Phase Change Transformations with Dynamically Addressable Aminal

Oct 9, 2017 - Dynamic polymers assembled through hemiaminal and aminal functionalities reversibly fragment upon binding to trivalent metals. Gels prod...
15 downloads 13 Views 5MB Size
Download PDF
Article Cite This: J. Am. Chem. Soc. 2017, 139, 15385-15391

pubs.acs.org/JACS

Phase Change Transformations with Dynamically Addressable Aminal Metallogels Peter J. Boul,*,† Peter D. Jarowski,‡ and Carl J. Thaemlitz† †

Aramco Research Center, Houston, Texas 77061, United States ChemAlive LLC, 1003 Lausanne, Switzerland

‡

S Supporting Information *

ABSTRACT: Dynamic polymers assembled through hemiaminal and aminal functionalities reversibly fragment upon binding to trivalent metals. Gels produced with these dynamic polymers are broken down to liquids after the addition of metal salts. Nuclear magnetic resonance spectroscopy studies and density functional theory calculations of intermediates reveal that the presence of these metals causes shifts in the energetic landscape of the intermediates in the condensation pathway to render stable nonequilibrium products. These species remain stable in the liquid phase at room temperature but convert to gels upon heating. With thermal activation, the fragmented ligands transform catalytically into closed-ring hexahydrotriazine products, which are macroscopically observable as new gels with distinct physical properties. The interplay between equilibrium and nonequilibrium gels and liquids and the ligands responsible for these transformations has been observed rheologically, giving controlled gel times dictated by the thermodynamics and kinetics of the system. This constitutionally dynamic macromolecular system offers the possibility of harnessing an equilibrium/nonequilibrium system in tandem with its inherent self-healing properties and triggered release functionality.

■

INTRODUCTION The stimulus-responsive nature of constitutionally dynamic materials (CDMs) unlocks powerful possibilities in advanced materials design. This attribute can be used to facilitate triggered release,1 reversible gelation,2 and self-healing in composite materials.3 CDMs use reversible covalent and/or noncovalent bonding such that they are able to undergo dynamic modifications to their constitutions through the dissociation and reassociation of their constituent building blocks.4,5 The dynamics can be addressed, depending on the reversible reaction and the material’s constitution, through changes in pH, temperature, light, or electric and magnetic fields. An assortment of different reversible covalent reactions has been investigated for this class of materials. These include reversible hydrazone formation,6 reversible Schiff base formation,7 reversible aminal formation,8 Diels−Alder condensations,9 disulfide exchange,10 dithioacetal exchange,11 dynamic boronic ester formation,12 olefin metathesis,13 and metal−ligand association.14 Aminal functionalities are of particular interest for their ability to undergo transformations in response to a wide variety of chemical stimuli. Hemiaminal and aminal polymers have been reported to undergo pH responsive phase change and network rearrangements, which has given them promise as recyclable plastics, self-healing polymers, and stimulus-responsive materials.15 In addition to pH responsivity, hemiaminal gels also show dynamics through aminal/thiol-exchange.16 Coordination of dynamic aminal bonds to metals has been demonstrated in the dynamic modification of triaza-adamantanes.17 © 2017 American Chemical Society

Both constitutionally and motionally covalent dynamic chemistries have been demonstrated through reversible imine formation and the transformation of imines into aminals.18 It has been observed that the complexation of certain transition metals to condensation products of polyamines and aldehydes can result in a switch from the aminal condensation product to the Schiff base.19 Other studies have shown the utility of metal templation to achieve thermodynamically specified chiral centers through hemiaminal intermediates.20 These investigations reveal untapped potential of aminal chemistry in CDMs. In this Article, the dynamics of hemiaminal and aminal polymers are examined for their responsivities to certain divalent and trivalent metal salts. The introduction of metal salts (aluminum and iron) to these systems provides a handle to exploit phase change dynamics and tailor mechanical properties of these dynamic material systems. Metals are used to modify the kinetics and thermodynamics of hemiaminal and aminal inclusive CDMs. In the interest of instructing the dynamics of aminal gels, a hemiaminal polymer gel, produced through the condensation of a multifunctional amine and formaldehyde, can be converted into a liquid through the addition of a trivalent metal. The resulting organometallic liquid can then be thermally transformed with a controlled and modifiable gelation time into a different kind of gel with different physical properties. It is then possible to Received: July 14, 2017 Published: October 9, 2017 15385

DOI: 10.1021/jacs.7b07053 J. Am. Chem. Soc. 2017, 139, 15385−15391

Article

Journal of the American Chemical Society transform this gel back into a liquid through the addition of a reactive compound such as a thiol or phosphine. The dynamics of this system are studied with small amplitude oscillatory shear (SAOS), nuclear magnetic resonance spectroscopy, and supported by quantum chemical computations. A simplified description of the effect of trivalent metal (M(III)) on gels formed from the condensation of a polyalkylene glycol amine and formaldehyde in a polar aprotic solvent is presented in Scheme 1. The breaking of the gel occurs because of

Scheme 2. Mechanism for the Condensation of Formaldehyde with Organic Amines Producing a Series of Aminals and Hemiaminalsa

Scheme 1. Equilibrium Dynamics of the Aminal Gels with Metals and withouta

a The solvents that have been tested for this chemistry are Nmethylpyrrolidone (NMP) and N,N′-dimethylformamide (DMF).

The concentration of a selected component can then be amplified relative to that of the other components. For example, product A could be amplified over product L. The outcome of this effect is that the physical properties (i.e., mechanical strength, melting point, etc.) of a gel or resin produced from these dynamic building blocks can be altered when in the presence of a chemical stimulus.

a

The yellow sphere represents a trivalent metal cation. The open red torus represents the hemiaminal structure, L (in Scheme 2). The closed red torus is the hexahydrotriazine ring (N in Scheme 2).

■

the coordination of the CDM to metal cation. It is proposed that this occurs because components from the equilibrium shown in Scheme 2 are not as networked into the gel fraction but rather the sol fraction is amplified. The extent of this amplification is reduced when divalent metal is added in place of trivalent metal due to a difference in the energetics of binding. This difference in energetics is confirmed by density functional theory computations at the M06-2x/6-31+g(d) level with COSMO solvation in water. In the case of trivalent metals, this dynamics can be exploited for triggered release as is demonstrated herein. As confirmed with DFT, the condensation pathway for the reaction of formaldehyde and a primary amine in NMP (Scheme 2) has product N (the hexahydrotriazine) as the thermodynamic product. The other products are kinetic products and are present in the pre-equilibrium reaction product to varying degrees. The concentration distribution of kinetic products can be altered by a change in pH or through the addition of a molecule that binds to any of the reaction products, A through N, is depicted in Scheme 2. When a molecule binds to any of these products it affects the free energy of the complex and changes the equilibria and activation energies of the formaldehyde/amine condensations.

RESULTS AND DISCUSSION Paraformaldehyde can be added to N-methyl pyrrolidone to render hemiaminal gels by heating the slurry to 60 °C for 30 min to crack the paraformaldehyde.15 Following this step, a polypropylene glycol triamine (Jeffamine T5000) is added to the slurry to render the hemiaminal gel. Different ratios of amine condensed with aldehyde show the differences in melting points and shear moduli as determined by the stoichiometries of the starting materials (Figures S-1 and S-2). Table 1 indicates the proportions of the components to gels made in this manner and tested for shear response in the presence and absence of divalent iron. When an iron(II) compound, Mohr’s salt, is added to each of samples 1−5, the gels are not transformed into liquids; however, the measured melting points (Figures S-1 and S-2) for the gels are lowered from their values in the absence of iron(II) with the exception of sample 5 which is increased by 0.5 °C. Furthermore, the presence of iron(II) appears to catalyze the formation of the thermodynamically favored hexahydrotriazine structure from the open hemiaminal structure. This is clear in the temperature ramp 15386

DOI: 10.1021/jacs.7b07053 J. Am. Chem. Soc. 2017, 139, 15385−15391

Article

Journal of the American Chemical Society Table 1. Formulations for the Different Experiment A Samplesa sample designation

Jeffamine T-5000 (mmol)

amine/aldehyde ratio

1 2 3 4 5

0.8 1.2 1.6 2.0 2.4

0.55 0.83 1.1 1.4 1.7

a

Variable amine:aldehyde ratio with a formaldehyde concentration of 0.216 M.

Figure 2. Addition of a trivalent metal salt, ferric ammonium sulfate 48 h, after metal salt addition. Samples 1−5 are presented from left to right.

experiment depicted in Figure S-2b where the sample 1 hemiaminal gel with iron(II) shows a melting point of 49.0 °C and then resolidifies at 77.3 °C. The addition of ferrous ammonium sulfate also has the effect of increasing the strain tolerance for the gels of samples 1 through 5. The effect of iron(II) on shear modulus is shown in Figure 1.

Figure 3. Gel Sample 1 (a) after reaction with aluminum chloride hexahydrate and (b) after heating the liquid to 70 °C for 2 h.

Whereas the gel before the addition of M(III) is referred to as the kinetic gel. The chemistry developed for the breakdown of the kinetic gels by M(III) can be used to render triggered release of cargo. In this experiment, a naphthalene-based polymer is used as the chemical cargo. When a hemiaminal gel of the composition in Table S-4 is placed in NMP (Figure S-3), there is no appreciable disintegration of the gel over days at room temperature. On the other hand, exposure to aluminum chloride hydrate solvated in NMP causes the gel to disintegrate rapidly. Figure 4 shows the Figure 1. Sample 3 and sample 3 with Fe(II) amplitude sweep and corresponding phase angle.

There is also an observed increase in the phase shift angle at low shear strain (100% strain) past the flow point of the gel, returning to 1% strain shows that the gel returns to its original state with the same complex shear modulus. Regarding the thermodynamic gels, the gel time for the liquid produced through the addition of aluminum chloride to the gel of sample 1 depends not only on the temperature but also on the aluminum concentration in the sample. Figure 6 shows the results of a SAOS experiment where different amounts of aluminum chloride are added to the same gel composition and the resulting gel time for the organometallic liquid at 70 °C. The materials are referred to as gels when the shear modulus loss factors are measured as less than or equal to one. As determined

Figure 7. Plotting the relative gel times for hemiaminal gels complexed with aluminum versus iron(III) shows clearly that iron is a more strongly retarding additive to hexahydrotriazine formation.

NMR studies of the gel of sample 1, where 1.94 equiv of aluminum chloride hydrate is added to the polypropylene glycol amine, lend evidence to a mechanism for the reversion of the gel after the addition of metal salt. After the addition of aluminum chloride hydrate to the gel, a number of peaks appear between 4.65 and 5.0 ppm in the 1H NMR (Figure 8). Peaks appear at 4.97, 4.90, 4.896, 4.80, 4.78, 4.74, and 4.73 ppm. The respective areas of these peaks are 0.19, 0.25, 0.14, 3.18, 0.89, 2.69, and 1.77. The most intense peaks appear at 4.80, 4.74, and 4.73 ppm. From the HSQC-TOCSY spectrum (Figure S-4) of sample 2 immediately after the addition of aluminum, the 1H NMR peaks at 4.80, 4.74, and 4.73 ppm correlate to 13C peaks at 82−89 ppm. The DEPT-135 experiment of the sample reveals that the 13 C peaks for 83−88 ppm all correspond to CH2 functional groups (Figure S-5). The downfield position of these peaks suggests their correspondence to R−NH−CH2−OH functionalities. This (Figure 8b) is evidence for the fragmentation of the hemiaminal condensation product through the addition of

Figure 6. Gelation times aluminum gel materials from varying amounts of aluminum added at 70 °C. The complex shear modulus loss factor is plotted with respect to time. The loss factor less than or equal to one describes the material in the gel state. 15388

DOI: 10.1021/jacs.7b07053 J. Am. Chem. Soc. 2017, 139, 15385−15391

Article

Journal of the American Chemical Society

The 14 ppm shift corresponds to the methyl group from the polypropylene glycol amine (Jeffamine T5000). The addition of aluminum to the gel sample shifts some of the 1H NMR signal from 1.11 pm downfield to 1.34−1.38 ppm. This is likely due to the donation of the nitrogen lone pair coordinating to the 3s orbital of aluminum chloride. There are also some peaks around 1.39 ppm, appearing as two doublets, which disappear after heating to 353 K. The integral area of the peaks from 1.34 to 1.38 increases from 8.06 to 9.65 after heating to 353 K, indicating further enrichment of this species with heat. Density functional theory (DFT) computations elucidate further and in greater detail, the thermodynamic observations on the dynamism of the gel−liquid transition. We chose to apply the well-tested M06-2x functional, especially suited to a balanced approach between main group and transition metal chemistry with an emphasis on quality energetics.21 Following the work of Jones and co-workers,22 the organic hemiaminal mechanism with R = methyl was remodeled at the M06-2x/6-31+g(d) level with COSMO solvation in water as implemented by the software NWchem.23 As we are primarily interested in ligation of these intermediates to cationic metals, we have chosen not to include explicit water molecules in our model and have avoided the consideration of various intermediates and transition states, which have been adequately dealt with in previous studies and have only marginal effects on the reaction. Scheme 3 presents the computationally derived reaction free energy diagram (kcal/mol, ΔG) along a generalized reaction

Figure 8. Region of the 1H NMR spectra from 5.10 to 4.30 ppm of the reaction product of paraformaldehyde with polypropylene glycol amine in DMF-d7. (a) Before the addition of aluminum chloride hydrate; (b) after the addition of aluminum chloride hydrate; (c) after heating for 14 h at 353 K with aluminum chloride. All spectra were collected at 303 K.

aluminum chloride hydrate. These peaks disappear after heating for 14 h at 353 K. The 1H NMR thus reveals the heating step eliminates many of the R−NH−CH2−OH functionalities which could be ascribed to intermediate condensation products. These shifts have been confirmed by 1H NMR modeling at the density functional level (Figure S-8) where large downfield shifts of over 1 ppm correspond to metal binding at N of adjacent proton bearing centers. Some new species that appear in the 1H NMR (Figure 9) immediately after the addition of aluminum chloride can be seen

Scheme 3. M06-2x/6-31+g(d) (COSMO, Water)//B3LYP/631+g(d) Reaction Free-Energy Coordinate (ΔG, kcal/mol) for the Different Intermediates Drawn in Scheme 1

coordinate. Each change in energy thus reflects the reaction energy necessary to proceed forward (or backward) from a given state to the next one. Only one reaction pathway is shown (through the methyl-dimethanolamine) while data for alternative reaction sequences can be found in the Supporting Information. In the absence of metal salt, the progression from formaldehyde and methyl amine (A and B, respectively) toward the final hexahydrotriazine product is consistently energetically downhill with a final total reaction free energy of −29.69 kcal/mol. We have assumed, as predicted by previous computational studies, that the rate-determining step is between intermediates L and N and represents a ring closure condensation/dehydration step to give the final product. This barrier has been predicted to be about 27 kcal/mol, and thus, the kinetic product of the process shown in Scheme 3 would likely accumulate at intermediate L prior to

Figure 9. Region of the 1H NMR spectra from 1.48 to 1.30 ppm of the reaction product of paraformaldehyde with polypropylene glycol amine in DMF-d7. (a) Before the addition of aluminum chloride hydrate; (b) after the addition of aluminum chloride hydrate; (c) after heating for 14 h at 353 K with aluminum chloride. All spectra were collected at 303 K.

as a cluster of doublets around 1.39 ppm with J couplings of 6.7 Hz. These peaks have an integral area of 3.14. The peaks are assigned to the methyl proton of the polypropylene glycol amine, alpha to the methyne group. The signal is shifted downfield due to binding between the nitrogen electron loan pairs to aluminum. Additional to these peaks is another multiplet of peaks further upfield at 1.38−1.34 ppm with an integral area of 8.06. According to the HSQC, these peaks correspond to carbon shift at 14 ppm. 15389

DOI: 10.1021/jacs.7b07053 J. Am. Chem. Soc. 2017, 139, 15385−15391

Article

Journal of the American Chemical Society

nearly −15 kcal/mol more stable. On the other end, complexation with N renders Fe(II) about 10 kcal/mol less stable. As a result, there is less contribution of these earlier/less cross-linked intermediates to the total material. This result is observed rheologically, as depicted in Figure 1, where it is shown that Fe(II) gels have a higher phase shift angle for the complex shear modulus. However, they are not liquids as in the case of the trivalent metals (illustrated in Figures 2 and 3). The presence of the nonequilibrium earlier intermediates (A−K in Scheme 2) is substantially higher when trivalent metals are complexed at room temperature. Ligands F and H did not form the expected complexes with Al(III) and Fe(II/III) (with the exception of [Fe(κ2-F)(H2O)3]3 (−19.28 kcal/mol stabilization as a bidentate ligand with flanking hydroxyl-M dative bonds). Both metals generally induce a barrierless transfer of the hydroxyl function from its position to the metal center creating a M−OH complex with an iminium ligand. The distinguishing difference between F and H to I and L is the accommodation by the metal of a full coordination sphere for the latter two. While the number of water molecules was held constant at three throughout for full comparison, it is possible that an additional water molecule might stabilize the metal to abstraction of the hydroxyl function in the case of F and H from the methanolamine moiety. However, these rearrangements point to an interesting aspect of the chemistry of these metal coordinated ligands−the metals facilitate dehydration and may play a strong role in the kinetics of the RDS ring closure that involves loss of water. To further explore this possibility transition-state calculations have been performed for the case of Al(III). The availability of the Al(III) ion allows a catalyzed reaction channel that transfers the hydroxide to the metal center (Scheme 4) yielding [Al((κ2-

overcoming the RDS and giving the final thermodynamic product, N. The metal chelation effects have been thoroughly modeled for both Al(III) and Fe(III) in the hydrate forms (chloride ions are assumed to be fully solvated and outer sphere anions) with three water molecules acting as ancillary ligands. Upon introduction of aluminum hydrate (Al(H2O)6), the free-energetic landscape of the organic intermediates (i.e., ligands) is altered significantly. Surprisingly, methylamine forms the most stable complex with Al(III) relative to the other available intermediates. For the purpose of direct comparison with the tridentate ligators (L and N), the aluminum complex with 3 equiv of methylamine was modeled [Al(CH3NH2)3(H2O)3]. This showed the aluminum(III) hydrate as having an exergonicity of 28.63 kcal/mol. Considering the required displacement of three water molecules in this process, the reaction can be described as enthalpically driven because the entropic factors are roughly balanced on each side (ΔS: ca. −5 kcal/mol at 298.15). This stability is about twice that for the complexation with hexahydrotriazine (N) at −18.10 kcal/mol (κ3-N(H2O)3Al(III)). The main difference is the strongly disfavored entropic requirement to render the relatively dynamic hexahydrotriazine (triazinane) ring as a rigid ligand, making the final complex less stable (ΔS: ca. +68 kcal/mol at 298.15 K). Importantly, intermediate I achieves a smaller freeenergy stabilization when complexed with aluminum ((κ3I(H2O)3Al(III))) (−6.92 and −5.59 kcal/mol, respectively), while L (κ3-L(H2O)3Al(III)) is exceptionally strongly complexed to aluminum by −17.92 kcal/mol reaching the same magnitude as for N. With respect to the complex geometries of L and N (see the Supporting Information), both form a similar complex with three nitrogen atoms able to bind the aluminum cation, whereas I forms two N-M bonds with a third ligand position occupied by the hydroxyl function, which is a less effective dative bond (all structures available in the Supporting Information). F and H are only able to establish two dative bonds with the metal centers, which impacts their stability (see below). The final effect of these metal−ligand interactions on the energetic landscape is to shift the equilibrium of the ligands toward the methylamine starting materials. This is made clear by comparison of the final energy of the methylamine complex with that of the complex of L, where L is a ca. 4−5 kcal/mol more stabilized. The uncomplexed ligand, L, in the absence of metal is −16.01 kcal/mol. Fe(III) complexes were computed in the high-spin sextet state also assuming an inner sphere of water molecules. Ligation to this metal cation is more stabilizing than for Al(III), in general, where the complexes with N ([Feκ3-N(H2O)3]3+), L ([Fe(κ3-L)(H2O)3]3+), and I ([Fe(κ3-I)(H2O)3]3+) are stabilized by −25.28, −27.22, and −11.35 kcal/mol, respectively. Complexation with 3 equiv of methylamine ([Fe(CH3NH2)3(H2O)3]3+) releases −36.32 kcal/mol. Compared to [Fe(κ3-L)(H2O)3]3+, the overall exergonicity of [Fe(CH3NH2)3(H2O)3]3+) is about 7 kcal/mol higher, while compared to ([Fe(κ3-I)(H2O)3]3+ it is significantly stabilized compared to the analogous comparison with Al(III). Thus, Fe(III) is predicted to turn the gel to a liquid with a greater thermodynamic driving forces compared to Al(III) and away from intermediate I and its complex. Finally, Fe(II) complexes were modeled in their high-spin quintet states to establish the reason for the absence of a phasechange effect when introducing this metal cation. As illustrated in Scheme 3, Fe(II) does not stabilize the intermediates as strongly as either Al(III) or Fe(III). Comparing Fe(II) to Fe(III) when complexed with three equivalents of methylamine, the latter is

Scheme 4. Transition State of the Aluminum Catalyzed Ring Closure Reaction to the Hexahydrotriazine Structure

M)(OH)(H2O)3]3+ from [Al(κ3-L)(H2O)3]3+. No transition state for this process could be localized and the free-energy of reaction is nearly isoenergetic at −0.55 kcal/mol, favoring the metal hydroxide. For complex M, the metal-hydroxide functions in a similar way to the water bridge found for the organic reaction pathway as it swings like a hinge between hydrogen bonding interaction with the methylene proton in M and later with the 15390

DOI: 10.1021/jacs.7b07053 J. Am. Chem. Soc. 2017, 139, 15385−15391

Article

Journal of the American Chemical Society secondary amino in [Al(NH)(OH)(H2O)3]3+. The reaction free energy for M to NH is −6.92 kcal/mol with the transition state between these states [Al(M-NH)(OH)(H2O)3]3+ presents are barrier height of 22.0 kcal/mol. Comparing this value to that for the pure organic mechanism calculated at a comparable level of theory shows significant stabilization on the order of 5 kcal/mol for the metal catalyzed case presented here. The structure of the transition-state shows a relatively weak imaginary frequency whose principle components lies along the forming cyclic bond with a noticeable components of increasing interaction between the hydroxide and the as formed tertiary ammonium NH bond. Loss of water yields [Al((κ3-N)(H2O)3]3+, bringing the thermodynamic driving force for this process to −13.3 kcal/mol.

(3) Wei, Z.; Yang, J. H.; Zhou, J.; Xu, F.; Zrínyi, M.; Dussault, P. H.; Osada, Y.; Chen, Y. M. Chem. Soc. Rev. 2014, 43, 8114−8131. (4) Lehn, J.-M. Chem. Soc. Rev. 2007, 36, 151−160. (5) Zhang, Y.; Barboiu, M. Chem. Rev. 2016, 116, 809−834. (6) (a) Skene, W. G.; Lehn, J.-M. Proc. Natl. Acad. Sci. U. S. A. 2004, 101, 8270−8275. (b) Schaeffer, G.; Buhler, E.; Candau, S. J.; Lehn, J.-M. Macromolecules 2013, 46, 5664−5671. (7) Belowich, M. E.; Stoddart, J. F. Chem. Soc. Rev. 2012, 41, 2003− 2024. (8) Fox, C.; ter Hurrne, G. M.; Wojtecki, R. J.; Jones, G. O.; Horn, H. W.; Meijer, E. W.; Frank, C. W.; Hedrick, J. L.; Garcia, J. M. Nat. Commun. 2015, 6, 7417. (9) Reutenauer, P.; Buhler, E.; Boul, P. J.; Candau, S. J.; Lehn, J. M. Chem. - Eur. J. 2009, 15, 1893−1900. (10) (a) Deng, G.; Li, F.; Yu, H.; Liu, F.; Liu, C.; Sun, W.; Jiang, H.; Chen, Y. ACS Macro Lett. 2012, 1, 275−279. (b) Ding, F.; Wu, S.; Wang, S.; Xiong, Y.; Li, Y.; Li, B.; Deng, H.; Du, Y.; Xiao, L.; Shi, X. Soft Matter 2015, 11, 3971−3976. (c) Yoon, J. A.; Kamada, J.; Koynov, K.; Mohin, J.; Nicolaÿ, R.; Zhang, Y.; Balazs, A. C.; Kowalewski; Matyjaszewski, K. Macromolecules 2012, 45 (1), 142−149. (11) Orrillo, A. G.; Escalante, A. M.; Furlan, R. L. E. Chem. - Eur. J. 2016, 22, 6746−6749. (12) Cromwell, O. R.; Chung, J.; Guan, Z. J. Am. Chem. Soc. 2015, 137, 6492−6495. (13) Lu, Y.-X; Guan, Z. J. Am. Chem. Soc. 2012, 134, 14226−14231. (14) (a) McConnell, A. J.; Wood, C. S.; Neelakandan, P. P.; Nitschke, J. R. Chem. Rev. 2015, 115, 7729−7793. (b) Enke, M.; Bode, S.; Vitz, J.; Schacher, F. H.; Harrington, M. J.; Hager, M. D.; Schubert, U. S. Polymer 2015, 69, 274−282. (15) (a) Fox, C. H.; ter Hurrne, G. M.; Wojtecki, R. J.; Jones, G. O.; Horn, H. W.; Meijer, E. W.; Frank, C. W.; Hedrick, J. L.; Garcia, J. M. Nat. Commun. 2015, 6, 7417. (b) García, J. M.; Jones, G. O.; Virwani, K.; McCloskey, B. D.; Boday, D. J.; ter Huurne, G. M.; Horn, H. W.; Coady, D. J.; Bintaleb, A. M.; Alabdulrahman, A. M. S.; Alsewailem, F.; Almegren, H. A. A.; Hedrick, J. L. Science 2014, 344, 732−735. (16) Wojtecki, R. J.; Jones, G. O.; Yuen, A. Y.; Chin, W.; Boday, D. J.; Nelson, A.; García, J. M.; Yang, Y. Y.; Hedrick, J. L. J. Am. Chem. Soc. 2015, 137, 14248−14251. (17) Clegg, J. K.; Harrowfield, J.; Kim, Y.; Lee, Y. H.; Lehn, J.-M.; Lim, W. T.; Thuéry, P. Dalton Trans. 2012, 41, 4335−4357. (18) Kovaříček, P.; Lehn, J.-M. J. Am. Chem. Soc. 2012, 134, 9446− 9455. (19) (a) Lee, Y.-M; Kim, E.-S.; Kim, H.-J.; Choi, H. J.; Kim, Y.-I.; Kang, S. K.; Choi, S.-N. Dalton Transactions 2009, 126−133. (b) Bréfuel, N.; Lepetit, C.; Shova, S.; Dahan, F.; Tuchagues, J.-P. Inorg. Chem. 2005, 44, 8916−8928. (20) (a) Jo, H. H.; Edupuganti, R.; You, L.; Dalby, K. N.; Anslyn, E.V. I Chem. Sci. 2015, 6, 158−164. (b) Zhou, Y.; Ren, Y.; Zhang, L.; You, L.; Yuan, Y.; Anslyn, E. Tetrahedron 2015, 71, 3515−3521. (21) (a) Cramer, C. J.; Truhlar, D. G. Phys. Chem. Chem. Phys. 2009, 11, 10757. (b) Zhao, Y.; Truhlar, D. G. Theor. Chem. Acc. 2008, 120, 215−241. (22) Jones, G. O.; Garcia, J. M.; Horn, H. W.; Hedrick, J. L. Org. Lett. 2014, 16, 5502−5505. (23) Valiev, M.; Bylaska, E. J.; Govind, N.; Kowalski, K.; Straatsma, T. P.; van Dam, H. J. J.; Wang, D.; Nieplocha, J.; Apra, E.; Windus, T. L.; de Jong, W. A. Comput. Phys. Commun. 2010, 181, 1477−1489.

■

CONCLUSIONS Through these studies, hemiaminal gels have been observed to transform into liquids through the addition of trivalent metals, such as aluminum or iron. Trivalent metals revert the gels back in the direction of the starting materials. The resulting organometallic liquid can be transformed into a different gel with unique physical properties through heating. The gel time for this transformation can be controlled and modified as needed by adjusting the amount of trivalent metal in the liquid. Aluminum chloride acts as both an accelerator to gel formation, catalyzing the formation of the hexahydrotriazine complex, and as a retarder by stabilizing the reactants prior to stabilizing the hexahydrotriazine product. Ferric ammonium sulfate does not accelerate the formation of the hexahydrotriazine as effectively as the aluminum chloride but the addition of increasing amounts of ferric ammonium sulfate to the solution also leads to a decrease in the rate of hexahydrotriazine formation. These mechanisms have been verified and further elucidated with DFT at the M062x/6-31+g(d) level with COSMO solvation in water. The dynamics described can be utilized to access triggered release of molecular cargo from polymer gels of these CDMs.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.7b07053. Experimental and computational details (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*[email protected] ORCID

Peter J. Boul: 0000-0002-7082-086X Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS The authors would like to thank Aramco for permission to publish this work. REFERENCES

(1) (a) Buchs, B.; Godin, G.; Trachsel, A.; de Saint Laumer, J.-Y.; Lehn, J.-M.; Herrmann, A. Eur. J. Org. Chem. 2011, 2011, 681−695. (b) Herrmann, A.; Giuseppone, N.; Lehn, J.-M. Chem. - Eur. J. 2009, 15, 117−124. (2) de Hatten, X.; Bell, N.; Yufa, N.; Christmann, G.; Nitschke, J. R. J. Am. Chem. Soc. 2011, 133, 3158−3164. 15391

DOI: 10.1021/jacs.7b07053 J. Am. Chem. Soc. 2017, 139, 15385−15391