POLYMER CHEMISTRY

Poly[n]catenanes: Synthesis ofmolecular interlocked chains

Qiong Wu,1,2 Phillip M. Rauscher,1 Xiaolong Lang,2 Rudy J. Wojtecki,2

Juan J. de Pablo,1,3 Michael J. A. Hore,2 Stuart J. Rowan1,2,3,4*

As the macromolecular version of mechanically interlocked molecules, mechanicallyinterlocked polymers are promising candidates for the creation of sophisticatedmolecular machines and smart soft materials. Poly[n]catenanes, where themolecular chains consist solely of interlocked macrocycles, contain one of thehighest concentrations of topological bonds. We report, herein, a synthetic approachtoward this distinctive polymer architecture in high yield (~75%) via efficient ringclosing of rationally designed metallosupramolecular polymers. Light-scattering,mass spectrometric, and nuclear magnetic resonance characterization of fractionatedsamples support assignment of the high–molar mass product (number-averagemolar mass ~21.4 kilograms per mole) to a mixture of linear poly[7–26]catenanes,branched poly[13–130]catenanes, and cyclic poly[4–7]catenanes. Increased hydrodynamicradius (in solution) and glass transition temperature (in bulk materials) were observed uponmetallation with Zn2+.

Mechanically interlockedmolecules (MIMs)such as catenanes, rotaxanes, and knotsare attracting increasing attention onaccount of their aesthetic topologicalstructures (1–3) and their potential in ap-

plications that range from molecular machines(4) to catalysis (5), drug delivery (6), and switch-able surfaces (7). An impressive array of small-moleculeMIMshas been successfully synthesized,primarily through proper orientation of the pre-cursor components via a noncovalent templatingevent that employs, for example, metal-ligand(1, 2, 8–10), donor-acceptor (11), or hydrogen bond-ing (12) interactions or the hydrophobic effect(3), followed by a covalent fixing step. High molarmass MIMs, or mechanically interlocked poly-mers (MIPs), such as polyrotaxanes and poly-catenanes, offer access to useful properties forfunctional soft materials platforms (13). Perhapsthe most notable example to date is providedby the slide-ring gels (a subset of polyrotaxanes),where the mechanically interlocked rings act asmobile cross-linking sites. These materials ex-hibit anti-scratch and healing characteristics andhave been successfully applied as anti-scratchcoatings for cell phones and automobiles (14).As highlighted by this example, the unusual mo-bility elements in MIPs (e.g., low-energy barrierfor the sliding of the ring along a thread or ro-tation of rings) make them a specialized class of

polymers, with the potential to display a veryunusual matrix of properties.[n]Catenanes consist of n consecutively inter-

locking rings. Poly[n]catenanes are polymer chainscomposed entirely of interlocked rings and canbe thought of as [n]catenanes for which n is largeand, generally, dispersityĐ > 1. These interlockedpolymers contain a very high concentration oftopological bonds and can be considered themolecular equivalent of a macroscopic chain(Fig. 1A). Such structures retain their flexibilityno matter the stiffness of their ring componentson account of the conformational mobility oftheir topological structures, allowing access toboth high strength and excellent flexibility. Fig-ure 1B shows themain conformationalmobilities(rotational, elongational, and rocking motions)in the backbone repeat unit of such a chain forappropriate ring sizes (13). Theoretical studieshave suggested that poly[n]catenanes with suchmobility elements could exhibit a large loss mod-ulus and a low activation energy for flow andcould potentially act as superior energy-dampingmaterials and/or elastomers with excellent tough-ness and stimuli-responsive mechanical proper-ties (13, 15, 16).The synthesis of poly[n]catenanes, however,

represents a major challenge. The confirmedsynthesis of these polymeric topologies of highmolar mass has yet to be reported. Most of theearlier work on polycatenanes has focused onpoly([2]catenane)s (generally accessedbypolymer-izing a bis-functionalized [2]catenane) (15, 17–23),poly- and oligocatenanes with unclearly definedarchitectures (24, 25), and polymeric [2]catenanes(i.e., two interlocked cyclic polymers) (26). Thesematerials have afforded insights into the po-tentially useful properties of MIPs. For example,poly([2]catenane)s have shown conformational

flexibility along the polymer chain in solution(indicated by a Kuhn segment length that isshorter than the length of its [2]catenane moiety)(15) and temperature-dependent motion of therings in solution (18) or in bulk materials (22).However, on account of the covalent linkingunits present in the poly([2]catenane) backbone,the degree of rotation of the ring componentsin the catenane moieties is limited (Fig. 1C).Poly[n]catenanes, on the other hand, have ringslinked solely by topological bonds, allowing fullrotational mobility of every ring (with sufficientring size), and, as such, should exhibit moreconformational flexibility. One of themost success-ful attempts to access oligo[n]catenanes withat least seven interlocked rings was recentlyachieved via the ring-opening polymerization ofmetallo-[2]catenanes by Meijer, Di Stefano, andco-workers (25). However, the specific architec-ture of the product could not be clearly defined.To date, the longest interlocked molecular chainwith clearly defined structure is a linear [5]catenane(olympiadane) reported by Stoddart and co-workers (27) and,more recently, by Iwamoto et al.(28), bothvia stepwiseapproaches. Stoddart and co-workers have also used a similar strategy to access abranched [7]catenane (29). These step-wise strat-egies, however, are not efficient methodologiesfor the synthesis of long-chain poly[n]catenanes.Onekeychallenge inthesynthesisofpoly[n]catenanesis the conflicting reaction conditions for access-ing macrocycles (requiring low concentration)andpolymers (requiring high concentration). Takata,Kihara, and co-workers proposed an elegantsolution to this conflict by converting a bridgedpoly([2]catenane) into poly[n]catenane (20), butto date there is no report of a poly[n]catenanesynthesized via this route.The synthetic strategy toward poly[n]catenanes

outlined herein, and illustrated in Fig. 1D, aimsat decoupling the above-mentioned conflictingsynthetic requirements through the preassemblyof a metallosupramolecular polymer (MSP) as atemplate, followed by an efficient ring-closing re-action and demetallation of the resulting metal-lated poly[n]catenane. Pioneered by Sauvage andco-workers, metal-ligand coordination has beenone of themost successful synthetic routes towardMIMs (1, 2, 8, 9). The terdentate ligand, 2,6-bis(N-alkyl-benzimidazolyl)pyridine (Bip) (Fig. 1E), whichbinds transition metal ions such as Zn2+ or Fe2+

in a 2:1 stoichiometry, was chosen because it hasalready been shown to access a wide selection offunctional metallosupramolecular polymers thatexhibit photo-healing (30), stimuli-responsive (31),shape-memory (32), and actuation properties (33).Furthermore, we have recently shown that [3]catenanes can be accessed via a near quantitativering-closing reaction of a pseudo[3]rotaxane as-sembled fromamonotopic Bip-containingmacro-cycle, Zn2+, and a ditopic thread (1with R = hexyl)in a ratio of 2:2:1 (9).Building on these studies, we targeted the MSP

assembly of the rationally designed 68-memberedditopic Bipmacrocycle 2with the threadingmol-ecule 1 and the subsequent ring closing of 1 (Fig.1E). Upon the addition of 2 equivalents of Zn2+ to

RESEARCH

Wu et al., Science 358, 1434–1439 (2017) 15 December 2017 1 of 6

1Institute for Molecular Engineering, University of Chicago,Chicago, IL 60637, USA. 2Department of Macromolecular Scienceand Engineering, Case Western Reserve University, Cleveland,OH 44106, USA. 3Materials Science Division and Institute forMolecular Engineering, Argonne National Laboratory, 9700South Cass Avenue, Lemont, IL 60439, USA. 4Department ofChemistry, University of Chicago, Chicago, IL 60637, USA.*Corresponding author. Email: stuartrowan@uchicago.edu

on March 12, 2020

http://science.sciencem

ag.org/D

ownloaded from

the 1:1 mixture of 1 and 2, the components self-assembled into the alternating supramolecularcopolymer [Fig. 1D; monitored by 1H nuclearmagnetic resonance (NMR), see fig. S1]. The al-ternation is a consequence of the principle ofmax-imal site occupancy (34): Two Bipmoieties in thesame (or different) macrocycle 2 cannot bind thesame Zn2+ ion; therefore, each Bip unit in2mustform a 2:1 Bip:Zn2+ ion complex with a Bip unitin 1 to maximize enthalpic gain (fig. S2). The for-mation of an MSP was confirmed by diffusion-ordered spectroscopy (DOSY) (fig. S3). The largebinding constant (>106 M–1 in acetonitrile) be-tween Bip and Zn2+ (35)maintains theMSP underthe relatively dilute conditions required to favorthe ring-closing reaction (although cyclic MSPsshould be favored under very dilute conditions).The ring-closing metathesis reaction showed ex-cellent olefin conversion at 2.5 mM (with respectto 1) (fig. S4), and the resulting reaction mixturecould be readily demetallatedwith tetrabutylam-monium hydroxide. Figure 2A shows a compar-ison of the partial 1H NMR spectrum of thedemetallated crude product (focusing on the pyr-idyl protons HA/a) with all possible noninterlockedby-products 4 (macrocycle obtained from 1), 2,and 5 [acyclic diene metathesis (ADMET) poly-merized 1, see fig. S5 for structure]. The majorityof the HA/a peaks of the crude product show anupfield chemical shift [in the region of 8.11 to8.27 parts permillion (ppm), highlighted in yellow]that does not correspond to any of the possiblenoninterlocked species but is in good agreementwith the chemical shifts observed in the previousstudies on Bip-containing [3]catenanes (9), con-sistentwith the formationof interlocked structures.The crude product also contains a small percentage(~20%) of noninterlocked molecules that were as-signed to unreacted 2 and polymer/oligomer 5 onthe basis of their 1H NMR chemical shift.To facilely remove thenoninterlockedby-product,

we developed a procedure (Fig. 2B) that took ad-vantage of the preorganized interlocked structureof the targeted poly[n]catenane. According to theinterlocked ligand effect (i.e., the catenand effect)(8, 36), metal ions preferentially bind with me-chanically interlocked ligands rather thannoninter-locked free ligands on account of the reducedentropic cost of bringing the ligands together.1H NMR titration studies (fig. S5B) confirmedthat upon addition of Zn2+ to the crude product,the species associated with the upfield-shiftedpyridyl protons disappear when ~70% of the Bipmoieties aremetallated, consistent with selectiveremetallation of the interlocked species. The non-interlocked by-products remained unmetallated,allowing them to be easily separated by solid-liquid extraction (with 2:1 v/v chloroform/hexane),leaving the pure poly[n]catenane 3 as a solid in~75% yield. This purification step also removessome oligomeric interlocked products, primarilyon account of the smaller number of interlockedmetal-ion binding sites making them statisticallyless likely to bind enough metal ions to altertheir solubility.All NMRpeaks of the interlocked product3have

been assigned on the basis of two-dimensional

NMR studies (figs. S6 to S8). The spectroscopyconfirms that 3 contains ~50% of macrocycle 2and ~50% of a component (or components) de-rived from 1 (on the basis of 1H NMR integration)(fig. S10), which is consistent with the formationof the targeted alternating poly[n]catenane struc-ture. The interlocked structure is further con-firmed by nuclear Overhauser effect spectroscopy(NOESY) (Fig. 2C). Lowering the temperature(−17°C) to diminish the ring motions in the poly-catenane gives rise to numerous NOE cross peaksin dilute (1% w/v) solutions of 3. Of particularinterest are intercomponent NOE cross peaksbetween protons in the different macrocycles(2 and 4). At least four intercomponent NOEcross peaks—including, for example, one betweenHL of 4 and Hn/o of 2—were observed in 3 (Fig.2C) but not in the simple mixtures of 2 and 4or 2 and 5 (figs. S11 and S12). Given that NOEsignals normally require the protons to be <0.5 nmapart, these data strongly suggest that macro-cycles 2 and 4 have to be interlocked and that3 is a poly[n]catenane.Having confirmed the interlocked structure,

we next used gel permeation chromatography

coupled to a multiangle light scattering (GPC-MALS) detector to measure the absolute molarmass of the polymers (37). TheGPC chromatogramof 3 (Fig. 3A, black line) shows a broad peak con-sistent with a averagemolarmassMn = 21.4 kg/moland dispersity Đ = 1.44 (Table 1). At each elutionvolume, the absolute molar mass is shown asthe red dotted line, indicating that the productdistribution 3 contains species with molar massup to ~200 kg/mol, which corresponds to a poly[130]catenane. Deconvolution of the broad peakin the chromatogram yields three subpeaks (I toIII), implying that the poly[n]catenane productmay contain three different architectures. Themeasured absolute molar masses (from MALS)of these three peaks are about 20 to 200, 10 to40, and 6 to 10 kg/mol, which would correspondto poly[13–130]catenanes, poly[7–27]catenanes,and poly[4–7]catenanes, respectively. The relativepercentage of the three peaks is estimated (onthe basis of peak areas) to be 28, 61, and 11%,respectively.One possible explanation for peaks I to III is

the formation of branched, linear, and cyclicarchitectures, respectively, as illustrated in Fig. 3B.

Wu et al., Science 358, 1434–1439 (2017) 15 December 2017 2 of 6

Fig. 1. Structure and conformational mobility of and synthetic approach to poly[n]catenanes.(A) Schematic representation of the poly[n]catenane architecture, which can be thought of asthe molecular equivalent of a robust and flexible metal chain. (B) Common conformational motionsobserved in catenanes. (C) Comparison of the ring rotational mobilities in a poly([2]catenane)versus a poly[n]catenane. (D) Targeted synthesis of poly[n]catenane 3 via assembling 1 and 2 intoa metallosupramolecular polymer (MSP), followed by ring-closing to yield a poly[n]catenate (i.e.,metallated poly[n]catenane) and demetallation. (E) Structure of 1, 2, and 4 (ring-closed productof 1). A few key protons are labeled with blue capital letters in 4 and red lowercase letters in 2. Thebox highlights the structure of the Bip ligand moiety.

RESEARCH | REPORTon M

arch 12, 2020

http://science.sciencemag.org/

Dow

nloaded from

In addition to the targeted linear poly[n]catenanes,it is certainly possible that interchain reactions(dimerization and oligomerization of 4) can oc-cur, leading to higher–molarmass branched poly[n]catenanes. Furthermore, it is also possiblethat a percentage of the MSPs are lower–molar

mass cyclic structures, which in turn would leadto cyclic polycatenanes. One piece of evidencesuggestive of cyclic polycatenanes is the observedtransition peak between peaks II and III in theabsolute molar mass measurement (Fig. 3A, at~9.2 ml). If peak III corresponds to a cyclic geom-

etry, we would expect these more compact struc-tures to co-elute with lower–molar mass linearpolymers (38).To more fully characterize the poly[n]catenane

mixture, we performed preparative GPC on 3,obtaining four fractions 3a to 3d. 1H NMR studiesconfirmed that all four fractions contain ~50%of macrocycle 2, consistent with them all beingpoly[n]catenanes (fig. S14). GPC-MALS of eachfraction confirmed successful size-based sepa-ration (Fig. 3C and fig. S15), and the four frac-tions showed decreasing molar mass (Table 1)with relatively narrow Đ (1.11 to 1.18) (Table 1).On the basis of the elution times, fractions 3a and3b mostly correspond to peak I, 3c predomi-nantly corresponds to peak II, and 3d containsamixture of peaks II and III. Matrix-assisted laserdesorption/ionization–time-of-flight (MALDI-TOF)mass spectrometry confirmed the presence ofhigh–molar mass polymers with molecular ionpeaks corresponding up to a poly[53]catenane in3b (Fig. 3D) and up to a poly[27]catenane for 3c(fig. S16).One way to characterize the architectures

(branched, linear, or cyclic) of the poly[n]catenanesin fractions 3a to 3d is to elucidate the averagenumber of chain ends (NC) of the polymers ineach fraction. By definition,NC equals 0 for cyclicpoly[n]catenanes, 2 for linear poly[n]catenanes,and ≥4 for branched poly[n]catenanes (Fig. 3B).With absolute molar mass of each fraction avail-able fromMALS, their degrees of polymerization(DPs, which we define here as equivalent to n)can be readily calculated by MMALS/1544 (re-sults listed in Table 1), whereMMALS is the molarmass fromGPC-MALS and 1544 g/mol is themeanmolar mass of 2 and 4 (assuming a 1:1 ratio in thepolymer). It is then possible to calculate NC bycombining these data with NMR chain-end anal-ysis (eq. S1). The key for such calculations is to finddiagnostic chain-end peaks of poly[n]catenanesin the 1H NMR spectra. To do this, a (predom-inantly) [3]catenane 6 was targeted and syn-thesized by reacting 1 in the presence of twoequivalents of 2 and Zn2+ ions (fig. S17), and itsNMR spectrum was compared with 3a to 3d inFig. 3E. The three groups of peaks for HA/a of 6with ~1:1:1 peak areas (fig. S17)were observed anddefined as regions a (8.235 to 8.270 ppm), b(8.195 to 8.235 ppm), and g (8.150 to 8.195 ppm),respectively. The lower intensity of region a in3a to 3d, comparedwith the intensity in 6, alongwith the longer NMR relaxation time of theprotons in region a (fig. S18), confirms that thisregion corresponds to the chain end (see supple-mentary materials). With this assignment, theaverage number of chain ends, NC, can be cal-culated by Eq. 1

NC ¼ DP � 2IaIA=a

ð1Þ

where IA/a is the total peak integration of HA/a

and Ia is the chain-end peak integration of HA/a

(i.e., region a; see eqs. S1 to S3 for derivation de-tails). The results of theNC calculations for 3 and

Wu et al., Science 358, 1434–1439 (2017) 15 December 2017 3 of 6

Fig. 2. NMR characterization and purification of the poly[n]catenanes. (A) Partial 1H NMR(600 MHz, CDCl3) spectrum of the HA/a region of the purified poly[n]catenane compared with spectraof individual rings 2 and 4 and by-product ADMETpolymer 5 (for structure, see fig. S5). The yellowregion highlights protons corresponding to interlocked Bip moieties. (B) Chromatography-freepurification of poly[n]catenanes via selective metallation and extraction. (C) Partial 1H-1H NOESYspectrum of the purified poly[n]catenane solution (1% w/v in CDCl3) measured at −17°C. Some NOEcross peaks between the 2 and 4 moieties are highlighted. Red and blue letters are as in Fig. 1E. For thefull spectrum, see fig. S13.

Mn (kg/mol)

Table 1. Molar mass and architecture of 3 and 3a to 3d. Mn, number-average molar mass; DP,

degree of polymerization; Đ, dispersity; NC, average number of chain ends; NCEx, average number of

chain ends for acyclic poly[n]catenanes.

RESEARCH | REPORTon M

arch 12, 2020

http://science.sciencemag.org/

Dow

nloaded from

3a to 3d are summarized in Table 1. Fractions3a(NC = 8.7) and 3b (NC = 4.7) are predominantlybranched, whereas 3c (NC = 1.7) and 3d (NC =1.2) appear to contain different amounts of cyclicpoly[n]catenanes that result in NC < 2. To iden-tify the proportion of cyclic poly[n]catenanesin these latter fractions, additional studieswere undertaken to specifically target a cyclicpoly[n]catenane 7. This was achieved by carryingout theMSP formation (usingFe2+ as themetal ion)and ring-closing reaction at much lower concen-trations (see supplementarymaterials and fig. S19).Comparison of the HA/a region in the 1H NMRspectra (Fig. 3E) of the cyclic poly[n]catenane–containing fractions (3c, 3d, and 7) shows anadditional upfield shift to 8.06 to 8.15 ppm (Fig.3E, green region), which is defined as region d. Incontrast, the acyclic poly[n]catenane–containingsamples (3a, 3b, and 6) do not have any peaks inthis region. Therefore, it is reasonable to concludethat region d is specific for cyclic poly[n]catenanes.Compared with the acyclic poly[n]catenanes (3aand3b) or [3]catenane6, it appears that ~50%ofthe A/a peaks of the cyclic poly[n]catenane 7 are

shifted more upfield. One possible explanationis the cyclic architecture restricts the confor-mational mobility, resulting in amore compactmolecule (fig. S19E) and allowing additional shield-ing effects from neighboring aromatic moieties.Given that the integral of the HA/a in 7 shows that~50% of its HA/a protons are present in the dregion (fig. S20), it is possible to determine theaverage number of chain ends for the acyclicpoly[n]catenanes (NC

Ex) present in each fractionby Eq. 2 (for derivation, see eqs. S3 to S5)

NCEx ¼ DP � 2Ia

IA=a � 2Idð2Þ

where Id is the peak integration of the d region inthe NMR spectra (Fig. 3E). The results of thesecalculations show that both3c and3d haveNC

Ex≈2 (with NC < 2), consistent with the fact thatthey both contain a mixture of linear and cyclicpolycatenaneswith the percentage of cyclic poly-catenanes in 3c and 3d being ~15 and ~34%,respectively (calculated via eq. S6). Table 1 sum-

marizes themolar mass and architectural data of3 and the four fractions 3a to 3d. The fractionwith the highest molar mass (3a) is assigned tohighly branched poly[32–130]catenanes (on thebasis of MALS analysis) (fig. S15A) with an aver-age DP = 55. Fraction 3b consists of a lowerdegree of branching poly[n]catenanes with anaverage DP = 25. Fractions 3c and 3d containmainly linear (with some cyclic) polycatenaneswith up to 27 (averageDP = 11) (fig. S15C) and 20(average DP = 8) (fig. S15D) interlocked rings,respectively. Based on the data from the four frac-tions, we can estimate that the overall percentageof branched, linear, and cyclic poly[n]catenanesis ~24, ~60, and ~16%, respectively, whichmatches well with the GPC peak deconvolutionresult (28, 61, 11%, respectively) (Fig. 3A). Takentogether these data are consistent with thepoly[n]catenane 3 being a mixture of branchedpoly[13–130]catenanes, linear poly[7–27]catenanes,and cyclic poly[4–7]catenanes, in approximatelythe aforementioned ratios.It is known that catenanes prepared via metal

templating exhibit metallo-responsive behavior

Wu et al., Science 358, 1434–1439 (2017) 15 December 2017 4 of 6

Fig. 3. Molar mass and architecture characterization of thepoly[n]catenanes. (A) GPC chromatogram, MALS absolute molar massmeasurement, and peak deconvolution of 3. a.u., arbitrary units. (B) Proposedarchitectures corresponding to the three deconvoluted peaks, the formationmechanism, and the corresponding average number of chain ends (NC). RCM,ring-closing metathesis. (C) GPC chromatograms of fractionated polycate-nane fractions (3a to 3d, intensity normalized by their relative masspercentage) compared with that of the unfractionated poly[n]catenane 3. The

elution volume ranges of the three deconvoluted peaks (I to III) are alsoshown. (D) MALDI-TOF spectra of polycatenane fraction 3b. The lower peaks(marked with asterisks) are from multiply charged molecular ions. m/z,mass/charge ratio. (E) Partial 1H NMR (500 MHz, CDCl3) spectra showingthe HA/a region of 3a to 3d compared with that of two controls: [3]catenane(6) and cyclic polycatenane (7). The four regions (a, b, g, d) are defined.Thechain-end peaks and specific peaks for cyclic catenanes are highlighted inyellow and green, respectively.

RESEARCH | REPORTon M

arch 12, 2020

http://science.sciencemag.org/

Dow

nloaded from

(36). As such, the metallo stimuli–responsiveproperties of these poly[n]catenanes were inves-tigated in both solution and bulk, by measuringthe hydrodynamic radius Rh [with dynamic lightscattering (DLS)] (Fig. 4A and fig. S21) and theglass transition temperature (Tg) (Fig. 4B andfig. S22), respectively. Addition of Zn2+ ions intopoly[n]catenanes should lock the conformation-al motions of the rings, thereby inducing a switchfrom a highly flexible polymer to a semi-rigidpolymer, where the flexibility comes primarilyfrom the tetraethylene glycol (TEG) moieties (Fig.4A). For predominantly linear poly[n]catenanes(3c, averageDP= 11), the addition of Zn2+ shows asubstantial (~70%) Rh increase from 3.9 to 6.6 nm(Fig. 4A). This result agrees well with all-atommolecular dynamics simulations of metallatedlinear poly[n]catenane (see supplementary mate-rials). Molecules were simulated with the OPLS(optimized potentials for liquid simulations) all-atom force field (39)with 1,1,2,2-tetrachloroethaneas the solvent, and the properties of the polymerwere probed using a force-extension experiment.Tomodel a poly[n]catenane under tension, an [8]catenane was connected to itself through periodicboundary conditions (Fig. 4C); the extension wasenforced through the system (box) length, whichis systematically varied; and the correspondingtension follows from the pressure tensor (eq. S9).The force-extension behavior is well describedby the extensible worm-like chain (EWLC)model

(40), as shown in Fig. 4C. According to this in-terpretation, each macrocycle’s contribution tothe contour length, l = 2.40 ± 0.03 nm, is con-siderably smaller than the persistence length, lp =7.46 ± 2.88 nm, which is characteristic of asemi-rigid polymer. Again, this flexibility canbe primarily attributed to the TEG segments ofmacrocycle 2, as evidenced by larger fluctuationsin the intermetal distances for 2 versus 4 (figs.S28 and S29). Based on simulations of a non-periodic metallated [4]catenane, Rh is ~34% ofthe end-to-end distance (Ree). For a metallatedpoly[n]catenane of arbitrary DP, Ree can be cal-culated using worm-like chain statistics (eq. S15)and the results of the force-extension simulations.Rh for an average poly[n]catenane in fraction 3cis thereby estimated at 6.89 ± 1.38 nm, in goodagreement with DLS results (6.6 nm), furthersupporting the conclusion that fraction 3c isprimarily linear poly[n]catenane (see supplemen-tary materials). The simulated structure of ametallated linear poly[n]catenane under ~15 pNtension is illustrated in Fig. 4D, showing a seg-ment of 12 interlocked rings.Similar to linear poly[n]catenane, metallation

of branched poly[n]catenane (3b, averageDP =24.9) with Zn2+ shows a ~50% increase in Rh uponmetallation (5.0 to 7.6 nm) (fig. S21B), presumablyas a consequence of an extension of its arms.Metallation of cyclic poly[n]catenane 7 (averageDP= 6) shows only a relatively small change ofRh

(~10%, from 2.4 to 2.6 nm) (fig. S21C), which isconsistent with its cyclic architecture restrictingchain extension. In the bulk, the Tg values ofpoly[n]catenane with predominantly linear (3c)and branched architectures (3b) were found tobe 97° and 104°C, respectively, compared with137°C for 5 (the linear ADMET polymer of 1)[Fig. 4B and fig. S22, determined by differentialscanning calorimetry (DSC)]. In general, polymerswith greater freedom of segmental motion showlower Tg values, which is consistent with themore facile conformational molecular motionsexpected throughout the poly[n]catenane back-bones. Metallation with Zn2+ locks these confor-mational motions and substantially reduces theflexibility of the poly[n]catenane; as such, no Tgvalues are observed upon heating up to 160°C forboth linear and branched samples (Fig. 4B andfig. S22).In conclusion, the successful synthesis of main-

chain poly[n]catenanes via an MSP templatedstrategy has been achieved in ~75% yield. Theisolated product mixture encompasses linear poly[7–27]catenanes, branchedpoly[13–130]catenanes,and cyclic poly[4–7]catenanes. This synthetic strat-egy opens the door to the design and synthesis ofa variety of new interlocked polymers.

REFERENCES AND NOTES

1. K. S. Chichak et al., Science 304, 1308–1312 (2004).2. J. J. Danon et al., Science 355, 159–162 (2017).

Wu et al., Science 358, 1434–1439 (2017) 15 December 2017 5 of 6

Fig. 4. Experimental studies on and simulation of the metallatedpoly[n]catenanes. (A) Illustration of metallo-responsive conformationalchange of a linear poly[n]catenane between flexible and semi-rigid structures,as indicated by hydrodynamic radius (Rh) change (fig. S21).The flexibility ofthe metallated poly[n]catenane is primarily due to the TEG moiety ofmacrocycle 2. (B) DSC data for linear poly[n]catenanes (demetallated andmetallated) compared with data for ADMETpolymer 5 (Tg calculated byhalf-height midpoint). “Exo Up” indicates that the direction of exothermic heat

flow is up. (C) Simulated force-extension curve produced by varying the boxlength of a metallated linear [8]catenane connected to itself through periodicboundary conditions in the z direction. Data from simulations (green circles,size of data point includes error) are fit by the extensible worm-like chain(EWLC) model (black dashes). (D) Visualization of a metallated linearpoly[n]catenane (showing a segment of 12 interlocked rings) under a tensionof ~15 pN from all-atom molecular dynamics simulations in 1,1,2,2-tetrachloroethane solvent (solvent and hydrogen atoms omitted for clarity).

RESEARCH | REPORTon M

arch 12, 2020

http://science.sciencemag.org/

Dow

nloaded from

3. N. Ponnuswamy, F. B. L. Cougnon, J. M. Clough, G. D. Pantoş,J. K. M. Sanders, Science 338, 783–785 (2012).

4. S. Erbas-Cakmak, D. A. Leigh, C. T. McTernan,A. L. Nussbaumer, Chem. Rev. 115, 10081–10206 (2015).

5. P. Thordarson, E. J. A. Bijsterveld, A. E. Rowan, R. J. M. Nolte,Nature 424, 915–918 (2003).

6. K. K. Cotí et al., Nanoscale 1, 16–39 (2009).7. J. J. Davis, G. A. Orlowski, H. Rahman, P. D. Beer, Chem.

Commun. 46, 54–63 (2010).8. C. O. Dietrich-Buchecker, J. P. Sauvage, J. P. Kintzinger,

Tetrahedron Lett. 24, 5095–5098 (1983).9. R. J. Wojtecki et al., Chem. Sci. 4, 4440 (2013).10. H. Lahlali, K. Jobe, M. Watkinson, S. M. Goldup, Angew. Chem.

Int. Ed. 50, 4151–4155 (2011).11. P. R. Ashton et al., Angew. Chem. Int. Ed. Engl. 28, 1396–1399

(1989).12. G. A. Breault, C. A. Hunter, P. C. Mayers, Tetrahedron 55,

5265–5293 (1999).13. Z. Niu, H. W. Gibson, Chem. Rev. 109, 6024–6046

(2009).14. Y. Noda, Y. Hayashi, K. Ito, J. Appl. Polym. Sci. 131, 40509

(2014).15. J. L. Weidmann et al., Chemistry 5, 1841–1851 (1999).16. T. Pakula, K. Jeszka, Macromolecules 32, 6821–6830

(1999).17. J. Weidmann et al., Chem. Commun. 1996, 1243–1244

(1996).18. C. Hamers, F. M. Raymo, J. F. Stoddart, Eur. J. Org. Chem.

1998, 2109–2117 (1998).19. D. Muscat et al., Macromolecules 32, 1737–1745 (1999).20. N. Watanabe, Y. Ikari, N. Kihara, T. Takata, Macromolecules 37,

6663–6666 (2004).

21. C. Hamers, O. Kocian, F. M. Raymo, J. F. Stoddart, Adv. Mater.10, 1366–1369 (1998).

22. C. A. Fustin et al., J. Am. Chem. Soc. 125, 2200–2207(2003).

23. B. N. Ahamed, P. Van Velthem, K. Robeyns, C. A. Fustin,ACS Macro Lett. 6, 468–472 (2017).

24. K. Endo, T. Shiroi, N. Murata, G. Kojima, T. Yamanaka,Macromolecules 37, 3143–3150 (2004).

25. J. A. Berrocal et al., Macromolecules 48, 1358–1363(2015).

26. P. F. Cao, J. D. Mangadlao, A. de Leon, Z. Su, R. C. Advincula,Macromolecules 48, 3825–3833 (2015).

27. D. B. Amabilino, P. R. Ashton, A. S. Reder, N. Spencer,J. F. Stoddart, Angew. Chem. Int. Ed. Engl. 33, 1286–1290(1994).

28. H. Iwamoto et al., Chem. Commun. 52, 319–322 (2016).29. D. B. Amabilino et al., Angew. Chem. Int. Ed. Engl. 36,

2070–2072 (1997).30. M. Burnworth et al., Nature 472, 334–337 (2011).31. R. J. Wojtecki, M. A. Meador, S. J. Rowan, Nat. Mater. 10, 14–27

(2011).32. J. R. Kumpfer, S. J. Rowan, J. Am. Chem. Soc. 133,

12866–12874 (2011).33. B. T. Michal, B. M. McKenzie, S. E. Felder, S. J. Rowan,

Macromolecules 48, 3239–3246 (2015).

34. R. Kramer, J. M. Lehn, A. Marquis-Rigault, Proc. Natl. Acad.Sci. U.S.A. 90, 5394–5398 (1993).

35. S. J. Rowan, J. B. Beck, Faraday Discuss. 128, 43–53(2005).

36. J. E. M. Lewis, M. Galli, S. M. Goldup, Chem. Commun. 53,298–312 (2016).

37. P. J. Wyatt, Anal. Chim. Acta 272, 1–40 (1993).

38. C. W. Bielawski, D. Benitez, R. H. Grubbs, Science 297,2041–2044 (2002).

39. W. L. Jorgensen, D. S. Maxwell, J. Tirado-Rives, J. Am. Chem.Soc. 118, 11225–11236 (1996).

40. T. Odijk, Macromolecules 28, 7016–7018 (1995).

ACKNOWLEDGMENTS

We thank K. Herbert for help with figure design andsuggestions on DSC data analysis, P. Agrawal for DOSYmeasurements, and F. Etheridge for helpful discussions.Special thanks to S. Shang for the picture of herbracelet for Fig. 1A and R. Rauscher for assistance withseveral figures. This work is funded by the NSF undergrants CHE 1700847, CHE 1402849, and CHE 1151423 and bythe University of Chicago Materials Research Science andEngineering Center, which is funded by the NSF under awardnumber DMR-1420709. Additional data are available in thesupplementary materials.

SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/358/6369/1434/suppl/DC1Materials and MethodsSupplementary TextFigs. S1 to S29Tables S1 to S3References (41–93)

2 September 2017; accepted 13 November 2017Published online 30 November 201710.1126/science.aap7675

Wu et al., Science 358, 1434–1439 (2017) 15 December 2017 6 of 6

RESEARCH | REPORTon M

arch 12, 2020

http://science.sciencemag.org/

Dow

nloaded from

]catenanes: Synthesis of molecular interlocked chainsnPoly[Qiong Wu, Phillip M. Rauscher, Xiaolong Lang, Rudy J. Wojtecki, Juan J. de Pablo, Michael J. A. Hore and Stuart J. Rowan

originally published online November 30, 2017DOI: 10.1126/science.aap7675 (6369), 1434-1439.358Science 

, this issue p. 1434Scienceflushed out.through flanking preformed macrocycles, after which metathesis catalysis closed up the first ring, and the metal could berings are consecutively interlinked. The key was using zinc ions to template the threading of one macrocycle precursor

now report successful synthesis of polycatenanes in which tens ofet al.of interlocked rings amid spacer segments. Wu macroscopic chain links have proven surprisingly challenging to make. Most approaches have settled for tethering pairs

Although polymer strands are often informally called chains, molecular topologies that actually resemble extendedA little zinc makes the rings all link

ARTICLE TOOLS http://science.sciencemag.org/content/358/6369/1434

MATERIALSSUPPLEMENTARY http://science.sciencemag.org/content/suppl/2017/11/29/science.aap7675.DC1

REFERENCES

http://science.sciencemag.org/content/358/6369/1434#BIBLThis article cites 82 articles, 6 of which you can access for free

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

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAAS.ScienceScience, 1200 New York Avenue NW, Washington, DC 20005. The title (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement ofScience

Science. No claim to original U.S. Government WorksCopyright © 2017 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of

on March 12, 2020

http://science.sciencem

ag.org/D

ownloaded from