probe of acidity and performed 31P magic-angle- that ranged from 1.02 to 1.05Å, depending on the particular cluster
spinning (MAS) solid-state NMR spectroscopy. TMPO interacts modelled, significantly longer than the O–H bond with no hydrogen
with Brønsted and Lewis acid sites via the lone pairs on its oxy- bonding. This was accompanied by an O–H…O angle of 163–166°
gen atom. Strong acid sites polarize the phosphorus–oxygen bond, and a short H…O hydrogen bonding distance of 1.50–1.66Å, which
which results in a linear relationship between the 31P chemical shift indicates that the proton was very weakly bound. Indeed, the sys-
values of adsorbed TMPO and the strength of the acid site, where tem can be viewed as a protonated conjugate of an adsorbed pair
a higher 31P chemical shift corresponds to a stronger acid site22–26
.
of hydroxide and sulfate, with the proton sitting between the two
MOF-808-SO4 with adsorbed TMPO showed a 31P resonance at groups, but localized mostly on the hydroxide. One example of this
69ppm associated with a strongly acidic site (Fig. 2b,i), consistent site on a modelled cluster is represented in Fig. 3, which was mod-
with previous observations for this material12. This resonance at elled with overall two water molecules and two chelating sulfate
69ppm was found to be absent when TMPO was used in dehydrated groups located on opposite sides of the zirconium cluster.
MOF-808-SO4 (Fig. 2b,ii). As the loss of a water molecule is associ-
ated with the loss of the strongest acid site, this result supports the acid site implies that the water participating in a hydrogen bond to
role of terminal water as the strongest Brønsted acid source. a chelating sulfate should have distinctly different spectroscopic sig-
The broken symmetry of the water molecule at this proposed
At this point, two key molecular features that decorate the zirco- natures for its two proton environments. We refer to these two sites
nium clusters are now identified as essential to the acidity of MOF- as Ha for the acidic proton on water that participates in the hydrogen
808-SO4: the chelating mode of sulfate and terminal water ligand. In bond to the chelating sulfate, and Hb for the other proton that points
isolation, neither of these two species is sufficient to account for the into free space. To probe these proton chemical environments
acidity of this MOF, and therefore its strong Brønsted acidity must directly, we performed 1H solid-state NMR. Figure 4a shows the 1H
arise from a specific arrangement of these species on the cluster sur- MAS NMR spectrum of MOF-808-SO4 at 6kHz MAS taken before
face. Given the many possible ways to decorate the belt of the cluster and after dehydration, and their difference. The difference spectrum
with terminal water, terminal hydroxide and chelating sulfate, sev- shows that two peaks at around 2.5ppm and 8.1ppm are lost as a
eral arrangements were chosen to be modelled and geometrically result of dehydration. Assigning the identity of these resonances is
1
optimized using density functional theory (DFT). The formula Zr6 informed by comparison with the DFT-calculated H NMR chemi-
O4(OH)4(C2H3O2)6(SO4)2(OH)2(OH2)x was used as a representation cal shifts of two of the modelled zirconium clusters (Supplementary
of an average cluster, where x=2 or 3. The restrictions on the struc- Section 7 and Supplementary Tables 4 and 5). The difference in
tural arrangement of the cluster included (1) the core [Zr6O4(OH)4 chemical shift (Δδ) between the Ha and Hb protons in the acid site
3
(C2H3O2)6]6+ being fixed with the μ -O and –OH groups arranged is calculated for the two cases to be Δδ=5.1 and 9.1ppm, respec-
in the commonly reported alternating arrangement to minimize tively. Water that lacks a strong hydrogen bonding interaction to the
charge repulsion, (2) modelling sulfate as chelating to zirconium chelating sulfate is calculated to have only Δδ=2.0ppm between the
as opposed to bridging, (3) using a terminal hydroxide to charge two protons. The changes in the spectra in Fig. 4a after dehydration
balance the cluster and (4) including 2–3 water molecules per clus- suggest that the two lost resonances belong to the Ha and Hb protons
ter. Additionally, individual clusters were modelled by truncating on the water molecule in the acid site with Δδ=5.6ppm, where Ha,
the linker with acetate groups, which assumes the clusters are elec- the acidic proton, is the downfield resonance.
tronically decoupled. The most enlightening result obtained from
To confirm that these two resonances are the Ha and Hb pro-
the different modelled arrangements on the clusters is from the tons that belong to the same water molecule, a rotor-synchronized
comparison of terminal water in isolation versus one adjacent to a double-quantum (DQ) MAS NMR experiment with a back-to-back
chelating sulfate group. An O–H bond length of 0.98Å was observed recoupling sequence was performed. This experiment correlates
NATURe CHeMISTRY | www.nature.com/naturechemistry

NATure CheMISTry
Articles
a
b
Hb
0.98 Å
166.2°
1.04 Å
Ha
1.55 Å
Fig. 3 | Depiction of the zirconium cluster and Brønsted acid site in MOF-808-SO4 as determined by DFT geometry optimization. a, The overall cluster.
b, A close-up view of the acid site with the relevant bond lengths and angles, and with the acidic proton that participates in the hydrogen bond labelled as
Ha and the other as Hb. Zr, blue; O, red; S, yellow; H, white; atoms not directly part of the active site are in light grey.
a
b
1
2
3
4
5
6
7
8
9
10
11
12
13
15
10
5
0
–5
12 11 10
9
8
7
6
5
4
3
2
1
0 –1 –2 –3
1H chemical shift (ppm)
1H SQ dimension (ppm)
Fig. 4 | Identification of the resonances attributable to adsorbed water using 1H solid-state NMR, comparing MOF-808-SO4 before and after
dehydration. a, The 1H MAS NMR spectrum of MOF-808-SO4 at 6ꢀkHz MAS (blue) and the 1H MAS NMR spectrum of dehydrated MOF-808-SO4 at 6ꢀkHz
MAS (red) show the loss of two prominent peaks assigned to the two inequivalent protons on a water molecule hydrogen-bonded to sulfate. b, 1H DQ-MAS
NMR spectrum of MOF-808-SO4 with SQ and DQ skyline projections (blue). The spectrum was recorded at 12.5ꢀkHz with two cycles of the back-to-back
recoupling sequence for excitation and reconversion of the DQ coherence. The two peaks that are lost on dehydration appear at 2.5 and 8.7ꢀppm and exhibit
a DQ coherence at 11.2ꢀppm, and are assigned as the inequivalent protons on the terminal water hydrogen bonded to a chelating sulfate. The prominent peak
along the autocorrelation diagonal at 5ꢀppm is assigned as a terminal water elsewhere on the zirconium cluster, not adjacent to a sulfate.
proton resonances in the standard single-quantum (SQ) spectrum tons on water molecules that do not neighbour a chelating sulfate
by their proximity to one another through space. A peak in the DQ is similar to the chemical environment of the Hb proton in the acid
dimension indicates that a pair of protons is in close enough prox- site and, accordingly, their chemical shifts should be similar. This is
imity to generate a double quantum coherence27. As the closest pairs supported by our DFT calculations, in which the chemical shifts of
1
of protons in MOF-808-SO4 belong to those on µ -water molecules, protons in these environments were calculated to be within about
we expect these to be the primary coherences observed. The inten-
1ppm of one another. The 1H solid-state NMR results reveal a pic-
sity of these peaks is dependent on the number of duplicate pairs ture consistent with the proposed molecular conformation of the
that exhibit this coherence as well as the efficiency at which this Brønsted acid site, in which water hydrogen bonded to sulfate has
coherence is excited, that is, the internuclear distance28. The SQ Ha two protons with inequivalent O–H bond lengths and inequivalent
and Hb resonances at 8.7 and 2.5ppm, respectively, exhibit strong chemical shifts. The subsequent loss of these peaks after dehydra-
cross peaks at a DQ frequency of 11.2 ppm, which indicates their tion at 220°C is correlated with a loss of acidity, which results in the
close spatial proximity and confirms that these two resonances must conclusion that the strong Brønsted acid site arises from this hydro-
arise from a single water species (Fig. 4b). The low-intensity cross gen-bonding interaction between water and a chelating sulfate.
peaks between 8.7 and 3.1 ppm may arise from a small subset of Ha
and Hb protons in acid sites with a slightly different local arrange- Removal of water at the acid site impacts catalytic performance.
1
3
ment of nearest-neighbour µ -OH and µ -OH groups. Along the These results suggest a structure–property relationship in MOF-
diagonal, a strong autocorrelation DQ peak at around 5.0ppm is 808-SO4, where water must be present and adjacent to a chelating
observed for an SQ resonance at around 2.5ppm, which arises from sulfate to yield a strong acidity. We sought to test this hypothesis
pairs of protons that belong to isolated terminal water at other sites by measuring the activity of MOF-808-SO4 in catalysing the dimer-
on the zirconium cluster. The chemical environment of the pro- ization of isobutene (2-methyl-1-propene), and to see whether
NATURe CHeMISTRY | www.nature.com/naturechemistry

NATure CheMISTry
Articles
a
c
100
80
60
40
20
S-MOF-808
Dehydrated S-MOF-808
Sulfated zirconia
Amberlyst
H-ZSM-5
Solid acid catalyst
+
Helium
0
150
200
50
100
Temperature (°C)
b
d
25
20
15
10
5
25
20
15
10
5
S-MOF-808
S-MOF-808, 80 °C
S-MOF-808, 120 °C
S-MOF-808, 160 °C
Sulfated zirconia, 80 °C
Dehydrated S-MOF-808
Sulfated zirconia
Amberlyst
H-ZSM-5
0
0
50
100
Temperature (°C)
150
200
0
50
100
150
200
250
300
350
Temperature (°C)
Fig. 5 | Comparison of the catalytic conversion, selectivity and long-term stability of MOF-808-SO4 and dehydrated MOF-808-SO4 against benchmark
catalysts. a, General reaction scheme for the dimerization of isobutene (2-methyl-1-propene) to isooctene (2,4,4-trimethyl-1-pentene and 2,4,4-trimethyl-
2-pentene). b, Plot of the percent conversion of isobutene into isooctene for MOF-808-SO4, dehydrated MOF-808-SO4, sulfated zirconia, Amberlyst and
H-ZSM-5 from room temperature up to 200ꢀ°C. Amberlyst is the most active at low temperatures, whereas MOF-808-SO4 has a strong temperature
dependence. Dehydrated MOF-808-SO4 has a significantly lower conversion efficiency, which indicates that the presence of water adjacent to a chelating
sulfate is responsible for the majority of the activity. c, Plot of the selectivity for dimer products over higher-order oligomers. Both Amberlyst and
H-ZSM-5 have poor selectivity and favour higher-order oligomers at all temperatures. MOF-808-SO4, dehydrated MOF-808-SO4 and sulfated zirconia
have a nearly 100% selectivity for dimer products up to 80ꢀ°C. d, Plot of the long-term catalytic performance of MOF-808-SO4 for the dimerization of
isobutene at 80ꢀ°C, 120ꢀ°C and 160ꢀ°C against sulfated zirconia at 80ꢀ°C. The conversion efficiency for MOF-808-SO4 is maintained at 80ꢀ°C, but at higher
temperatures the material loses activity with an increasing rate, probably due to desorption of the terminal water from the clusters at these temperatures.
Sulfated zirconia at 80ꢀ°C has approximately one-third of the activity of MOF-808-SO4 at 80ꢀ°C, but falls to about half this value by 240ꢀh, whereas MOF-
808-SO4 maintains its conversion level throughout this period.
removing the water molecule in the active site by dehydration would for continuous production, and negates the need to purify isooctene
affect this activity. The dimerization of isobutene may yield two prod- from solvent mixtures. Our benchmark materials were chosen based
ucts, either 2,4,4-trimethyl-1-pentene or 2,4,4-trimethyl-2-pentene, on their capacity to operate under these conditions, and their cata-
both referred to as isooctene (Fig. 5a). The terminal alkene product lytic activities were evaluated with respect to the mass of the catalyst.
is prized as a starting material to synthesize terminal aldehydes and MOF-808-SO4 was found to be active even at room temperature,
alcohols, but both alkene products may be hydrogenated to form with conversion peaking at 160°C at 21.5%, which outperformed
2,4,4-trimethylpentane, known as isooctane, a valuable gasoline Amberlyst, sulfated zirconia and H-ZSM-5 under these conditions
octane booster29–31. In the process of dimerizing isobutene, higher- (Fig. 5b). The C8 selectivity of MOF-808-SO4 was 100% at 80°C and
order alkene oligomer products greater than C8 may form, which is lower, yet remained at 92.8% at 160°C, similar to sulfated zirconia
typically disfavoured, as a separation step is required to isolate the (Fig. 5c). The C8 product distribution for both MOF-808-SO4 and
C8 species. Selectivity for C8 products is crucial if isooctane is the sulfated zirconia runs at about 4:1 in favour of the terminal alkene
desired product32,33. To that end, MOF-808-SO4 was benchmarked product (Supplementary Figs. 28–30). H-ZSM-5 and Amberlyst
against other solid acid catalysts for C8 selectivity and conversion exhibit a C8 selectivity under 35% at all temperatures, to form a
efficiency (sulfated zirconia, Amberlyst and H-ZSM-5) using a con- mixture of many different higher-order oligomers. Although the
tinuous gas flow set-up, with isobutene diluted in helium and at C8 selectivity and product distribution for MOF-808-SO4 and sul-
atmospheric pressure (Supplementary Section 11). The advantage of fated zirconia are comparable, under longer experiments of up to
using a gas flow set-up over a solvent-based process is that it allows 15days at 80°C, MOF-808-SO4 does not lose activity or selectivity,
NATURe CHeMISTRY | www.nature.com/naturechemistry

NATure CheMISTry
Articles
maintaining a constant 15% conversion, whereas the conversion effi- References
1. Arata, K. Solid superacids. Adv. Catal. 37, 165 (1990).
ciency of sulfated zirconia drops by around 60% from its starting
value of 5.2% (Fig. 5d). MOF-808-SO4 does begin to lose activity at
120°C and 160°C, and at a faster rate with increasing temperature,
consistent with the notion that the desorption of water from the zir-
conium clusters at higher temperatures should affect the Brønsted
acid site. Indeed, when dehydrated MOF-808-SO4 was tested as a cat-
alyst, the conversion was 80% less than that of MOF-808-SO4 at 80°C
(Fig. 5b). The great majority of the catalytic activity of the material
can thus be attributed to this acid site, where water is adjacent and
hydrogen bonded to the chelating sulfate. The C8 selectivity and
product distribution for dehydrated MOF-808-SO4 are almost iden-
tical to that of MOF-808-SO4 and sulfated zirconia, which indicates
that this acid site alone is not responsible for the selectivity.
We conclude that perturbing the strong Brønsted acid site by
removing the water adjacent to the chelating sulfate has a signifi-
cant negative impact on the catalytic performance of MOF-808-SO4.
The remaining activity of the material in the absence of this water
molecule suggests that Lewis acid sites in the material may also con-
tribute to its activity, but to a lesser extent, a possibility supported
by the observation of open metal sites in the PND refinement of
the structure (Supplementary Section 2). Regeneration of the cata-
lyst thus only requires replacing the water molecule adjacent to the
sulfate, which can be done by repeating the solvent exchange and
activation process. Future work may find a more efficient process
whereby the catalyst is regenerated continuously during the opera-
tion by the addition of water vapour into the product stream, and so
maintain the active site even at higher temperatures.
2. Ward, D. A. & Ko, E. I. One-step synthesis and characterization
of zirconia–sulfate aerogels as solid superacids. J. Catal. 150,
18–33 (1994).
3. Haase, F. & Sauer, J. Te surface structure of sulfated zirconia: periodic
ab initio study of sulfuric acid adsorbed on ZrO2(101) and ZrO2(001).
J. Am. Chem. Soc. 120, 13503–13512 (1998).
4. Bensitel, M., Saur, O., Lavalley, J. C. & Morrow, B. A. An infrared study of
sulfated zirconia. Mater. Chem. Phys. 19, 147–156 (1988).
5. Clearfeld, A., Serrette, G. P. D. & Khazi-Syed, A. H. Nature of hydrous
zirconia and sulfated hydrous zirconia. Catal. Today 20, 295–312 (1994).
6. Kustov, L. M., Kazansky, V. B., Figueras, F. & Tichit, D. Investigation
2−
of the acidic properties of ZrO2 modifed by SO4 anions. J. Catal. 150,
143–149 (1994).
7. Adeeva, V. et al. Acid sites in sulfated and metal-promoted zirconium dioxide
catalysts. J. Catal. 151, 364–372 (1995).
8. Bolis, V., Magnacca, G., Cerrato, G. & Morterra, C. Microcalorimetric
characterization of structural and chemical heterogeneity of superacid SO4/
ZrO2 systems. Langmuir 13, 888–894 (1997).
9. Hino, M., Kurashige, M., Matsuhashi, H. & Arata, K. Te surface structure of
sulfated zirconia: studies of XPS and thermal analysis. Termochim. Acta 441,
35–41 (2006).
10. Arata, K. Organic syntheses catalyzed by superacidic metal oxides: sulfated
zirconia and related compounds. Green Chem. 11, 1719–1728 (2009).
11. Yadav, G. D. & Nair, J. J. Sulfated zirconia and its modifed versions as
promising catalysts for industrial processes. Microporous Mesoporous Mater.
33, 1–48 (1999).
12. Jiang, J. et al. Superacidity in sulfated metal—organic framework-808. J. Am.
Chem. Soc. 136, 12844–12847 (2014).
13. Furukawa, H. et al. Water adsorption in porous metal–organic frameworks
and related materials. J. Am. Chem. Soc. 136, 4369–4381 (2014).
14. Goesten, M. G. et al. Sulfation of metal–organic frameworks: opportunities
for acid catalysis and proton conductivity. J. Catal. 281, 177–187 (2011).
15. Osborn Popp, T. M. & Yaghi, O. M. Sequence-dependent materials. Acc.
Chem. Res. 50, 532–534 (2017).
The relative strength of this acid site compared to that of other
acids is of interest, as its structure may serve as a model for the
design of new strong acid sites. Although MOF-808-SO4 has been
previously stated to be superacidic by colorimetric methods12, these
methods can sometimes prove unreliable for acid sites that exist at
the interface of the solid and gas phase34. We can provide a thermo-
dynamically based estimate of the acidity of this specific Brønsted
acid site based on a previously calculated relationship between
the 31P chemical shift of adsorbed TMPO and the deprotonation
energy of simulated Brønsted acid sites21,22. The TMPO resonance
at 69ppm correlated to the acid site in this material corresponds
to a deprotonation energy of 1,214kJmol–1. It is generally accepted
that a superacid is a medium in which the chemical potential of
the proton is higher than that in sulfuric acid35. The deproton-
ation energy for gas-phase sulfuric acid has been experimentally
determined to be 1,295kJmol–1 (ref. 36), which suggests that this
Brønsted site in MOF-808-SO4 is at the very least comparable to sul-
furic acid, and may even be considered superacidic by this measure.
At its core, the structure of this site in MOF-808-SO4 is characterized
16. Cairns, A. B. & Goodwin, A. L. Structural disorder in molecular framework
materials. Chem. Soc. Rev. 42, 4881–4893 (2013).
17. Furukawa, H., Müller, U. & Yaghi, O. M. 'Heterogeneity within order' in
metal–organic frameworks. Angew. Chem. Int. Ed. 54, 3417–3430 (2015).
18. Trickett, C. A. et al. Defnitive molecular level characterization of defects in
UiO-66 crystals. Angew. Chem. Int. Ed. 54, 11162–11167 (2015).
1
19. Åberg, M. & Glaser, J. 17O and H NMR study of the tetranuclear hydroxo
zirconium complex in aqueous solution. Inorg. Chim. Acta 206, 53–61 (1993).
20. Springborg, J. Hydroxo-bridged complexes of chromium (iii), cobalt (iii),
rhodium (iii), and iridium (iii). Adv. Inorg. Chem. 32, 55–169 (1988).
21. Hall, J. Lab Manual for Zumdahl/Zumdahl’s Chemistry 656 (Brooks Cole,
Pacifc Grove, 2002).
22. Zheng, A., Zhang, H., Lu, X., Liu, S. B. & Deng, F. Teoretical predictions of
31P NMR chemical shif threshold of trimethylphosphine oxide absorbed on
solid acid catalysts. J. Phys. Chem. B 112, 4496–4505 (2008).
23. Zheng, A. et al. 31P chemical shif of adsorbed trialkylphosphine oxides
for acidity characterization of solid acids catalysts. J. Phys. Chem. 112,
7349–7356 (2008).
24. Zheng, A., Huang, S. J., Liu, S. B. & Deng, F. Acid properties of solid acid
catalysts characterized by solid-state 31P NMR of adsorbed phosphorous
probe molecules. Phys. Chem. Chem. Phys. 13, 14889 (2011).
1
by the pairing of two bases (chelating sulfate and μ -OH) supported
25. Chen, W. H. et al. A solid-state NMR, FT-IR and TPD study on acid
properties of sulfated and metal-promoted zirconia: infuence of promoter
and sulfation treatment. Catal. Today 116, 111–120 (2006).
on two neighbouring zirconium atoms and sharing a weakly bound
proton between them. Thus, such a Brønsted acid site construct may
be quite generalizable, as it may be possible to reduce the deproton-
ation energy of this proton to yield even stronger acidity by manipu-
lating the identities of these two bases or of the support atoms.
26. Lunsford, J. H., Sang, H., Campbell, S. M., Liang, C. H. & Anthony, R. G. An
NMR study of acid sites on sulfated-zirconia catalysts using
trimethylphosphine as a probe. Catal. Lett. 27, 305–314 (1994).
27. Gottwald, J., Demco, D. E., Graf, R. & Spiess, H. W. High-resolution
double-quantum NMR spectroscopy of homonuclear spin pairs and proton
connectivities in solids. Chem. Phys. Lett. 243, 314–323 (1995).
28. Schnell, I., Brown, S. P., Low, H. Y., Ishida, H. & Spiess, H. W. An
investigation of hydrogen bonding in benzoxazine dimers by fast magic-angle
Data availability
Synthetic and experimental procedures, as well as crystallographic, spectroscopic
and computational data are provided in the Supplementary Information.
Crystallographic data for the structures reported in this Article have been
deposited at the Cambridge Crystallographic Data Centre, under deposition
numbers CCDC 1871192 (MOF-808), 1871193 (MOF-808-SO4) and 1871194
(MOF-808-SeO4). Copies of the data can be obtained free of charge via www.ccdc.
cam.ac.uk/data_request/cif. All other data supporting the findings of this study
are available within the article and its Supplementary Information, or from the
corresponding author upon reasonable request.
1
spinning and double-quantum H NMR spectroscopy. J. Am. Chem. Soc. 120,
11784–11795 (1998).
29. Mahdi, H. I. & Muraza, O. Conversion of isobutylene to octane-booster
compounds afer methyl tert-butyl ether phaseout: the role of heterogeneous
catalysis. Ind. Eng. Chem. Res. 55, 11193–11210 (2016).
30. Takahashi, K., Yamashita, M. & Nozaki, K. Tandem hydroformylation/
hydrogenation of alkenes to normal alcohols using Rh/Ru dual catalyst or Ru
single component catalyst. J. Am. Chem. Soc. 134, 18746–18757 (2012).
31. Behr, A. Ullman’s Encyclopedia of Industrial Chemistry 223–269 (Wiley,
Hoboken, 2010).
Received: 13 December 2017; Accepted: 12 October 2018;
Published: xx xx xxxx
NATURe CHeMISTRY | www.nature.com/naturechemistry

NATure CheMISTry
Articles
32. Izquierdo, J. F., Vila, M., Tejero, J., Cunill, F. & Iborra, M. Kinetic study of
isobutene dimerization catalyzed by a macroporous sulphonic acid resin.
Appl. Catal. A 106, 155–165 (1993).
through the German Research Foundation (DFG, KA 4484/1-1). We acknowledge
B. Rungtaweevoranit for his assistance with electron microscopy, and S. Teat and
L. McCormick for the synchrotron X-ray diffraction data acquisition support at beamline
11.3.1 of the Advanced Light Source, Lawrence Berkeley National Laboratory.
33. Kamath, R. S., Qi, Z., Sundmacher, K., Aghalayam, P. & Mahajani, S. M.
Process analysis for dimerization of isobutene by reactive distillation. Ind.
Eng. Chem. Res. 45, 1575–1582 (2006).
Author contributions
34. Song, X. & Sayari, A. Sulfated zirconia-based strong solid-acid catalysts:
recent progress. Cat. Rev. 38, 329–412 (1996).
C.A.T. and T.M.O.P. co-wrote the manuscript. C.A.T. performed the PND
modelling, single-crystal X-ray diffraction and PXRD experiments. T.M.O.P.
performed the solid-state NMR experiments and NMR DFT calculations, with
support and advice from J.A.R. J.S., Q.L. and J.B. performed the dimerization
catalysis experiments with the support and advice of G.S. C.Y. performed the
infrared experiments. J.W. performed the DFT calculations on the cluster models,
with support and advice from M.P.H.-G. A.H. performed the PND experiments.
P.U. performed the PXRD Rietveld refinements. M.J.K. helped with the
thermogravimetric analysis experiments. J.J. supported and advised the synthesis. O.M.Y.
supervised the project. All the authors reviewed and edited the manuscript
and contributed to useful discussions.
35. Himmel, D., Goll, S. K., Leito, I. & Krossing, I. A unifed pH scale for all
phases. Angew. Chem. Int. Ed. 49, 6885–6888 (2010).
36. Viggiano, A. A., Henchman, M. J., Dale, F., Deakyne, C. A. & Paulson, J. F.
Gas-phase reactions of weak Brønsted bases I–, PO3 , HSO4 , FSO3 , and
–
–
–
–
CF3SO3 with strong Brønsted acids H2SO4, FSO3H, and CF3SO3H, a
quantitative intrinsic superacidity scale for the sulfonic acids XSO3H (X=HO,
F, and CF3). J. Am. Chem. Soc. 114, 4299–4306 (1992).
Acknowledgements
This work, including the synthesis, characterization and crystal structure analysis, was
funded by BASF SE and the US Department of Defense, Defense Threat Reduction
Agency (HDTRA 1-12-1-0053). Work performed at the Advanced Light Source is
supported by the Director, Office of Science, Office of Basic Energy Sciences, of the US
Department of Energy under Contract no. DE-AC02-05CH11231. A portion of this
research used resources at the Spallation Neutron Source, a DOE Office of Science User
Facility operated by the Oak Ridge National Laboratory. NMR work was supported
as part of the Center for Gas Separations Relevant to Clean Energy Technologies, an
Energy Frontier Research Center funded by the US Department of Energy, Office of
Science, Basic Energy Sciences under Award no. DE-SC0001015. T.M.O.P. acknowledges
funding from the NSF Graduate Research Fellowship Program. C.Y. acknowledges
support from a Hewlett-Packard Stanford Graduate Fellowship. P.U. acknowledges the
German Research Foundation (DFG, PU 286/1-1). M.J.K. is grateful for financial support
Competing interests
The authors declare no competing interests.
Additional information
Supplementary information is available for this paper at https://doi.org/10.1038/
s41557-018-0171-z.
Reprints and permissions information is available at www.nature.com/reprints.
Correspondence and requests for materials should be addressed to O.M.Y.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
© The Author(s), under exclusive licence to Springer Nature Limited 2018
NATURe CHeMISTRY | www.nature.com/naturechemistry