Improved Control Through a Semi-Batch Process in RAFT-Mediated Polymerization Utilizing Relatively Poor Leaving Groups

Materials and methods N-Vinylpyrrolidone (NVP) was purchased from Merck and vacuum distilled over 5% ground potassium hydroxide before use. Styrene and methyl methacrylate were obtained from Sigma Aldrich and were passed through a basic alumina column to remove inhibitor immediately before use. Potassium Oethyldithiocarbonate, anhydrous calcium chloride, anhydrous magnesium sulfate, potassium hydroxide pellets and anhydrous calcium sulfate were purchased from Merck and used as received. N-(Bromomethyl)phthalimide, bromobenzene, N-methylaniline, methyl 2-bromopropionate, bromoacetic acid, carbon disulfide, sodium hydride (60 % dispersion in mineral oil), 1,1azobis(cyclohexanecarbonitrile) (V-88), and chloroform-d 99.8 atom % were purchased from Sigma Aldrich and used as received. Azobisisobutyronitrile (AIBN) was purchased from Merck and recrystallized from methanol before use. Benzene-d6 (99.6 atom %) was purchased from ACROS Organics and used as received. Methanol, chloroform, dichloromethane, 1,4-dioxane and diethyl ether were purchased from KIMIX. Methanol was fractionally distilled before use. Chloroform and dichloromethane were washed with concentrated sulfuric acid followed by 5% sodium bicarbonate solution and finally water before being dried over anhydrous calcium chloride and fractionally distilled from anhydrous calcium sulfate. Diethyl ether and 1,4-dioxane were distilled from sodiumbenzophenone ketyl. Dimethylacetamide (Chromosolv® Plus, for HPLC ≥ 99.9%) was purchased from Sigma Aldrich and used as received. Butylated hydroxytoluene (BHT) 99% (GC) was purchased from Sigma Aldrich and used as received. Lithium chloride ≥ 98% was purchased from Riedel-de Haën and used as is. All NMR spectra were recorded using a Varian 300 MHz VNMRS, Varian 400 MHz Inova or Varian 600 MHz Unity spectrometer. SEC analysis, in DMAc, was performed on a setup consisting of a Waters 717 plus autosampler connected to a Shimadzu LC-10AT pump with the following column configuration: 1×PSS GRAM analytical precolumn (10 μ particle size, 8.0×50 mm), 1×PSS GRAM analytical column (10 μ particle size, 100 Å pore size, 8.0×300 mm), 2×PSS GRAM analytical column (10 μ particle size, 3000 Å pore size, 8.0×300 mm.) A Waters 2487 dual wavelength absorbance detector and a Waters 410 differential refractometer were connected in series. The SEC system was calibrated using narrow dispersity poly(methyl methacrylate) standards. SEC in THF was performed as follows: A SEC instrument equipped with a Waters 717plus Autosampler, Waters 600E system controller and Waters 610 fluid unit were used to perform SEC analyses. A Waters 2414 differential refractometer was used for detection. Two PLgel 5 μm Mixed-C columns and a PLgel 5 μm guard column were used. The oven temperature was maintained at 30 °C and 100 μL of 2 mg mL−1 sample was injected into the column set. Tetrahydrofuran (THF) (HPLC grade, BHT stabilised) was used as the eluent for the analyses at a flow rate of 1 mL min−1. Narrow polystyrene standards with molar masses ranging from 800–2 × 106 g mol−1 were used to calibrate the instrument. Molar masses obtained from SEC are reported as h-polystyrene equivalents.


Introduction
2][3] A high tolerance to impurities and many different functional groups makes radical polymerization preferred over ionic polymerization for industrial processes.Reaction temperatures also play a less crucial role for radical polymerization than for ionic polymerization.However, there are drawbacks to conventional radical polymerization.][7][8][9][10] Arguably, among the most efficient of the RDRP techniques is RAFT (Reversible Addition-Fragmentation chain Transfer) mediated polymerization.RAFTmediated polymerization is compatible with virtually all monomer classes that are accessible via conventional radical polymerization.2][13] For example, several researchers have referred to more activated monomers (MAMs) such as styrene and acrylates and less activated monomers (LAMs) such as vinyl acetate and N-vinylpyrrolidone (NVP).Where MAMs are typically mediated by dithiobenzoates and trithiocarbonates  19 and others has shown that also the choice of the so-called leaving group (or R-group) is essential in obtaining good control.It is commonly known that the choice of a RAFT agent with a large chain transfer constant is essential to obtain a narrow molar mass distribution.However, the factor that controls the degree of control is the probability of chain transfer.It has earlier been pointed out by Moad and coworkers that in a polymerization controlled via degenerative chain transfer, apart from a chain transfer agent (CTA) with a large chain transfer constant, also a low ratio of monomer concentration to CTA concentration will lead to a large probability of chain transfer and therefore a narrow molar mass distribution. 6,22,23 Mthacrylic macromonomers are relatively poor CTAs, which Moad and co-workers used to mainly study the synthesis of block copolymers.In starved feed emulsion polymerization experiments they reach dispersities (Đ) as low as 1.2-1.3. 23Similar experiments conducted in solution show values of Đ ≅ 1.5. 22The manipulation of monomer-to-RAFT agent ratio to improve control over the polymerization can be used to address either the R-or Z-group effect of the RAFT agent.In the event of a relatively poor Z-group, as in the case of a xanthate-mediated polymerization of MAMs, a continuous slow addition of monomer can overcome the inherently low C tr , maximizing the RAFT-to-monomer ratio at any instant during the polymerization.Monteiro and coworkers later confirmed the use of slow monomer addition to improve the level of control (as judged by a low Đ) in a RAFT-mediated emulsion polymerization utilizing a xanthate as a chain transfer agent. 24,25 Hwever, in the event of a relatively poor R-group, a discrete semi-batch process can lead to improved Đ values, in contrast to the case of a poor Z-group which requires a continuous semi-batch process.Surprisingly, this effect has, to the best of our knowledge, never been systematically investigated using typical RAFT agents such as dithiobenzoates, xanthates, and dithiocarbamates for a solution polymerization.As such, the focus of the current study is to demonstrate the Please do not adjust margins Please do not adjust margins  In the present contribution we will show examples of RAFTagent/monomer combinations that provide poor control when conducted as a batch polymerization (Ð ≅ 1.5).In addition we will show that the degree of control can be significantly improved by performing the reaction in semi-batch mode.The first example will be the RAFT-mediated polymerization of NVP in the presence of O-ethyl-S-(phthalimidylmethyl)xanthate (RAFT agent 1 -Figure 1), which was used previously to provide poly(N-vinylpyrrolidone) (PVP) with an aminefunctionalized α-end group (after deprotection). 16

Results and discussion
As was pointed out earlier by Moad and coworkers, the probability of chain transfer can be approximated by equation 1.
For a RAFT-mediated polymerization, the equilibrium constant for chain transfer to the initial RAFT agent is defined as a composite term, shown in equation 2. This, however, has no effect on the general applicability of equation 1.

Please do not adjust margins
Please do not adjust margins The general strategy to get good control (i.e.low Ð) in a RAFTmediated polymerization is by the use of RAFT agents with a high chain transfer constant C tr = k tr /k p , which leads to a large probability of chain transfer.However, inspection of equation 1 leads to the conclusion that an alternative strategy to get good control is selection of a low ratio of monomer concentration to RAFT agent concentration ([M]/[RAFT]).In order to test the efficacy of RAFT agent 1 as a RAFT agent for the polymerization of NVP, an initialization experiment was conducted according to the procedure described in the supporting information.Figure 2 shows the fractional conversion profiles of NVP and RAFT agent 1. Figure 2 clearly shows that the RAFT agent does not get fully converted into macro-RAFT agent.The most plausible explanation for this behaviour is that the oligo-NVP chains are better leaving groups than the original phthalimidomethyl leaving group.In other words, RAFT agent 1 possesses a low chain transfer constant.Next, two batch-wise RAFT agent 1-mediated NVP polymerizations were carried out in which two different monomer-to-RAFT agent ratios were used, i.e. 117 and 196 (entries 1 and 2, Table 1).Samples were taken from the reaction mixture at various stages (Tables 1 and S1).Two important observations can be made.First, the conversion of the RAFT agent into macro-RAFT agent is only occurring very gradually, and even at the end of the experiments, small amounts of the original RAFT agent are still present (Table S1).Evolution of molecular weight and dispersity with increasing monomer conversion for target DP of 117 and 196 is shown in Figure 3. Second, the dispersity values of the polymers are consistently around 1.5 throughout the polymerizations, in agreement with results previously reported by Postma et al.   16   This is a frequently observed phenomenon that is particularly common for polymerization in which RAFT agents with relatively low chain transfer constant are employed.1) To overcome the effect of a low chain transfer constant, polymerizations were carried out in semi-batch mode.The essence is that the monomer is fed in stages into the reaction.As a consequence, the initial monomer to RAFT agent ratio is low compared to a batch reaction, and similar degrees of polymerization can still be reached.In the case where NVP polymerization was mediated with RAFT agent 1, the monomer was fed in a stepwise fashion instead of utilizing a continuous feed.The stepwise additions were performed in such a way that the monomer conversion was kept around 60%. Increasing the conversion to even higher values would lead to a larger probability of termination reactions, which have been neglected in equation 1. Tables S3 and S5 show the results of two semi-batch experiments where the target degrees of polymerization were 124 and 166, respectively.The final dispersity values for the target degree of polymerization of 124 and 166 are shown in Table 1, entries 3 and 4, respectively.Compared to the batch experiments, it is immediately clear that the conversion of the RAFT agent is much larger already at early stages of the polymerization and reaches full conversion well before the end of the experiment This journal is © The Royal Society of Chemistry 20xx Please do not adjust margins Please do not adjust margins (Tables S3 and S5).Simultaneously, the dispersities are relatively low from early stages of the reaction.Figure 4 shows the evolution of M n and Ð for a polymerization carried out in semi-batch mode, for an overall target degree of polymerization of 124 (entry 3, Table 1).The monomer conversion values in Figure 4 are based on overall monomer used, and the theoretical molar masses are calculated based on the overall monomer-to-RAFT ratio.A better correlation between expected and measured number average molecular weight values is seen in Figure 4 (semibatch mode polymerization) compared to Figure 3 (batch mode polymerization).Further comparisons between batch and semi-batch mode of polymerization are exemplified in Table 1 (and Tables S2 -S11) for various RAFT agent/monomer combinations.In cases where batch polymerization yields polymers with Ð > 1.6, a switch to semi-batch mode of polymerization does not improve the control over the polymerization.This is clearly evidenced in cases of polymerizations of styrene and methyl methacrylate mediated by RAFT agents 2 (entries 9-11, Table 1) and 3 (entries 15 and 16, Table 1), respectively.Previously reported batch polymerization of styrene and MMA in solution mediated by RAFT agent 3 exhibited dispersity values in agreement with those obtained in this study. 26In the case of MMA polymerization, it can be postulated that the oligo-MMA tertiary radical is a far better leaving group than the R-group of RAFT agent 3, thereby affording no improvement by switching to semi-batch mode of polymerization.The improved control, obtained by switching to a semi-batch polymerization process, is significant in that it eliminates the need for one to use RAFT agents with excellent leaving groups, which often require demanding synthetic protocols.

Conclusions
Control over RAFT mediated polymerizations was improved by performing the experiment in a semi-batch mode.In doing so, the ratio of monomer concentration to RAFT agent concentration is kept low at early stages of the polymerization.This directly increases the probability of chain transfer and therefore mitigates the negative effect of a low chain transfer constant on the width of the molar mass distribution.Our preliminary assessment on employing the concept of varying monomer-to-RAFT agent ratio to improve control is that such a protocol is suited for systems in which the batch process yields Ð ≅ 1.5.

Figure 1 .
Figure 1.Structures of RAFT agents employed in the batch and semi-batch polymerizations

Figure 3 .
Figure 3. Evolution of M n and Ð with conversion for NVP polymerization in batch mode, using RAFT agent 1.The employed monomer to RAFT agent ratios are 117 (•) and 196 (○), entries 1 and 2 inTable 1, respectively.

Figure 4 .
Figure 4. Evolution of M n and Ð as a function of conversion for NVP polymerization in semi-batch mode, using RAFT agent 1 (entry 3, Table1)

Table 1 .
Experimental Results for RAFT-Mediated Polymerization of N-Vinylpyrrolidone, Styrene and Methyl Methacrylate in Batch and Semi-Batch Mode.