REPORTS

produce analogs of 2 differing in the alcohol 9. K. Zechmeister et al., Tetrahedron Lett. 11, 4075–4078 (1970). 31. K. Chen, P. S. Baran, Nature 459, 824–828 (2009).
functionalities (C-3, C-4, C-5, and C-20) to better 10. R. W. Alder, S. P. East, Chem. Rev. 96, 2097–2112 (1996). 32. A. Nakamura, M. Nakada, Synthesis 45, 1421–1451 (2013).
11. S. M. Ogbourne et al., Cancer Res. 64, 2833–2839 (2004). 33. T. Konegawa, Y. Ohtsuka, H. Ikeda, T. Sugai, H. Ohta,
understand the structure-activity relationship of 12. M. Fujiwara et al., Antimicrob. Agents Chemother. 40, Synlett 1997, 1297–1299 (1997).
ingenanes. 271–273 (1996). 34. K. M. Brummond et al., Org. Lett. 4, 1931–1934 (2002).
The synthesis of ingenol (1) presented herein 13. A. Vasas, D. Rédei, D. Csupor, J. Molnár, J. Hohmann, 35. K. Shibuya, Synth. Commun. 24, 2923–2941 (1994).
illustrates the power of two-phase logic to deliver Eur. J. Org. Chem. 2012, 5115–5130 (2012). 36. P. A. Wender, V. A. Verma, T. J. Paxton, T. H. Pillow,
14. C. M. Hasler, G. Acs, P. M. Blumberg, Cancer Res. 52, Acc. Chem. Res. 41, 40–49 (2008).
an efficient, concise synthesis even in architec- 202–208 (1992). 37. N. Z. Burns, P. S. Baran, R. W. Hoffmann, Angew. Chem.
turally complex settings. The usefulness of the 15. U.S. Food and Drug Administration, 2012 Novel New Int. Ed. 48, 2854–2867 (2009).
two-phase approach will undoubtedly continue Drugs Summary (FDA, Washington, DC, 2013); www.fda.
to expand as new methods for C–C bond for- gov/downloads/Drugs/DevelopmentApprovalProcess/ Acknowledgments: Financial support for this work was
DrugInnovation/UCM337830.pdf. provided by LEO Pharma. The Carlsberg Foundation and the
mation and C–H oxidation are developed. Fur- 16. G. Siller et al., Australas. J. Dermatol. 51, 99–105 (2010). Danish Council for Independent Research (Technology and
thermore, this report provides a strong rebuttal 17. “European public assessment report: Picato” [European Production Sciences) provided postdoctoral fellowships for
to the presumption that chemical synthesis is ill- Medicines Agency (EMA) publication 650464, London, L.J., and the Alexander von Humboldt Foundation provided
equipped to deal with the preparation of struc- 2012]; www.ema.europa.eu/docs/en_GB/document_ a postdoctoral fellowship for C.A.K. We are grateful to
library/EPAR_-_Public_assessment_report/human/ A. Rheingold and C. E. Moore (University of California, San
turally complex terpenoid drug molecules. Rather,
002275/WC500135329.pdf. Diego) for x-ray crystallographic analysis and G. Siuzdak
in this instance, total chemical synthesis holds 18. X. Liang, G. Grue-Sørensen, A. K. Petersen, T. Högberg, (the Scripps Research Institute) for mass spectrometry
promise as the best method to both prepare ingenol Synlett 23, 2647–2652 (2012). assistance. We thank H. Bladh (LEO Pharma) for valuable
mebutate (2) and enable the development of 19. J. Hohmann, F. Evanics, L. Berta, T. Bartók, Planta Med. discussions and LEO Pharma for generous donations of
therapeutic analogs with broader utility in the 66, 291–294 (2000). natural 20-deoxyingenol and ingenol. Metrical parameters
20. G. Appendino, G. C. Tron, G. Cravotto, G. Palmisano, for the structures of compounds 14, 20, and 21 are available
treatment of human diseases. J. Jakupovic, J. Nat. Prod. 62, 76–79 (1999). free of charge from the Cambridge Crystallographic Data
Centre under reference numbers CCDC-943074, 943076,

Downloaded from http://science.sciencemag.org/ on February 10, 2018

21. V. Hale, J. D. Keasling, N. Renninger, T. T. Diagana,
References and Notes Am. J. Trop. Med. Hyg. 77 (suppl.), 198–202 (2007). and 943075, respectively. A provisional patent application
1. F. E. Koehn, G. T. Carter, Nat. Rev. Drug Discov. 4, 22. R. J. Schmidt, J. Linn. Soc. Bot. 94, 221–230 (1987). has been filed and is available under patent application
206–220 (2005). 23. J. Kirby et al., Phytochemistry 71, 1466–1473 (2010). no. U.S. 61/829,861.
2. I. Raskin et al., Trends Biotechnol. 20, 522–531 (2002). 24. I. Kuwajima, K. Tanino, Chem. Rev. 105, 4661–4670 (2005).
3. P. G. Mountford, in Green Chemistry in the 25. J. D. Winkler, M. B. Rouse, M. F. Greaney, S. J. Harrison, Supplementary Materials
Pharmaceutical Industry, P. J. Dunn, A. S. Wells, Y. T. Jeon, J. Am. Chem. Soc. 124, 9726–9728 (2002). www.sciencemag.org/cgi/content/full/science.1241606/DC1
M. T. Williams, Eds. (Wiley, Weinheim, Germany, 2010), Materials and Methods
26. K. Tanino et al., J. Am. Chem. Soc. 125, 1498–1500 (2003).
pp. 145–160. Supplementary Text
27. A. Nickel et al., J. Am. Chem. Soc. 126, 16300–16301
4. C. J. Paddon et al., Nature 496, 528–532 (2013). Figs. S1 to S4
5. C. Zhu, S. P. Cook, J. Am. Chem. Soc. 134, 13577–13579 (2004).
28. K. Watanabe et al., J. Org. Chem. 69, 7802–7808 (2004). Tables S1 to S7
(2012). Data
6. J. D. Keasling, A. Mendoza, P. S. Baran, Nature 492, 29. O. L. Epstein, J. K. Cha, Angew. Chem. Int. Ed. 44,
188–189 (2012). 121–123 (2004). 7 June 2013; accepted 18 July 2013
7. J. D. Keasling, Science 330, 1355–1358 (2010). 30. G. Appendino et al., Eur. J. Org. Chem. 1999, Published online 1 August 2013;
8. E. Hecker, Cancer Res. 28, 2338–2349 (1968). 3413–3420 (1999). 10.1126/science.1241606

neutron, and electron) and remains unexamined

Mapping of Functional Groups in with other characterization methods. We show that
multidimensional solid-state nuclear magnetic res-
Metal-Organic Frameworks onance (SSNMR) combined with molecular simu-
lations can be used to generate three-dimensional
Xueqian Kong,1,2* Hexiang Deng,3,4*† Fangyong Yan,1,4* Jihan Kim,4‡ Joseph A. Swisher,1,4 (3D) maps of the apportionment of functional groups
Berend Smit,1,3,4 Omar M. Yaghi,3,4,5§ Jeffrey A. Reimer1,2§ (Fig. 1) within and between the pores of MTV-
MOF-5. These MTV-MOFs adopt MOF-5 structure
We determined the heterogeneous mesoscale spatial apportionment of functional groups in a series containing BDC (1,4 benzenedicarboxylate) linkers
of multivariate metal-organic frameworks (MTV-MOF-5) containing BDC (1,4-benzenedicarboxylate) with different functional groups–B (BDC-NH2), E
linkers with different functional groups—B (BDC-NH2), E (BDC-NO2), F [(BDC-(CH3)2], H [BDC-(OC3H5)2], (BDC-NO2), F [(BDC-(CH3)2], H [BDC-(OC3H5)2],
and I [BDC-(OC7H7)2]—using solid-state nuclear magnetic resonance measurements combined and I [BDC-(OC7H7)2]. This method can be ap-
with molecular simulations. Our analysis reveals that these methods discern between random (EF), plied to other ordered or disordered systems to
alternating (EI and EHI), and various cluster (BF) forms of functional group apportionments.
This combined synthetic, characterization, and computational approach predicts the adsorptive
1
properties of crystalline MTV-MOF systems. This methodology, developed in the context of Department of Chemical and Biomolecular Engineering, Uni-
ordered frameworks, is a first step in resolving the more general problem of spatial disorder in versity of California, Berkeley, CA 94720, USA. 2Environmental
Energy Technologies Division, Lawrence Berkeley National Lab-
other ordered materials, including mesoporous materials, functionalized polymers, and defect oratory, Berkeley, CA 94720, USA. 3Department of Chemistry,
distributions within crystalline solids. University of California, Berkeley, CA 94720, USA. 4Materials
Sciences Division, Lawrence Berkeley National Laboratory,
strategy for optimizing the properties of interior (2). This heterogeneity can enhanced se- Berkeley, CA 94720, USA. 5NanoCentury KAIST Institute and

A synthetic crystalline materials is to delib-

erately introduce heterogeneity by increas-
ing the number of constituents (1). This approach
lectivity for carbon dioxide (CO2) capture and
hydrogen uptake relative to mixtures of pure ma-
terials. On a fundamental level, these materials
Graduate School of Energy, Environment, Water, and Sustain-
ability (World Class University), Daejeon 305-701, Republic of
Korea.
*These authors contributed equally to this work.
can be compromised by the dual challenges of present characterization challenges, in that there †Present address: College of Chemistry and Molecular Sci-
phase separation, lack of order, or both. Multi- are no experimental techniques that can elucidate ences, Wuhan University Luojiashan, Wuhan, 430072 China.
variate metal-organic frameworks (MTV-MOFs) the intermingling of functional groups that char- ‡Present address: Department of Chemical and Biomolecular
Engineering, Korea Advanced Institute of Science and Tech-
can meet this challenge by incorporating multiple acterizes the heterogeneity within the crystalline nology, 291 Daehak-ro Yuseong-gu, Daejeon, 305-701 Korea.
linkers that bear different functional groups into MTV-MOFs. For example, this “heterogeneity prob- §Corresponding author. E-mail: [email protected] (O.M.Y.);
the same crystal, which creates a heterogeneous lem” is intractable for diffraction methods (x-ray, [email protected] ( J.A.R.)

882 23 AUGUST 2013 VOL 341 SCIENCE www.sciencemag.org

 REPORTS
map the distributions of functional groups, ligands, nature of apportionments. The matching of these considers the averaging of point-to-point distances
particles, or defects. NMR-derived average distances with Monte Carlo over the entire sample.
Our analysis indicates that, depending on the and molecular dynamics simulations allows us In the REDOR method, a period of multinu-
combination of various kinds of functionality and to uncover these various apportionment scenar- clear irradiation is incrementally increased so that
their ratios, three distinct apportionments of those ios. We invoked two separate strategies for dis- the pairwise coupling between NMR-active iso-
functionalities exist: random, alternating, and clus- tance measurements in the MTV-MOF system: topes evolves to yield “decay curves,”. These curves
tered. Upon discerning these apportionments, we 2D correlation experiments (3, 4) and rotational- are governed by the dipolar coupling constant be-
can then predict and explain (i) the CO2 adsorption echo double-resonance (REDOR) NMR (5–9). tween two spins (i and j), which is proportional to
isotherm and the observed differences in selectivity, The former method enjoys broad applicability in 1/rij3, where rij is their internuclear distance (11).
(ii) the changes in apportionment of functionalities polymer science (4), yet was inconclusive in the We selectively labeled the spins at different linkers—
as a function of composition and functionality ratio, present study (10). The latter method proved to for example, 13C on linker X and 15N on linker Y—
and (iii) the apportionment in ternary functionality be decisive. to probe the intermolecular separation of these
mixtures based on their binary NMR data. The REDOR NMR method provides a quan- linkers. If the linkers are close, the average coupling
Different interactions between linkers can lead titative measure of distances between atoms. A is large and the decay is rapid; that is, a fast decay
to different apportionment scenarios. Linkers with pairwise measurement of the distance between means that most of the linkers are in close prox-
preferential interactions with each other form clus- two spectroscopically resolved nuclear spins and, imity. Figure 2 provides a sketch of how these decay
ters, whereas linkers with preferential interactions hence, the corresponding atoms might seem an curves depend on the apportionment of function-
with different linkers lead to alternating struc- unlikely way to assess the apportionment of the alities for the different scenarios possible in MTV-
tures. Linkers with no preferences give a random functionalities of linkers because of the large num- MOF-5-BF. Indeed, faster decay is expected for
structure. The SSNMR distance measurements ber of possible linker configurations. The key to the scenarios in the order of alternating, random,
represent an average of pairwise interactions through- recovering apportionment of the functional groups small, and large clusters (Fig. 2B).

Downloaded from http://science.sciencemag.org/ on February 10, 2018

out the entire sample and thus reflect the average from this measurement, however, is in how one To measure the REDOR curves, we 15N la-
beled, in the case of MTV-MOF-5-BF, each amine
group of linker B and measured its coupling to
13
C spins of the methyl groups [19 parts per
million (ppm)]. Figure 2C shows the 13C spectra
measured in a 13C{15N} REDOR experiment at a
given recoupling time with and without 15N di-
polar modulation, S and S0, respectively. The S/S0
ratio for the resonance at 19 ppm is plotted against
a series of recoupling times to give S/S0 decay data
and shown in Fig. 3A. Similar data were obtained
for MTV-MOF-5-EF, EI, and EHI and are over-
laid on the predicted REDOR curves (Fig. 3A).
The apportionment can be derived from a
model if we assume that linkers with different
functional groups are predisposed to have differ-
ent neighbor preferences and that we can express
these preferences as effective interaction energies.
A Monte Carlo simulation (12) is then conducted
Fig. 1. MTV-MOF-5 linkers. Six linkers are combined with a Zn4O “secondary building unit” to form an whereby linkers between metal sites are randomly
MTV-MOF-5. We use the linkers B, E, F, H, and I. Each linker has the same length but differs by their exchanged in a simulated lattice structure; a new
functional groups so that they can be interchanged to form a material with the same crystal lattice but apportionment is accepted according to the energy
different apportionment of the linkers. The amino and nitro groups on linker B and E, respectively, were difference between the new and old apportionment.
100% isotopically enriched with 15N, whereas all other nuclei are present at natural isotopic abundance. In this way, the effective interaction energies

Fig. 2. Solid-state NMR results for MTV-MOF-5-BF. (A) The arrow rep- REDOR experiment at a given recoupling time. Blue, the 13C spectrum taken
resents the specific distance measured by our REDOR experiments. (B) without 15N dipolar modulation, S0; pink, the 13C spectrum with 15N dipolar
Molecular dynamics simulations of the 13C{15N} REDOR S/S0 decay curves for modulation, S; red, the difference spectrum D S obtained by subtracting S from
different BF apportionments. (C) A suite of 13C spectra obtained from a 13C{15N} S0. *, Spinning side bands.

www.sciencemag.org SCIENCE VOL 341 23 AUGUST 2013 883

REPORTS
determine the apportionment in a lattice model for
which many configurations are generated using
conventional Monte Carlo methods. The differ-
ent apportionments lead to the distinct REDOR
curves because these scenarios result in different
average distances between atoms on different func-
tional groups. In an alternating system, the majority
of 13C (on linker X)–15N (on linker Y) are paired
next to each other, and we observed fast decay. In
a clustered system, the number of neighboring
C- N pairs is far less, and the REDOR S/S0
13 15

decay is relatively flat. A quantitative measure of

the REDOR curves was obtained by taking the
apportionments of the functional groups from
our lattice simulations and replacing the lattice
model by the corresponding all-atom model. The
subsequent molecular dynamics optimization pro-
vided the ensemble average of 1/r3 for all labeled
pairs from which we computed the REDOR curves.
Both the experiments and the simulation show
that the decay was not sensitive to temperature.

Downloaded from http://science.sciencemag.org/ on February 10, 2018

Figure 3B shows that for different materials, we
could obtain a quantitative match with experimen-
tal data for the different apportionment scenarios,
which yield materials that are visually quite dif-
ferent. For example, the EF linkers had a random
distribution, the EI linkers were alternating, and
clusters were formed for the BF linkers. To ratio-
nalize these results, we carried out density func-
tional theory (DFT) calculations (10) to address
the effect of functional group identity on the ener-
gy of formation of the metal-linker complex (table
S3). Because of the electronegative -NO2 group,
the E linker has the least favorable energy of for-
mation. This accounts for why we have been un-
able to synthesize a pure E MTV-MOF and why
binary and ternary MTV-MOFs have a lower linker
E content compared with their synthesis solution
(table S1). The implication of these DFT calcula-
tions for apportionment is that the structural build-
ing units will minimize the number of E linkers,
resulting in an apportionment biased toward an
alternating structure. The B linker differentiates
itself from others in that this linker can form hy-
drogen bonds both with other B linkers and with
the synthesis solvent (13). One therefore would
anticipate clustering in solution as an antecedent
for the formation of clusters in the MTV-MOF.
With these apportionments, we can predict the
adsorption properties of these materials. Figure 4 Fig. 3. REDOR results and apportionments. (A) Experimental 13C{15N} REDOR curves (squares) are
shows that the experimental isotherms of MTV- compared with the molecular dynamics simulations for the different linker apportionments, large clusters
MOF-5-EI and EHI agree well with the ones de- (LC), small clusters (SC), random (Ran), and alternating (Alt). (B) The derived apportionments for the
rived from our best-fit models to the REDOR data different MTV-MOF-5s. Yellow, linker B; blue, linker F; red, linker E; purple, linker I; green, linker H.
and that alternative apportionments of large clus-
ters fail to describe the adsorption isotherms. We
can also use this result to rationalize the surprising for the fourfold enhancement of the selectivity of compositions that were not yet synthesized
enhancement of the selectivity for CO2 over carbon observed previously but unexplained (2). (fig. S17). In addition, we predicted the effect of
monoxide (CO) observed in the MTV-MOF-5-EHI Until now, we have used the effective inter- changes in the linker composition and the ternary
material compared with the single-linker MOF-5 action energies as fitting parameters for the NMR linker apportionment on the basis of only binary
material (2). The MOF-5 with E linkers lacks se- REDOR curves. We surmise that these effective interaction parameters (tables S6 and S7).
lectivity, while the ones with I linkers are too bulky energies describe the underlying interactions of Our strategy to resolve apportionment in a
and therefore block the pores. The combination of the linkers during the formation of the MOF. We disordered system consists of three parts: the syn-
the two in an alternating functionality apportionment, can use this model to predict the apportionment. thesis of systems with disorder distributed upon
however, creates the highly selective adsorption For example, we successfully predicted the ap- an ordered lattice (in the present study MTV-
sites and sufficient pore space that are responsible portionments and corresponding REDOR curves MOFs); experimental measurement of pairwise

884 23 AUGUST 2013 VOL 341 SCIENCE www.sciencemag.org

 REPORTS
6. S. D. Cady et al., Nature 463, 689–692 (2010).
7. L. Cegelski et al., J. Am. Chem. Soc. 132, 16052–16057
(2010).
8. C. P. Jaroniec, B. A. Tounge, C. M. Rienstra, J. Herzfeld,
R. G. Griffin, J. Magn. Reson. 146, 132–139 (2000).
9. E. A. Louie, P. Chirakul, V. Raghunathan, S. T. Sigurdsson,
G. P. Drobny, J. Magn. Reson. 178, 11–24 (2006).
10. Materials and methods are available as supplementary
materials on Science Online.
11. K. T. Mueller, J. Magn. Reson. A 113, 81–93 (1995).
12. D. Frenkel, B. Smit, Understanding Molecular Simulation
from Algorithms to Applications (Academic Press,
San Diego, CA, 2002).
13. S. Loos, W. Seichter, E. Weber, F. Mertens, Acta
Crystallogr. Sect. E Struct. Rep. Online 68, o2501
(2012).

Acknowledgments: This research was supported through

the Center for Gas Separations Relevant to Clean Energy
Technologies, an Energy Frontier Research Center funded by
Fig. 4. CO2 adsorption isotherms. (A) MTV-MOF-5-EI and (B) MTV-MOF-5-EHI. Experimental (blue) the U.S. Department of Energy, Office of Science, Office of
and simulated (red) at 298 K. Solid circles correspond to simulated isotherms for those linker ap- Basic Energy Sciences under award DE-SC0001015, and the
synthesis effort (H.D.) was partially supported by BASF SE
portionments that best fit the REDOR experiments. For reference, we have added simulation results for a
(Ludwigshafen, Germany). For all computations, we used the
large-cluster apportionment. resources of the National Energy Research Scientific
Computing Center (NERSC). We thank W. Morris, C. Stevens,

Downloaded from http://science.sciencemag.org/ on February 10, 2018

and H. Furukawa for their invaluable assistance.
distances (here via NMR REDOR); and a com- References and Notes
putational method that allows us to directly sim- 1. J. Hemminger et al., From Quanta to the Continuum:
Opportunities for Mesoscale Science (U.S. Department of Supplementary Materials
ulate the disorder. This strategy may be employed Energy, Washington, DC, 2012). www.sciencemag.org/cgi/content/full/science.1238339/DC1
for many issues in materials engineering, such as 2. H. Deng et al., Science 327, 846–850 (2010). Materials and Methods
defect distributions in crystals and functional group 3. J. Jeener, B. H. Meier, P. Bachmann, R. R. Ernst, Supplementary Text
J. Chem. Phys. 71, 4546–4553 (1979). Figs. S1 to S31
distributions in block copolymers. It is our cur- Tables S1 to S12
rent vision to design and synthesize materials with 4. K. Schmidt-Rohr, H. W. Spiess, Multidimensional
References (14–42)
Solid-State NMR and Polymers (Academic Press,
crafted heterogeneity so as to carry out complex San Diego, CA, 1994). 26 March 2013; accepted 3 July 2013
molecular operations such as recognizing, sorting, 5. T. Gullion, J. Schaefer, J. Magn. Reson. 81, 196–200 Published online 25 July 2013;
and counting substrates. (1989). 10.1126/science.1238339

ture, recent observations of long-lived nanometer-

Microscopic Evidence for sized clusters present before nucleation have led
to a predominant view of CaCO3 formation in
Liquid-Liquid Separation in which the nucleation pathway is “nonclassical”
(2, 3, 5–8), involving prenucleation clusters that
Supersaturated CaCO3 Solutions are stable (or metastable) with respect to both dis-
solution and growth. In this model, bulk CaCO3
forms primarily through aggregation of these
Adam F. Wallace,1,2,3* Lester O. Hedges,1,2 Alejandro Fernandez-Martinez,1,4 clusters.
Paolo Raiteri,5 Julian D. Gale,5 Glenn A. Waychunas,1 Stephen Whitelam,2 An alternative route to mineralization, liquid-
Jillian F. Banfield,1,6 James J. De Yoreo2,7,8* liquid separation, has also been proposed based on
light scattering, electron microscopy, and nuclear
Recent experimental observations of the onset of calcium carbonate (CaCO3) mineralization suggest magnetic resonance (NMR) (9–11). This hypothe-
the emergence of a population of clusters that are stable rather than unstable as predicted by classical sis is supported by experimental work document-
nucleation theory. This study uses molecular dynamics simulations to probe the structure, dynamics, and ing the formation of liquid magnesium sulfate
energetics of hydrated CaCO3 clusters and lattice gas simulations to explore the behavior of cluster (MgSO4) at high temperature (12), as well as the
populations before nucleation. Our results predict formation of a dense liquid phase through liquid-liquid known persistence of a polymer-stabilized liquid
separation within the concentration range in which clusters are observed. Coalescence and solidification
of nanoscale droplets results in formation of a solid phase, the structure of which is consistent with 1
Earth Sciences Division, Lawrence Berkeley National Laboratory,
amorphous CaCO3. The presence of a liquid-liquid binodal enables a diverse set of experimental Berkeley, CA 94720, USA. 2The Molecular Foundry, Lawrence
observations to be reconciled within the context of established phase-separation mechanisms. Berkeley National Laboratory, Berkeley, CA 94720, USA. 3De-
partment of Geological Sciences, University of Delaware, Newark,
DE 19716, USA. 4Institut des Sciences de la Terre (ISTerre), Univer-
alcium carbonate (CaCO3) has been in- classical expectations. In the classical picture, nu-

C tensely studied over the past century, and

its most stable polymorph at ambient
conditions, calcite, is often cited as a model of
cleation is a stochastic process in which thermal
fluctuations induce the formation of clusters that
are unstable with respect to dissolution. Clusters
sité de Grenoble 1, CNRS, 38041 Grenoble, France. 5Nanochem-
istry Research Institute, Department of Chemistry, Curtin University,
Post Office Box U1987, Perth, WA 6845, Australia. 6Department
of Earth and Planetary Science, University of California, Berkeley,
Berkeley, CA 94720, USA. 7Materials Sciences Division, Lawrence
classical crystal growth behavior (1). However, increase in both size and free energy until a thresh-
Berkeley National Laboratory, Berkeley, Berkeley, CA 94720,
findings from titration, ultracentrifugation (2, 3), old is crossed, whereupon the energy gained by USA. 8Physical Sciences Division, Pacific Northwest National
and cryogenic transmission electron microscopy forming the bulk material overcomes the penalty Laboratory, Richland, WA 99352, USA.
(cryo-TEM) (3, 4) now suggest that the onset of for creating an interface, and growth proceeds *Corresponding author. E-mail: [email protected] (A.F.W.);
CaCO3 mineralization—nucleation—contradicts spontaneously. Contrasting with this classical pic- [email protected] (J.J.D.Y.)

www.sciencemag.org SCIENCE VOL 341 23 AUGUST 2013 885

Mapping of Functional Groups in Metal-Organic Frameworks
Xueqian Kong, Hexiang Deng, Fangyong Yan, Jihan Kim, Joseph A. Swisher, Berend Smit, Omar M. Yaghi and Jeffrey A.
Reimer

Science 341 (6148), 882-885.

DOI: 10.1126/science.1238339originally published online July 25, 2013

Mapping Molecular Linkers

In metal-organic framework compounds, inorganic centers (metal atoms or clusters) are linked by bidentate
organic groups. Normally, the same group is used throughout the structure, but recently, synthesis with linkers bearing
different functional groups has produced well-defined materials. Kong et al. (p. 882, published online 25 July) combined

Downloaded from http://science.sciencemag.org/ on February 10, 2018

solid-state nuclear magnetic resonance and molecular simulations to map the distributions of linkers in these materials as
random, well-mixed, or clustered.

ARTICLE TOOLS http://science.sciencemag.org/content/341/6148/882

SUPPLEMENTARY http://science.sciencemag.org/content/suppl/2013/07/24/science.1238339.DC1
MATERIALS

REFERENCES This article cites 37 articles, 2 of which you can access for free
http://science.sciencemag.org/content/341/6148/882#BIBL

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Use of this article is subject to the Terms of Service

Science (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement of
Science, 1200 New York Avenue NW, Washington, DC 20005. 2017 © The Authors, some rights reserved; exclusive
licensee American Association for the Advancement of Science. No claim to original U.S. Government Works. The title
Science is a registered trademark of AAAS.