Isabell S. R.
Karmel
a,
Maxim
Khononov
a,
Matthias
Tamm
*b and
Moris S.
Eisen
*a
aSchulich Faculty of Chemistry, Institute of Catalysis Science and Technology, Technion – Israel Institute of Technology, Technion City, Haifa, 32000 Israel. E-mail: chmoris@tx.technion.ac.il
bInstitut für Anorganische und Analytische Chemie, Technische Universität Braunschweig, Hagenring 30, 38106 Braunschweig, Germany. E-mail: m.tamm@tu-bs.de
First published on 3rd September 2015
The ring-opening polymerization (ROP) of the cyclic ester ε-caprolactone was studied using the uranium(IV) complexes [(ImDippN)2U(NMeEt)2] (3), [(C5Me5)2U(NMe2)2] (4) and [(C5Me5)2U(NCMePh)2] (5) as initiators. While the bis(imidazolin-2-iminato) complex 3 displayed a surprisingly high catalytic activity of 1.2 × 107 g (PCL) mol−1 h−1 at room temperature, compounds 4 and 5 exhibited lower catalytic activities even at 90 °C. The activity of the uranium complex 3 was further compared to the imidazolin-2-iminato uranium(IV) complexes [(ImtBuN)4U] (1) and [(ImMesN)3U(NMeEt)] (2), which display catalytic activities of 7.9 × 103 g (PCL) mol−1 h−1, and 5.3 × 103 g (PCL) mol−1 h−1, respectively at an elevated temperature of 90 °C. In order to shed light on the operative mechanisms, kinetic studies were carried out with complexes 3–5.
The great interest in this polymer over the last eight decades can be attributed to its low melting point (59–64 °C), high solubility in a large variety of organic solvents, exceptional miscibility, and mechanical compatibility with a large number of polymers, as well as its biodegradability and biocompatibility.6 In addition, the extensive research carried out during the 1970s and 1980s in the field of biodegradable polymers led to interesting correlations between the molecular weight of the polymer, its biodegradation conditions and degradation kinetics.7 Therefore, the application of PCL in the field of biomedicine is widespread and includes the scaffolds in tissue engineering,8 long-term drug delivery systems7b and contraceptive delivery systems.9 Additionally, PCL is used as a packaging material,9 in microelectronics10 and in adhesives.7e The availability of the monomer ε-caprolactone, and the wide applicability of the corresponding polyester renders PCL an environmentally friendly, low-cost polymer with an increasing demand over the last two decades.11
The polymerization of ε-caprolactone has been investigated with a variety of main group5,12 and transition metals,5,13 as well as with lanthanide catalysts,5,14 affording insights into the mechanistic details, the thermodynamic and kinetic parameters as well as the control of the molecular weight and crystallinity of the resulting PCL. Despite the large variety of metal catalysts examined in the ROP of ε-caprolactone, only a few examples involving actinide-based catalysts can be found in the literature,15 which can be attributed to the high oxophilicity of these elements. The oxophilic nature should result in a decrease in catalytic activity towards oxygen-containing substrates, since a reaction between the actinide centre and the oxygen atom of the substrate can occur, leading to the formation of thermodynamically stable, catalytically inactive actinide-oxo species as reported by Marks et al.16 Since the low catalytic activity of the early actinides towards oxygen-containing substrates is attributed to their high electrophilicity, decreasing the electrophilic nature of the metal should lead to an increased reactivity towards oxygen-containing molecules such as cyclic esters. Our method of choice for making the actinide centre less electrophilic is based on using highly nucleophilic and strongly electron-donating ligands, i.e. the imidazolin-2-iminato motif (ImRN−), which is obtained by the deprotonation of imidazolin-2-imine (ImRNH). This strongly basic and highly nucleophilic ligand class can be considered as 2σ, 4π electron donors towards early transition metals and metals in high oxidation states and therefore as monodentate isolobal analogues to the widely used cyclopentadienyl ligand (Scheme 2).17 Accordingly, the resulting transition metal and lanthanide metal complexes with Ln (Ln = Sc, Y, Gd, Lu),18 Ti,19 Zr,20 V,21 Mo,22 W,22,23 and Re24 usually exhibit short M–N bonds and large, almost linear M–N–C angles.
Recently, we reported the syntheses and structures of the imidazolin-2-iminato uranium(IV) complexes [(ImtBuN)4U] (1), [(ImMesN)3U(NMeEt)] (2) and [(ImDippN)2U(NMeEt)2] (3).25 This series of complexes was obtained by an acid–base reaction between the homoleptic [U(NMeEt)4] and neutral imidazolin-2-imine ImRNH, which furnished the respective uranium complexes in dependence of the steric demand of the R substituent on the imidazolin-2-imine ligand (Scheme 3). Furthermore, we reported the selective preparation of mono(imidazolin-2-iminato) thorium(IV) and uranium(IV) complexes by a selective protonolysis reaction of actinide metallacycles with neutral imidazolin-2-imines.26 The uranium complexes 1–3 display short U–N bond distances (2.174(11)–2.177(11) Å) and almost linear U–N–C angles (165.0(4)°–172.3(4)°), suggesting a higher bond order of the U–N bond.25
Scheme 3 Synthesis of imidazolin-2-iminato uranium(IV) complexes.25 |
Herein, we report the reactivity of these complexes in the ROP of ε-caprolactone, giving rise to mechanistic, thermodynamic and kinetic details. Moreover, we compare the reactivity and kinetics of the imidazolin-2-iminato complexes 1–3 to two analogous cyclopentadienyl uranium(IV) complexes Cp*2U(NMe2)2427 and Cp*2U(NCMePh)25 (Fig. 1),28 focusing on the differences in reactivities, mechanisms and rates, despite the isolobal analogy between the respective complexes.
Entry | Time (min) | Activity (g mol−1 h−1) | M w (dalton) | PDI | Yield (%) |
---|---|---|---|---|---|
a Polymerization conditions: 5 mL of toluene, r.t., 0.216 μmol of 3, complex 3/ε-CL: 1/60000. b Conditions as in “a” but at 90 °C. c Carried out in THF. d Polymer insoluble in THF; no GPC analysis possible. e The relative calibration of the Mn values was done using polystyrene standards; the Mn values were multiplied by a factor of 0.56 (Mark–Houwink coefficient) and correlated to the actual PCL values.30 | |||||
1 | 10 | 1.2 × 107 | 21 970 | 1.86 | 28 |
2 | 30 | 1.1 × 107 | 23 680 | 1.86 | 78 |
3 | 60 | 6.8 × 106 | 30 660 | 2.54 | 99 |
4 | 120 | 3.4 × 106 | 223 660 | 2.51 | 99 |
5 | 300 | 1.4 × 106 | 327 860 | 3.59 | 99 |
6 | 30 | 1.3 × 107 | 355 280 | 2.36 | 98 |
7 | 720 | 5.6 × 105 | 99 | ||
8 | 60 | 2.2 × 106 | 37 890 | 2.03 | 32 |
9 | 120 | 3.3 × 106 | 60 110 | 2.29 | 95 |
The yield of the obtained polymer increases linearly with time until the monomer is fully consumed after ~60 minutes (Fig. 2), suggesting a living polymerization (expected PDI = 1.0); however, the molecular weights of the polymers do not increase linearly. In addition, the activity of the catalyst remains constant until all the monomer is polymerized. Additional polymerization time reduces the activity almost linearly since there is no additional monomer. Interestingly, after additional time, the molecular weight of the polymer clearly increases, indicating that the complex is able to continue performing a transesterification, which causes also an increase in the PDI (entry 5, Table 1). Hence, the polydispersity of the obtained polymers at the beginning of the polymerization is close to 2, indicative of a single site polymerization mechanism. These results suggest that the polymerization initiated by complex 3 is in a rapid competition with a chain transfer mechanism (transesterification) between the catalytically active species. The transesterification reactions of this type have been previously observed in the ROP of lactides and lactones, as well as in the co-polymerization of these monomers.29
Moreover, the reinsertion of the polymer chain obtained after 720 minutes leads to a polymer with an ultrahigh molecular weight, which is not soluble. When the reaction is carried out at higher temperatures, there is an increase in the activity and in the molecular weight of the polymer. A variation of the solvent to THF resulted in lower activities, suggesting competitive coordination of THF to the active catalytic species, which hampers the coordination of the substrate, ε-caprolactone.
For investigating the mechanism of the polymerization reaction mediated by complex 3, we performed kinetic measurements, which exhibit a first order dependence on ε-caprolactone and the catalyst (eqn (1), Fig. 3).
(1) |
The thermodynamic parameters were determined from the Arrhenius plot (Ea = 12.8(5) kcal mol−1) and the Eyring plot (ΔS‡ = −33.9(8) cal mol−1 K−1, ΔH‡ = 12.2(8) kcal mol−1) which is presented in Fig. 4. A plausible mechanism for the polymerization of ε-caprolactone is shown in Scheme 4. In order to determine, whether both amido groups are active in the polymerization, we performed NMR experiments with stoichiometric amounts of the monomer, which led to the observation that two equivalents of free amines were released per mole of catalyst. After the protonolysis step, the uranium-alkoxo-caprolactonate intermediate B undergoes a reaction with an incoming caprolactone monomer, leading to the open chain intermediate D, which can insert further monomers into the growing polymer chain, leading to the growing polymer chain E. The polymerization is terminated by an additional equivalent of the monomer, ε-caprolactone, leading to the formation of a polymer with caprolactonyl end-group F (see the ESI†) and regeneration of the active catalyst A (Scheme 4).
Scheme 4 Plausible mechanism for the ROP of ε-caprolactone mediated by complex 3. The second NMeEt unit has been omitted for clarity. |
The large discrepancy between the activity of the isolobal cyclopentadienyl uranium(IV) complex Cp*2UMe2 (ref. 15) and complex 3 towards ε-caprolactone raised the question, whether the high activity of 3 could be attributed to the replacement of the cyclopentadienyl moiety by imidazolin-2-iminato ligands or to the replacement of the methyl ligands by amido groups. Therefore, we synthesized the respective isolobal complex Cp*2U(NMe2)2 (4),27 and compared the kinetic data and reactivity with 3. The polymerization results are shown in Table 2.
Entry | Time (min) | Activity (g mol−1 h−1) | M w (dalton) | PDI | Yield (%) |
---|---|---|---|---|---|
a Polymerization conditions: 5 mL of toluene, 90 °C, 4.08 μmol of complex 4, complex 4/ε-CL:1/1000. b Conditions as in “a” but at r.t. c The relative calibration of the Mn values was done using polystyrene standards; the Mn values were multiplied by a factor of 0.56 (Mark–Houwink coefficient) and correlated to the actual PCL values.30 | |||||
1 | 30 | 4.6 × 103 | 41 040 | 2.78 | 2 |
2 | 60 | 3.9 × 104 | 58 680 | 1.39 | 34 |
3 | 120 | 2.6 × 104 | 73 520 | 2.60 | 45 |
4 | 180 | 3.2 × 104 | 97 190 | 1.65 | 85 |
5 | 300 | 2.2 × 104 | 99 840 | 1.48 | 98 |
6 | 840 | 7.9 × 103 | 148 540 | 1.92 | 98 |
7 | 840 | 978 | 9000 | 1.32 | 12 |
In comparison to complex 3, the cyclopentadienyl analogue 4 displays lower activity at 90 °C and almost no activity at room temperature. The molecular weights of the polymers obtained are lower than those obtained for complex 3. An increase in the average molecular weight of the polymer can be observed as a function of time (Fig. 5), and the polydispersity values indicate a single-site polymerization process. As with complex 3, when the polymerization is complete, additional time reduces linearly the activity; thus, the catalyst is able to perform a transesterification, as indicated by the larger molecular weight and increased PDI.
The kinetic measurements performed with complex 4 show a first order dependence on monomer and catalyst (eqn (2); Fig. 6).
(2) |
NMR experiments with stoichiometric amounts of the substrate, confirmed an intermolecular mechanism, initiated by the amido ligands (Scheme 5). However, the metal centre does not react with the acidic hydrogen atom in the α-position to the carbonyl leading to the release, of free amine. Instead, the uranium centre reacts with the oxygen atom of the carbonyl group over intermediate B, (Scheme 5) and nucleophilic attack at the carbonyl carbon atom leads to an amido end-group in the first polymer chain F, generating catalytically active uranium-alkoxocaprolate species G. Intermediate G can now react with a further equivalent of ε-caprolactone, leading to the formation of the open-chain intermediate H. After insertion of additional ε-caprolactone monomers into the growing polymer chain of H, the reaction is terminated by an incoming monomer, yielding a polymer with a caprolactonyl end-group (I) (see the ESI†) and regenerating the active catalyst G.
Scheme 5 Plausible mechanism for the ROP of ε-caprolactone mediated by complex 4. The second NMe2 unit has been omitted for clarity. |
The energy of activation for the ROP of ε-caprolactone mediated by complex 4 was determined from the Arrhenius plot (Fig. 7) with a value of Ea = 19.0(7) kcal mol−1. In comparison to the activation barrier of the polymerization catalysed with complex 3, compound 4 has a much higher barrier of activation, which explains the high temperature required for the polymerization. The entropy of activation is comparable and just slightly larger than for complex 3 (ΔS‡ = −27.8(8) cal mol−1 K−1).
The structural and electronic similarity of the imidazolin-2-iminato ligands to the ketimido ligand reported as ancillary ligands in actinide complexes by Kiplinger et al.28 should also result in a comparable reactivity. Due the π-character attributed to the U–N bond in the latter, the ketimido ligand similar to imidazolin-2-iminato ligands should not initiate the polymerization by promoting a nucleophilic attack on the substrate, in contrast to the amido moieties in 3 and 4. Therefore, complex 5 should in theory exhibit a very low activity in the ROP of ε-caprolactone, if the U–N bond displays a higher bond order. Therefore, the uranium(IV) bis(ketimido) complex 5 was synthesized and its reactivity towards the ROP of ε-caprolactone was studied. The results are summarized in Table 3. Similar to 4, complex 5 showed only a catalytic activity at higher temperatures. The activities obtained were higher than those of complex 4, but lower than those found for 3. The isolated polyester exhibits high molecular weights which increase over time, and narrow polydispersities (~2.0) indicating a single-site catalyst mechanism. When the polymerization was carried out at room temperature or in THF, no product was obtained. For elucidating the mechanism of this reaction, an NMR scale reaction with stoichiometric amounts of ε-caprolactone was carried out. Neither the ketimido ligands nor the cyclopentadienyl ligands could be observed as a free ketimine, or cyclopentadiene, respectively, which suggests a Lewis acid-catalysed mechanism.
Entry | Time (min) | Activity (g mol−1 h−1) | M w (dalton) | PDI | Yield (%) |
---|---|---|---|---|---|
a Polymerization conditions: 5 mL of toluene, 90 °C, 3.36 μmol of complex 5; complex 5/ε-CL 1/1000. b Conditions as in “a” but at r.t. c Couldn't be determined due to low conversion. d The relative calibration of the Mn values was done using polystyrene standards; the Mn values were multiplied by a factor of 0.56 (Mark–Houwink coefficient) and correlated to the actual PCL values.30 | |||||
1 | 30 | 4.6 × 103 | <2 | ||
2 | 60 | 6.6 × 104 | 84 140 | 1.87 | 58 |
3 | 120 | 5.1 × 104 | 156 580 | 1.80 | 89 |
4 | 300 | 2.1 × 104 | 270 050 | 1.72 | 94 |
5 | 840 | 7.9 × 103 | 108 700 | 2.54 | 97 |
6 | 840 | 0 | <1 |
The mechanism presented in Scheme 6 involves an activation of the monomer by the Lewis acidic metal complex, which was previously observed with other main group and transition metals, followed by a nucleophilic attack of an incoming monomer unit B, leading to the growing polymer chain D.30 The polymerization process is terminated by an additional equivalent of ε-caprolactone, leading to the formation of a polymer with a caprolactonyl end-group (E) (see the ESI†) under regeneration of the active catalyst A.
Kinetic and thermodynamic NMR studies have shown a first order dependence on the monomer and the catalyst (eqn (3), Fig. 8).
(3) |
The energy of activation (Ea = 23.55 kcal mol−1) was determined as described previously from the Arrhenius plot (Fig. 9); the enthalpy of activation (ΔH‡ = 22.8(5) kcal mol−1) and the entropy of activation (ΔS‡ = −15.0(9) cal mol−1 K−1) were determined from the Eyring plot. The large value for the energy of activation is reflected in the high temperatures required and provides an explanation for the lack of reactivity at room temperature.
For the kinetic 1H NMR studies, a J-Young NMR tube was loaded with the respective amount of catalyst from a stock solution, ε-caprolactone and toluene-d8 inside the glove box, then the tube was subsequently sealed, and the reaction mixture was frozen at the liquid nitrogen temperature until the start of the 1H NMR measurements. The sample was heated (if required) in the NMR spectrometer.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5cy01162k |
This journal is © The Royal Society of Chemistry 2015 |