Indium Catalysts for Low-Pressure CO2/Epoxide Ring-Opening Copolymerization: Evidence for a Mononuclear Mechanism?

The alternating copolymerization of CO2/epoxides is a useful means to incorporate high levels of carbon dioxide into polymers. The reaction is generally proposed to occur by bimetallic or bicomponent pathways. Here, the first indium catalysts are presented, which are proposed to operate by a distinct mononuclear pathway. The most active and selective catalysts are phosphasalen complexes, which feature ligands comprising two iminophosphoranes linked to sterically hindered ortho-phenolates. The catalysts are active at 1 bar pressure of carbon dioxide and are most effective without any cocatalyst. They show low-pressure activity (1 bar pressure) and yield polymer with high carbonate linkage selectivity (>99%) and isoselectivity ( Pm > 70%). Using these complexes, it is also possible to isolate and characterize key catalytic intermediates, including the propagating indium alkoxide and carbonate complexes that are rarely studied. The catalysts are mononuclear under polymerization conditions, and the key intermediates show different coordination geometries: the alkoxide complex is pentacoordinate, while the carbonate is hexacoordinate. Kinetic analyses reveal a first-order dependence on catalyst concentration and are zero-order in carbon dioxide pressure; these findings together with in situ spectroscopic studies underpin the mononuclear pathway. More generally, this research highlights the future opportunity for other homogeneous catalysts, featuring larger ionic radius metals and new ligands, to operate by mononuclear mechanisms.


Experimental section
All reactions were conducted under an inert atmosphere of dry nitrogen, either using standard Schlenk line or glovebox techniques. All solvents and reagents were obtained from commercial sources (Aldrich and Merck) and used as received, unless stated otherwise. Tetrahydrofuran, hexane and toluene were distilled from sodium/benzophenone, under dry argon ([H2O] < 5 ppm).
Dichloromethane and diethyl ether, used for ligands syntheses, were taken directly from an MBraun MB-SPS-800 solvent purification system. All dry solvents and reagents were stored under nitrogen and degassed by several freeze-pump-thaw cycles. Cyclohexeneoxide (CHO) was dried overnight over CaH2, fractionally distilled and degassed. KO t Bu was sublimed prior to use. 1 All other reagents used as received from suppliers.

NMR Spectroscopy
1 H (400.2 MHz) and 31 P (162.0 MHz) NMR spectra were collected using a Bruker Advance III HD nanobay NMR equipped with a 9.4T magnet. 13 C NMR, COSY, HMBC and HSQC NMR spectra were collected with a Bruker Avance NMR equipped with a 11.75T magnet and a 13 C (125.8 MHz) detect cryoprobe.

DOSY
The Diffusion-Ordered Spectroscopy (DOSY) NMR experiments were performed, at 298 K, on a Bruker 500 AVANCE III NMR spectrometer operating, at a frequency of 499.9 MHz for proton resonance and equipped with a z-gradient bbfo/5mm tuneable "SmartProbe" TM probe and a GRASP II gradient spectroscopy accessory providing a maximum gradient output of 53.5 G/cm (5.35G/cmA).
Diffusion ordered NMR data was acquired using the Bruker pulse program ledbpgp2s with a spectral width of 5500Hz (centred on 4.5 ppm) and 32768 data points. A relaxation delay of 12 s was employed along with a diffusion time (large delta) of 100 ms and a longitudinal eddy current delay (LED) of 5 ms. Bipolar gradients pulses (little delta/2) of 2.2 ms and homospoil gradient pulses of 0.6 ms were used. The gradient strengths of the 2 homospoil pulses were -17.13% and -13.17%. 32 experiments were collected with the bipolar gradient strength, initially at 5% (1 st experiment), linearly increased to 95% (32 nd experiment). All gradient pulses were smooth-square shaped (SMSQ10.100) and after each application a recovery delay of 200 µs used. The experiment was run with 24 scans per increment, employing one stimulated echo with two spoiling gradients. DOSY plots were generated by using the DOSY processing module of TopSpin. Diffusion coefficients were calculated by fitting intensity data to the Stejskal-Tanner expression. 2 (BDI)Zn[N(SiMe3)2], 643.4 g mol -1 , Table S1}. From the diffusion coefficients of the external standards, linear calibration graphs were obtained by plotting logD vs logMW (Graph S1). Following DOSY analysis of the product, the diffusion coefficient obtained for the signals corresponding to the product allowed an estimate of the MW of the species present in solution. Table S1. Diffusion coefficients of standards in d8-THF solution compared to their molecular weight.

S7
High pressure copolymerization studies were performed in a 25 mL Parr 5500 HP Compact Reactor.
Research grade carbon dioxide (99.999 %) was purified by passing through two high pressure drying columns, connected in series (VICI,Thames Restek).

In situ ATR-IR measurements
In situ ATR-IR measurements were performed on a Mettler-Toledo ReactIR ic.10 spectrometer equipped with a MCT detector and a silver halide DiComp probe.

Typical Polymerization Procedure at 1 bar of CO2:
In a glove box, the indium catalyst (1 equiv.) was dissolved in CHO (1 mL, 1000 equiv.), under a nitrogen atmosphere in a Schlenk tube with a magnetic stirrer bar. The Schlenk tube was sealed, brought outside of the glove box and connected to a vac-CO2 line. The reaction mixture was subjected to three cycles of vac-CO2 fill and then heated to the desired temperature (60 o C or 80 o C), under 1 bar of CO2 pressure, for the desired time (24 h or 48 h). The crude polymer was isolated by evaporation of the excess CHO, under reduced pressure. Polymers were dissolved in THF (1 mL), then purified by precipitation using pentane (2 x 5 mL) to yield white powders.

Typical Polymerization Procedure at high CO2 pressure:
In a glove box, the indium catalyst (1 equiv.) was dissolved in CHO (6 mL, 1000 equiv.), under a nitrogen atmosphere and added into a 25 mL Paar reactor (equipped with an overhead mechanical stirrer). The reactor was sealed, brought outside of the glove box and connected to a high pressure CO2 line. The reactor was heated to 80 o C and pressurized. Polymerizations were stirred for 24 h, at 80 o C, at 10-40 bar of CO2. After the allotted time period, the reactor was cooled to ambient temperature over 5 min and depressurized. The crude polymer was isolated by evaporation of the S8 excess CHO, under reduced pressure. The crude polymer was dissolved in THF (5 mL) then purified by precipitation from pentane (2 x 20 mL) to yield a white powder.

In-situ ATR-IR monitoring of CO2/CHO ROCOP, at 1 bar of CO2
The ATR-IR probe was inserted into a dry Schlenk tube and dried overnight, under vacuum. In a glove box, the indium catalyst (1 equiv.) was dissolved in CHO (1.5 mL, 500-1200 equiv.) and the solution added to a dry Schlenk tube, equipped with a magnetic stirrer bar. The Schlenk tube was sealed, removed from the glove box and connected to a vac-CO2-N2 line, where the reaction mixture was subjected to three cycles of vac-CO2 fill. The IR probe was transferred from the empty Schlenk tube to the one containing the reaction mixture, under a dynamic flow of CO2 gas. The IR probe was immersed in the reaction solution and the Schlenk tube was immersed in a pre-heated oil bath at 80 o C. As soon as the Schlenk tube was immersed in the oil-bath, data acquisition was started (at a rate of 1 scan/15 min). The reaction was maintained at 80 o C, under 1 bar of CO2, for the desired time (24 h or 48 h). When complete, an aliquot was removed and used to determine the polymer conversion (by 1 H NMR spectroscopy). The crude polymer was isolated by evaporation of the excess CHO, under reduced pressure. The crude polymer was dissolved in THF (1 mL) then purified by precipitation using pentane (2 x 5 mL) to yield a white powder.

Kinetic analyses
The IR data acquired in the above experiments was used to determine initial rates. A brief outline of the methods used is provided and further information is available in Figs. S61, S78 and S79 (which show the plots of initial rates vs. time). The initial rates were monitored using in situ ATR-FTIR spectroscopy, by analysis of two IR vibrational modes: the C=O stretch (1787-1731 cm -1 ) and the C-O stretch (O-C=O = 1014 cm -1 ). An example of the data treatment is provided below:

Monitoring of absorption at 1789 cm -1 and [In] = 20.24 mM.
First the absorption vs. time data was collected, with absorption corresponding to the raw instrument data. Subsequently, the absorption data was normalized

(Normalized Absorption)t= (Absorption)t -(Absorption)i
The monomer conversion was determined from the aliquot analysed by 1 H NMR spectroscopy (at time = F). The absorption data was converted to conversion vs. time data using the following relationship: The initial rate of the polymerization was determined by plotting [PCHC] (over the conversion range =0 -10 %) vs. time. A linear fit was applied to the data and the gradient corresponds to the initial rate.
kobs was determined by plotting [cat] (mM) against υ (mM.min -1 ) and a linear fit was applied. The gradient is labelled, kobs (where kobs relates to kp according to the equation shown below):

In-situ ATR-IR monitoring of the CO2 insertion reaction
The ATR-IR probe was dried overnight, under reduced pressure, in a dry Schlenk tube. In a glove box,

S10
Compound S1: At 0 o C, N-bromosuccinimide (1 equiv.) was added to a solution of the desired phenol (5 g 3                S28 Figure S17. DOSY NMR analysis of LInCl in d8-THF (298 K). Determined by comparison of the integrals of signals arising from the methylene protons in the 1 H NMR spectra due to copolymer carbonate linkages (δ = 4.65 pmm) and copolymer ether linkages (only for % polymer selectivity, δ = 3.45 ppm), and the signals due to the trans-cyclic carbonate (δ = 4.00 ppm), and cis-cyclic carbonate (δ = 4.67 ppm). b Determined by SEC, in THF, at 40 o C, calibrated using polystyrene standards.   Figure S21. 1 H and 31 P{ 1 H} NMR spectra of LInBr in C6D6 (298 K) after the addition of a slight excess of THF. It can clearly be seen that the THF resonances stayed unchanged compared to "free" THF, indicating that the THF is not coordinated to indium in solution. Figure S22. DOSY NMR analysis of LInBr in d8-THF (298 K). As it can be seen from the DOSY, there is no mixing of the d8-THF and LInBr resonances indicating that THF is not coordinating to indium in solution.              Figure S42. 1 H NMR spectra (d8-THF 298 K) of LInOAc before the addition of CHO (higher spectrum), after CHO addition (second spectrum from the top), after heating the mixture at 60 o C for 24 h (third spectrum from the top) and after heating the mixture at 60 o C for 48 h (lower spectrum). Figure S43. Molecular structure of Amyl LInO t Bu with thermal ellipsoids at the 50% probability level, hydrogen atoms and a THF molecule omitted for clarity. Figure S44. 1 H and 31 P{ 1 H} NMR spectra of Amyl LInO t Bu (C6D6, 298 K). Figure S45. 13 C{ 1 H} NMR spectrum of Amyl LInO t Bu (C6D6, 298 K). Figure S46. DOSY NMR analysis of Amyl LInO t Bu (d8-THF, 298 K). Figure S47. 1 H and 31 P{ 1 H} NMR spectrum of Cumyl LInO t Bu (C6D6, 298 K).      Statistical analysis of tetrad distributions observed in the 13 C{ 1 H] NMR spectra in the carbonyl region was used to determine the stereocontrol of the polymerization. 6 The probability of racemic (Pr) or meso (Pm) enchainment was determined by solving the following equations. The predicted tetrads sequences are governed by: 8 [mmm] = Pm 4 + (1-Pm) 4 [S1]

S44
[mmr] = After rearrangement the equation is simplified to: The Bernoullian test B described by the equation below was used to assess the accuracy of the model. A value equal to 1 is indicative of a chain-end control mechanism: 9 [S11]            Figure S67. DOSY NMR spectrum of the reaction of LInO t Bu and CHD (C6D6, 298 K). All the resonances corresponding to the LIn are aligned, while the resonances corresponding to the coordinated CHD are lower than expected. There is also some mixing between the resonances of the butanol (at 1.06 ppm) and CHD. Figure S68. Ellipsoid representation at 50 % probability of the molecular structure of LInCHD. The CHD coordinated to indium and a molecule of uncoordinated CHD are present in two orientations in the crystal lattice (upper structure and lower structure). Hydrogen atoms, a hexane molecule and a THF molecule were removed for clarity. Figure 69. 13 C{ 1 H} NMR spectrum (d8-THF) of LInO2CO t Bu. Figure S70. 1 H NMR spectrum of the product of the CO2 inserted product into LInO t Bu (d8-THF, 298 K); x side product attributed to partial hydrolysis of LInO t Bu due to residual water in CO2; * residual hexane.

Structural Refinement Details
LInNO3, LInOAC, LInCHD, and LInBr were found to have disordered t Bu moieties. The two predominate positions were found in the difference map and their occupancies were refined with a free variable. In most cases bond distances and thermal parameters were restrained to allow for a stable refinement.
The C24-and C53-based i-pentyl groups, the O70-based t-butoxide ligand, and the O80-based included THF solvent molecule in the structure of Amyl LInO t Bu were all found to be disordered, and in each case two orientations were identified, of ca. 92:8, 72:28, 59:41 and 74:26% occupancy respectively. The geometries of each pair of orientations were optimized, the thermal parameters of adjacent atoms were restrained to be similar, and only the non-hydrogen atoms of the major occupancy orientations were refined anisotropically (those of the minor occupancy orientations were refined isotropically).
LInCHD showed severe disorder of the O-Cy-OH moiety attached to indium. This group could be modelled in two locations with a refined occupancy of nearly 50%. This disorder subsequently affected the cocrystallized CHD, which is involved in hydrogen bonding. This too was modelled in two locations with the same free variable. Bond distances and thermal parameters were restrained to allow for stable refinement. After modelling this disorder, additional electron density still remained elsewhere in the unit cell which was too severely disordered to be modelled as solvent, thus the SQUEEZE 17 protocol of the PLATON 18 suite of programs was utilized.
The bridging dicarbonate, O3C-propylene-CO3, in (LInCO2)2POD was disordered over an inversion center resulting in ½ of the propylene methyl present in the asymmetric unit. This methyl was further disordered over two positions resulting in the identification of two methyl groups within the asymmetric unit, each of ¼ occupancy. Bond distances and thermal parameters were restrained to allow for stable refinement.