Scott C.
Radzinski‡
,
Jeffrey C.
Foster‡
and
John B.
Matson
*
Department of Chemistry and Macromolecules and Interfaces Institute, Virginia Tech, Blacksburg, Virginia 24061, USA. E-mail: jbmatson@vt.edu
First published on 14th January 2015
A novel dithiocarbamate chain transfer agent (CTA1) with a directly polymerizable Z-group was synthesized for use in reversible addition–fragmentation chain transfer polymerization (RAFT). This CTA effectively mediated RAFT polymerization of styrenic and acrylic monomers with dispersities (Đ) < 1.08. Utilizing the polymerizable Z-group on the ω-chain end that is inherited from the RAFT process, bottlebrush polymers were synthesized via ring-opening metathesis polymerization (ROMP) in a grafting-through process. The effect of a number of parameters on the grafting process was studied, and optimized conditions yielded bottlebrush polymers of controllable molecular weights, narrow molecular weight distributions, and high conversions (>90%). Bottlebrush polymers made by a transfer-to strategy were also synthesized from CTA1. In this case, ROMP was first carried out to produce poly(CTA1) (PCTA1), then RAFT was performed from the PCTA1 backbone. This technique allows for the preparation of high molecular weight bottlebrush polymers without radical coupling between bottlebrush polymers. Lastly, regardless of the synthetic method, all bottlebrush polymers produced using CTA1 are composed of polymeric side chains that are attached to the bottlebrush backbone through a labile dithiocarbamate linkage that can be cleaved in the presence of nucleophiles such as amines. The unique combination of these capabilities allows for the study of bottlebrush polymer formation by both transfer-to and grafting-through strategies using a single agent.
Bottlebrush polymers can be prepared via one of four approaches: (1) the grafting-from strategy, whereby polymer side chains are grown from a polymeric backbone decorated with initiating functionalities; (2) the grafting-to methodology involving the attachment of pre-formed polymers to reactive sites on a polymer backbone; (3) the grafting-through or macromonomer (MM) approach, in which polymers fitted with a polymerizable moiety are utilized as MMs in an subsequent polymerization; and, (4) the transfer-to strategy (sometimes called the RAFT Z-group approach). Transfer-to is a unique hybrid of the grafting-from and grafting-to strategies in which polymeric radicals detach from the bottlebrush backbone, propagate freely in solution, and return to the backbone through a chain-transfer reaction with a pendant CTA.14 While the grafting-from, grafting-to, and transfer-to strategies yield macromolecules with a bottlebrush topology, the grafting process is hindered by steric interactions between adjacent polymer chains, resulting in low grafting densities.15 However, despite this shortcoming, grafting-from (and to a greater extent transfer-to) can be employed to synthesize bottlebrush polymers with relatively higher molecular weights (on the order of ≥106 Da) than are possible with grafting-through. In contrast, grafting-through results in “perfectly grafted” (i.e. the highest possible grafting density) bottlebrush polymers, as each repeat unit bears a polymeric side chain. In view of the high grafting density and synthetic versatility of the grafting-through technique, recent efforts have focused on its application.11,16–31
To prepare well-defined bottlebrush polymers via the grafting-through strategy, reversible-deactivation radical polymerization techniques such as atom transfer radical polymerization (ATRP) and reversible addition–fragmentation chain transfer polymerization (RAFT) are often employed.1 Generally, semi-telechelic MMs of predefined MW and low dispersity (Đ) are synthesized via one of these techniques. In a second step, the resulting MMs are functionalized with a polymerizable moiety in a post polymerization reaction, and the MMs are subsequently polymerized using an additional polymerization method in the third and final step.20,23 Alternatively, in a two-step method, MM synthesis can be conducted in the presence of an initiator or chain-transfer agent (CTA) containing an orthogonal functionality.17,21,22 For example, norbornene and derivatives thereof have been coupled to dithioester and trithiocarbonate CTAs.19 In this case, RAFT was utilized to prepare the MM grafts, and ring-opening metathesis polymerization (ROMP) was employed in a subsequent grafting-through step. ROMP is particularly well suited for this purpose because of the high functional group tolerance and rapid propagation rate of several ruthenium-based olefin metathesis catalysts (Fig. 1).16 Bottlebrush polymers have been prepared via a combination of these two polymerization techniques in separate steps as described above, or more simply via a one-pot strategy wherein a ROMP catalyst is added to a terminated RAFT reaction mixture.18
RAFT polymerization is mediated by a thiocarbonylthio-containing agent with an activating Z-group and a leaving group (R-group) that is comparable in radical stability to that of the monomer-derived radical. A number of R- and Z-groups have been utilized to gain control over the RAFT polymerization of a wide range of vinyl monomers.32 However, to our knowledge, there exists no report in the literature on the incorporation of a directly polymerizable Z-group such as exo-norbornene imide into a RAFT agent (Scheme 1). We envisioned that such a CTA (CTA1) could be employed in both RAFT transfer-to and ROMP grafting-through methodologies.
While directly polymerizable R-groups have been incorporated into RAFT CTAs,18,19 a directly polymerizable Z-group would experience the benefits inherent in the Z-group, or transfer-to, approach.33–36 Such benefits are a result of the RAFT mechanism when applied in a graft polymerization. During a RAFT transfer-to polymerization, growing polymer chains detach from the bottlebrush polymer backbone during propagation and then add back to the backbone through reaction with a pendant thiocarbonylthio group. Because the growing “arms” are free in solution, coupling between growing adjacent “arms” attached to the bottlebrush backbone does not occur. In addition, radical coupling between growing bottlebrush polymers cannot occur, as the propagating radical resides on the detached polymeric side chains. As a result, the transfer-to approach affords bottlebrush polymers with lower dispersities and higher possible conversions relative to conventional RAFT grafting-from using the R group approach.36,37 An additional advantage of the incorporation of a directly polymerizable Z-group is the location of the thiocarbonylthio group in the bottlebrush polymer. In the case of bottlebrush polymers prepared from CTA1, the thiocarbonylthio group would link the polymeric arms to the backbone polymer, in contrast to systems using directly polymerizable R-groups, which leave the thiocarbonylthio group on the bottlebrush surface. Given the wealth of literature concerning the post-polymerization removal of RAFT end groups,38 we envisioned the possibility of thiocarbonylthio degradation-driven side chain dissociation. Herein, we investigate the efficacy of CTA1 as a mediator of RAFT polymerization and the availability of the norbornene functionality for ROMP. Additionally, we evaluate the preparation of bottlebrush polymers via both transfer-to and grafting-through strategies. Finally, we explore side-chain cleavage by aminolysis to investigate side chain molecular weights and molecular weight distributions.
![]() | ||
Scheme 2 One-pot synthesis of CTA1. aExperimental conditions: (i) CS2, KOH, DMF; (ii) benzyl bromide. |
Column purification was required to remove unreacted starting materials, and isolated yields ranged from 35–48%. While this range is lower than reported values for the synthesis of other dithiocarbamates,41,42 many such CTAs are derived from electron-rich secondary amines in contrast to the relatively electron-deficient imide employed in our study. This electron-deficiency results in the limited nucleophilicity of the imide nitrogen,43 hindering its addition to electrophiles such as CS2. Longer reaction times or increased temperatures were not found to increase yields.
To evaluate our hypothesis, RAFT polymerizations of styrene and n-butyl acrylate (nBA) were carried out in the presence of CTA1 (Scheme 3). The polymerizations were conducted in THF (1:
1 v/v% THF–monomer) at 75 °C for styrene or 60 °C for nBA in the presence of 2,2′-azobis(isobutyronitrile) (AIBN). To maintain a high level of chain end fidelity, a [CTA]/[AIBN] ratio of 10
:
1 was chosen.54 Kinetic analysis was performed by removing aliquots of the polymerization solution at various time points via N2-purged syringe. The polymerizations were quenched by exposing the reaction mixture to air and submerging the reaction vessel into liquid N2. Molecular weight (MW) and dispersity (Đ) were determined by size-exclusion chromatography (SEC), and conversions were measured by 1H NMR spectroscopy. CTA1-mediated RAFT polymerization of styrene and nBA yielded polymers of controllable molecular weights with narrow molecular weight distributions (Đ < 1.08). As previously reported for similar dithiocarbamate CTAs,55 polymerization of methyl methacrylate was uncontrolled. On the opposite end of the monomer reactivity spectrum, polymerization of vinyl acetate was completely inhibited in the presence of CTA1.
![]() | ||
Scheme 3 CTA1-mediated RAFT polymerization of styrene or nBA. aExperimental conditions: AIBN, THF, 75 °C for styrene or 60 °C for nBA. |
Kinetic analysis of CTA1-mediated polymerization of styrene and nBA is shown in Fig. 2. Molecular weight distributions determined by SEC were monomodal with low Đ, indicative of a well-controlled polymerization and a high chain transfer efficiency of CTA1 under the conditions investigated. Conversion increased linearly with time for nBA, and Đ decreased over the course of the reaction for both nBA and styrene. In the case of styrene, a non-linear relationship was observed in the semi-logarithmic plot (Fig. 2C) and a maximum conversion of only 48% was obtained after 24 h, indicative of the occurrence of termination reactions during the polymerization. The polymerization was repeated under more dilute conditions (2:
1 v/v% THF–styrene), resulting in a reduction of termination reactions and a higher terminal conversion (81% after 24 h). For both monomers, the linear relationship between MW and conversion corroborated the controlled nature of the polymerization (Fig. 2D).
Chain extension of a polymer prepared using CTA1 was attempted to further validate its capability to mediate controlled RAFT polymerization. For the first block, poly(nBA) of MW = 6900 Da (Đ = 1.06) was synthesized using CTA1. This polymerization yielded a macroCTA with a dithiocarbamate at the ω-chain end. The poly(nBA) macroCTA was utilized in a second step to control the polymerization of styrene ([M]/[CTA]/[AIBN] = 1000:
1
:
0.1), ultimately resulting in the formation of a block copolymer. After 4 h, the polymer had grown to 27.0 kDa as measured by SEC (Đ = 1.05) (Fig. S9†). This measured MW agreed with the expected value of 26.6 kDa based on conversion as measured by 1H NMR spectroscopy.
To prepare macromonomers for ROMP grafting-through, a series of polymers of differential MW were synthesized, and the isolated polymers were characterized by SEC and 1H NMR spectroscopy (Fig. S8 and S10†). A summary of our analysis is provided in Table 1. The subscripts in the polymer name assignments refer to the degree of polymerization of the polystyrene component of the macromonomers. Polymers with relatively narrow molecular weight distributions were obtained. In addition, MWs measured by SEC were in good agreement with those calculated from conversion using 1H NMR spectroscopy.
Polymera | M n (SEC) (Da)b |
M
n (theo)![]() |
Đ |
---|---|---|---|
a Average degree of polymerization shown as a subscript calculated from SEC data using the formula DP = (Mn − MWCTA)/MWstyrene. b Measured by SEC using absolute MW determined by light scattering. c Determined by 1H NMR spectroscopy using the formula Mn (theo) = MWstyrene × ([styrene]/[CTA]) × % conv. | |||
MM29 | 3300 | 3100 | 1.05 |
MM52 | 5700 | 6000 | 1.02 |
MM113 | 12![]() |
13![]() |
1.04 |
An important outcome of CTA1-mediated RAFT polymerization is the preservation of the reactive norbornene olefin moiety on the ω-end of the resulting polymer chain, allowing for subsequent bottlebrush formation via ROMP grafting-through. The orthogonality of the norbornene olefin with reversible-deactivation radical polymerization has been previously shown.56 The poor reactivity of the internal norbornene olefin in RAFT relative to the terminal, vinyl groups of acrylic monomers is attributed to their differences in electronics. 1H NMR spectroscopic analysis of the pure MMs confirmed the preservation of the norbornene olefin, evident as a singlet at 6.2 ppm (Fig. S8†).
CTA1 can be utilized for both transfer-to and grafting-through strategies, creating a unique opportunity to evaluate the differences between these two techniques while using identical chemistry. To evaluate the ability of CTA1 to control the growth of high MW side chains from a polymeric backbone by RAFT transfer-to polymerization, CTA1 was first polymerized by ROMP (Scheme 4 left) using Grubbs’ 3rd generation catalyst ((H2IMes)(Cl)2(pyr)2RuCHPh) (G3) (50
:
1 [CTA1]/[G3]) to yield a poly(CTA1) (PCTA1) consisting of a poly(norbornene) backbone with a dithiocarbamate group on each repeat unit (Fig. S7†). Following our characterization of the resulting polymeric CTA, PCTA1 was then subjected to RAFT polymerization conditions ([styrene]/[macroCTA]/[AIBN] = 50
000
:
1
:
0.1, THF, 75 °C), and the progress of the reaction was monitored by SEC (Fig. 3). Molecular weight distributions of the growing bottlebrush polymer were monomodal, and the corresponding polymers exhibited successively shorter retention times, indicative of increasing MW. After 91 h, the polymer had grown from 25.1 kDa (for PCTA1) to 1250 kDa, with Đ remaining low (<1.02) throughout the course of the polymerization. A second peak in the SEC traces at ∼15 min corresponds to dead polymer arising from termination reactions between detached polymeric radicals. This phenomenon is discussed in further detail below.
A series of Ru-based metathesis catalysts were evaluated for their efficacy towards the grafting-through polymerization of CTA1-derived polystyrene (MM29) (Scheme 4 right). Toward this end, ROMP of MM29 was conducted in the presence of various ROMP catalysts at a [MM]/[catalyst] ratio of 50:
1 (Fig. S11†). Grafting-through polymerizations initiated by Grubbs’ 1st ((PCy3)2(Cl)2Ru
CHPh) (G1) and 2nd generation ((H2IMes)(Cl)2(PCy3)Ru
CHPh) (G2) catalysts showed poor conversion to the corresponding bottlebrush polymer after 1 h in CH2Cl2 (Table 2, entries 1 and 2, respectively). The Hoveyda-Grubbs’ 2nd generation catalyst ((H2IMes)(Cl)2(PCy3)Ru
CH(2-OiPrPh)) (HG2) effected higher conversion to bottlebrush polymer (78%); however, the resulting polymer had a broad molecular weight distribution (Đ = 1.50) (Table 2, entry 3). Grafting-through by the modified Grubbs’ 2nd generation catalyst (G3) showed 91% conversion to bottlebrush polymer with Đ = 1.03 and a monomodal molecular weight distribution (Table 2, entry 5). Based on this analysis, G3 was utilized for further experiments.
Entry | MM | MM Mn,SEC![]() |
Cat. | [MM]/[I] | [MM] (mg mL−1) | % Conv. to BBb | BB Mn,theo![]() |
BB Mn,SEC![]() |
BB Đa |
---|---|---|---|---|---|---|---|---|---|
a Measured by SEC using absolute MW determination by light scattering. b Determined from SEC by comparing the integrations of BB and MM peaks. c Calculated using the formula % conv. × ([M]/[I]) × Mn,MM. All polymerizations were conducted for 1 h in CH2Cl2. | |||||||||
1 | MM29 | 3.3 | G1 | 50![]() ![]() |
100 | 8 | 13.2 | 21.1 | 1.11 |
2 | MM29 | 3.3 | G2 | 50![]() ![]() |
100 | 16 | 26.4 | 353 | 1.48 |
3 | MM29 | 3.3 | HG2 | 50![]() ![]() |
100 | 78 | 129 | 334 | 1.50 |
4 | MM29 | 3.3 | G3 | 25![]() ![]() |
100 | 91 | 75.0 | 85.6 | 1.04 |
5 | MM29 | 3.3 | G3 | 50![]() ![]() |
100 | 91 | 150 | 152 | 1.03 |
6 | MM29 | 3.3 | G3 | 75![]() ![]() |
100 | 89 | 220 | 240 | 1.14 |
7 | MM29 | 3.3 | G3 | 100![]() ![]() |
100 | 87 | 330 | 340 | 1.25 |
8 | MM29 | 3.3 | G3 | 50![]() ![]() |
25 | 70 | 116 | 100 | 1.10 |
9 | MM29 | 3.3 | G3 | 50![]() ![]() |
50 | 80 | 132 | 125 | 1.09 |
10 | MM52 | 5.6 | G3 | 50![]() ![]() |
100 | 65 | 182 | 190 | 1.09 |
11 | MM113 | 12.1 | G3 | 50![]() ![]() |
100 | 40 | 242 | 180 | 1.16 |
Macromonomer MW was found to influence ROMP grafting-through polymerization (Table 2, entries 5, 10–11, Fig. S12†). Comparing the grafting-through polymerization of three MMs of differing MW (3.3 KDa, 5.6 KDa, and 12 KDa) revealed an inverse correlation between MW and conversion. This observed dependence of conversion on MW is attributed to steric factors.57 The MW of the isolated bottlebrush polymer made from MM29 was 152 kDa (Table 2, entry 5), which is in good agreement with the expected molecular weight of 150 kDa. Despite lower conversions, polymerizations of MM52 and MM113 showed reasonable agreement between theoretical and observed Mn values and maintained fairly low dispersities.
To further confirm the livingness of grafting-through polymerization of CTA1-derived MMs, the [MM29]/[G3] ratio was varied from 25:
1 to 100
:
1 (Table 2, entries 4–7). The resulting bottlebrush polymers were monomodal and their SEC traces exhibited the expected inverse relationship between MW and retention time (Fig. 4A). Mn values determined by SEC are in good agreement with theoretical MWs. A plot of Mnvs. [M]/[I] showed the anticipated linear trend (Fig. 4B).61 Additionally, dispersity values were low, ranging from 1.04–1.25 throughout the series.
Macromonomer concentration had a small but significant influence on the conversion to bottlebrush polymer (Table 2, entries 5, 8–9, Fig. S13†). Measured conversions varied from 70% to 80% to 91% for the evaluated concentration range (25 mg mL−1 to 100 mg mL−1), with increasing concentration resulting in higher conversion to BB.
It is important to note that MM conversion could not be increased beyond 91%, even under optimized conditions. Given the wealth of literature regarding the dependence of RAFT chain-end fidelity on a number of factors including [M]/[I] and [CTA]/[I] ratios,49 it can be hypothesized that the observed conversion limit of 91% originates from the existence of “non-living” chains, or those that do not possesses a thiocarbonylthio group on the ω-chain end, in the MM samples. It has been shown that ca. 8% of polymer chains prepared by RAFT polymerization under similar conditions are of this “non-living” type.54,62 Polymers of this “non-living” variety do not possess a norbornene on the chain end and thus will not polymerize during ROMP grafting-through.
The presence of the dithiocarbamate moiety adjacent to the bottlebrush backbone allowed for the cleavage of the polystyrene side chains (Scheme 5). A series of bottlebrush polymers prepared by RAFT transfer-to were dissolved in THF and exposed to a 40% w/v solution of methylamine in H2O for 72 h. The resulting mixture of backbone and dimerized side chain polymers was separated from residual reactants via precipitation. SEC analysis of the aminolyzed bottlebrush polymers showed a clean shift of the molecular weight distribution to a longer retention time, indicating that the side chains had been quantitatively cleaved (Fig. S14†). Interestingly, the measured MWs of the dissociated side chains were higher than the expected values calculated by dividing the experimental BB MW by the DP of PCTA1 (Table 3). We attribute this deviation to three independent phenomena. First, oxidation of ω-chain end thiols liberated during aminolysis resulted in disulfide bond formation, as has been previously reported.63 This reaction doubles the observed molecular weight of the side chains. Second, increased steric crowding of the dithiocarbamate CTAs near the bottlebrush backbone by the attached polystyrene sidechains likely led to radical termination reactions between detached polymeric radicals during the transfer-to process, as has also been observed.65,66 This side reaction not only yielded a significant amount of dead polymer, evident as a low MW peak in the SEC trace (Fig. 3), but also likely resulted in lower than perfect grafting density and a higher than expected average side chain MW. Lastly, “dead” polymer arising during the RAFT polymerization is also incorporated into this sample. Although the concentration of “dead” polymer is surely eclipsed by the more abundant aminolyzed sidechains, these “dead” polymer chains could explain the observed broadening of the molecular weight distributions of the aminolyzed polymers relative to those of the bottlebrush polymers prior to aminolysis. While ultrahigh MW polymers can be obtained via RAFT transfer-to, the limitations of this methodology (i.e., limited control over grafting density) are made clear by this aminolysis experiment.
![]() | ||
Scheme 5 Displacement of polymeric side chains by methylamine. Experimental conditions: (i) CH3NH2, H2O–THF. |
Polymerization time (h) |
M
n,SEC![]() |
Đ |
M
n,sidechains theo![]() |
M
n,sidechains SEC![]() |
Đ
aminolysis![]() |
---|---|---|---|---|---|
a Measured by SEC using absolute MW by light scattering. b Determined using the formula: Mn,sidechains = Mn,BB/DPMCTA1. | |||||
0 | 25.1 | 1.02 | — | — | — |
19 | 356 | 1.01 | 4.7 | 19.5 | 1.15 |
42 | 737 | 1.01 | 9.9 | 35.5 | 1.21 |
67 | 1000 | 1.01 | 13.2 | 47.4 | 1.23 |
91 | 1250 | 1.01 | 16.5 | 56.9 | 1.28 |
To further evaluate bottlebrush polymers prepared by the grafting-through approach, we subjected these polymers to aminolysis as well. ROMP of MM29 at a [MM]/[I] ratio of 50:
1 was carried out to give a BB with degradable side chain linkages. A MW of 164 kDa was determined for the BB, with Đ = 1.07. Aminolysis of the BB proceeded rapidly and quantitatively at rt in the presence of an excess of methylamine in THF–H2O. SEC analysis of the precipitated reaction mixture revealed a narrow molecular weight distribution and a MW approximately double that of the starting MM (5400 Da, Đ = 1.01), as expected for polystyrene side chain disulfide dimers (Fig. 5). Quantitative aminolysis was confirmed by the complete disappearance of the bottlebrush peak in the SEC trace. Additionally, DLS analysis of the cleaved bottlebrushes exhibited a shift in the size of the macromolecules from 11.8 ± 3.1 nm for the BB to 2.6 ± 0.4 nm for the dissociated side chain dimers (original MM29 = 2.0 ± 0.5 nm) (Fig. S15†).
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4py01567c |
‡ These authors contributed equally to this work. |
This journal is © The Royal Society of Chemistry 2015 |