The anion exchange resin Ambersep^® 900 OH (FLUKA, 20-50 mesh) was dried for several days under high vacuum. By titration of an aqueous suspension with hydrochloric acid, its capacity was determined to be 3.5 mmol g⁻¹. To remove the inhibitor, styrene and butyl acrylate (ALDRICH) were passed through a basic alumina (ALDRICH, Brockmann I, ∼150mesh, 58 Å) column. 2,2′-Azobis(isobutyronitrile) (AIBN, AKZONOBEL) was recrystallized from methanol and dried under high vacuum before use. Tetrahydrofuran (THF) used as the eluent in SEC (Carl Roth, stabilized with 2,6-di-tert-butyl-4-methylphenol) was used as received. Unless otherwise specified, all other chemicals were purchased from ALDRICH and used without further purification.

Molecular weight distributions were measured by means of SEC using a system composed of a JASCO AS-2055PLUS autosampler, a WATERS 515 pump, three PSS SDV columns (8 × 300 mm, nominal particle size: 5 μm, pore sizes: 10⁵, 10³ and 10² Å), a WATERS 2410 refractive index detector and a VISCOTEK VE 3210 ultraviolet (UV) detector (set to a wavelength of 310 nm). THF at 35 °C was used as the eluent at a flow rate of 1.0 mL min⁻¹. The SEC set-up was calibrated against PSS polystyrene standards of very low dispersity. The molecular weight distributions were adjusted according to the principle of universal calibration using Mark-Houwink parameters for linear pBA (K = 1.22 × 10⁻⁴dLg⁻¹, a = 0.700) [38].

Nuclear magnetic resonance (NMR) spectra were recorded at room temperature on a VARIAN UNITY 300 (¹H, ¹³C) or a VARIAN INOVA 500 (¹H) spectrometer, using deuterated chloroform as solvent and internal standard.

Differential scanning calorimetry was carried out on a METTLER TOLEDO DSC 820 apparatus which was operated with the STARe 9.01 software. Samples of 10–20 mg were heated under nitrogen atmosphere (flow rate 14 mL min⁻¹) from −52 °C to 170 °C and cooled back to −52 °C. To obtain the thermograms the samples were heated again to 170 °C. The heating and cooling rates were 10 °C min⁻¹. The glass transition temperature was taken as the onset temperature of the steps in the thermograms.

In the synthesis of polytrithiocarbonate A with the lowest degree of polymerization (see Table 1), cesium carbonate was employed as the base to produce trithiocarbonate ions [39]. Carbon disulfide (409 μL, 6.79 mmol) was added to a solution of cesium carbonate (2.21 g, 6.79 mmol) in 5 mL acetonitrile at room temperature. The color of the suspension changed to red. After 10 min of stirring, dimethyl 2,6-dibromoheptanedioate (739 μL, 3.40 mmol) was added and the color changed to yellow immediately. After another 21 h of stirring, dimethyl 2,6-dibromoheptanedioate (100 μL, 460 μmol) was added again. The suspension was stirred for 45 min and poured into water (100 μL). A yellow oil precipitated which was isolated and washed with water (2 × 100 mL) and methanol (2 × 100 mL). The product (665 mg, 67 %) was dried under vacuum for 2 d. See synthesis of B–E for NMR spectroscopic data.

In the synthesis of the polytrithiocarbonates B-E with higher degrees of polymerization (see Figure 2 and Table 1), the trithiocarbonate ions were generated at the surface of an anion exchange resin [40,41]. A typical procedure is reported here. Dry AMBERSEP^® 900 OH (13 g, 46 mmol OH⁻) was suspended in carbon disulfide (65 mL). The surface of the resin beads turned dark red immediately. Dimethyl 2,6-dibromoheptanedioate (5.00 mL, 23.0 mmol) was added. The suspension was heated to boiling and was stirred for 65 h under reflux. The color of the resin beads and the surrounding solution turned dark yellow over time. The reaction mixture was cooled to room temperature, poured in a fritted funnel and rinsed subsequently first with carbon disulfide (300 mL) and then with THF (300 mL). Both obtained fractions were concentrated under vacuum, washed with water (3 × 200 mL) and methanol (3 × 150 mL) and dried under high vacuum and were thus obtained as yellowish-orange oils, whereas the fraction obtained from rinsing with carbon disulfide (1.55 g, 23 %) was less viscous and paler in color than the THF fraction (771 mg, 11%).

¹H-NMR (300 MHz, CDCl₃): δ = 1.40–1.70 (m, CH₂–CH₂–CH₂), 1.85–2.14 (m, CH₂–CH₂–CH₂), 3.75 (s, COOCH₃), 4.14–4.24 (m, CH–Br), 4.59–4.79 (m, CH–S) ppm. ¹³C–NMR (75.58 MHz, CDCl₃): δ =22.1 (CH₂–CH₂–CH₂), 30.5 (CH–CS₃), 44.8 (CH–Br), 53.0 (COO–CH₃, CH–SCS₂), 170.4 (COO–CH₃), 219.2 (S–CS–S) ppm.

Polymerizations 1–6 of styrene and 7–9 of BA with the prepared polytrithiocarbonates as RAFT agents were carried out in bulk at 60 °C and ambient pressure using AIBN as initiator. Polymerizations I–VI of butyl acrylate with the prepared multiblock styrene-homopolymers were carried out in solution of toluene under the same conditions. See Tables 2–4 for the individual conditions. The indicated concentrations of the RAFT agents are referred to the molar masses which were measured by means of SEC and to the number of molecules, rather than the number of trithiocarbonate groups.

All polymerization mixtures were thoroughly degassed via three freeze-pump-thaw cycles and then evenly portioned into three to five capped glass vials in an argon-filled glove-box which were then inserted into a heating block. The reactions were stopped by cooling the vials in ice water after distinct time periods and residual monomer and solvent (I–VI) were removed by evaporation. Monomer to polymer conversion was determined by gravimetry.

The general structure of the polyfunctional RAFT agents that were used in this study is depicted in Figure 1. It is comparable to the one used by You et al. but contains one methylene group less per unit. These polytrithiocarbonates were generally prepared by the polycondensation reaction of dimethyl 2,6-bromoheptanedioate and trithiocarbonate anions. Two different strategies were followed to produce the trithiocarbonate anions via the reaction of carbon disulfide with a base: The first one (procedure 1) involves the production of the trithiocarbonate anions with cesium carbonate as the base [39]. However, it could only be successfully employed to prepare polyfunctional RAFT agents of very low degree of polymerization at room temperature. At elevated temperatures or longer reaction times, a side-product was formed, presumably because of the reaction of the dibromide's ester groups [43]. The second strategy (procedure 2) was based on the formation of the trithiocarbonate anions on the surface of an anion exchange resin. You et al. published a synthesis for polytrithiocarbonates using a chloride ion exchange resin [40] and it can be assumed that they used the products for their above mentioned study. Here, a hydroxy-functionalized exchange resin was employed directly [41], whose capacity was measured by titration with hydrochloric acid prior to use in order to be able to add the stoichiometrically exact amount of hydroxide ions to the system. The fact that the solubility of the produced trithiocarbonates in carbon disulfide decreased with increasing chain length enabled the possibility of fractionating them directly within the work-up of the reaction by first washing the resin with carbon disulfide and then with THF.

The prepared trithiocarbonates were examined by means of SEC. The theoretically expected positions of the individual low molar mass peaks in the SEC curves of the polytrithiocarbonates, which correlate to the individual degrees of polymerization, can be rationalized on the basis of their general structure which is shown in Figure 1: (1) M ¯ n , RAFT = g ¯ M TTC + 2 M EG + ( g ¯ − 1 ) M MG = 346 . 01 g mol − 1 + g ¯ ⋅ 294 . 42 g mol − 1 ⋅M̅_n,RAFT is the average molar mass of the RAFT agents with an average of ḡ trithiocarbonate groups, M_TTC, M_EG and M_MG are the molar masses of trithiocarbonate groups, end groups and middle groups (see Figure 1).

Because no appropriate calibration data were available, the SEC curves were measured using the calibration for linear polystyrene. The SEC curve of polyfunctional RAFT agent A (see Table 1), which was synthesized via procedure 1, is shown in Figure 2(a). The numbers in the diagram indicate the positions of the maxima in the number-weighted distribution. The UV detector signal—measured at 310 nm, where trithiocarbonates are known to absorb radiation [41]—nicely matches the refractive index (RI) detector signal. This indicates that molecules with higher masses possess more trithiocarbonate groups [44], which is also underlined by the expected structure of almost equidistant peaks. The first peak, though, can only be seen in the RI signal wherefore it has to be assigned to a reactant or a byproduct without trithiocarbonate groups. For the analysis of all SEC curves, the peak at 387 g mol^–1 was the first one that was considered. Comparing the positions of the peaks with the expected ones, it can be seen that the measured absolute positions and the distances between the peaks are considerably lower than the theoretical values, most likely due to the non-appropriate calibration. Nevertheless, the measured average molar masses of the prepared RAFT agents were used to calculate their average number of trithiocarbonate groups according to Equation (1). It has to be kept in mind that these values are only a rough approximation and probably too low. For polytrithiocarbonate A this yields an average number of ḡ_GPC,A = 1.34 trithiocarbonate groups per chain. Figure 2(b) shows the SEC curves of the polyfunctional RAFT agents C and D (see Table 1) that were prepared within the same reaction performed according to procedure 2. C was obtained by rinsing with CS₂ and D by rinsing with THF. These curves illustrate that by procedure 2, RAFT agents with higher degree of polymerization could be produced than with procedure 1 and that the fractionation of the reaction products was successful. Again, the good agreement of the RI signal and the UV signal demonstrates that molecules with higher molar mass also possess more trithiocarbonate groups, which is in line with the structural expectations. Generally, SEC analysis gave values of ḡ_GPC = 4.09–5.33 for the fraction obtained by rinsing with CS₂ and ḡ_GPC = 12.0–18.5 for the fraction obtained by rinsing with THF.

It was already shown by You et al. that the average number of trithiocarbonate groups within the prepared polyfunctional RAFT agents can also be determined by measuring their NMR spectra and calculating the ratio of the intensity I_4.7ppm at 4.7 ppm (proton in α-position to a trithiocarbonate group) and the intensity I_4.2ppm at 4.2 ppm (proton in α-position to a bromine atom) [40]: (2) g ¯ NMR = I 4 . 7 ppm I 4 . 2 ppm

The thus determined value for polytrithiocarbonate A lies under the minimum value of 1, which can possibly be explained by residual dibromide in the sample contributing only to I_4.2ppm. This is a considerable factor of inaccuracy of this method when polytrithiocarbonates with low degree of polymerization are examined. On the other hand, also the examination of trithiocarbonates with a very high number of trithiocarbonate groups becomes very unprecise because in these cases, it has to be divided by very small values of I_4.2ppm. The determined average numbers of trithiocarbonate groups for procedure 2 were ḡ_NMR = 2.89–4.09 for the CS₂ fraction and ḡ_NMR = 8.19–15.9 for the THF fraction (see Table 1).

To make sure that complete relaxation of all nuclei was achieved within the NMR measurements, one polytrithiocarbonate sample was re-measured with an additional relaxation delay of 15 seconds between the pulses. This measurement gave an identical spectrum which demonstrates that the spectra can indeed be interpreted quantitatively.

In order to produce segmented copolymers of styrene and BA, styrene is preferentially used for the first polymerization step, as macromolecular RAFT agents with the better leaving groups are formed. Nevertheless, to gain more insight into the system, three polytrithiocarbonates were also employed in polymerizations with BA. All homopolymerizations were performed in bulk and initiated thermally with AIBN as initiator. The resulting polymers were examined by SEC (see Tables 2 and 3) using the calibration data for the pure homopolymers which seem reasonably well applicable since the RAFT agents constitute only a minor part of the final polymer's main chain.

In all polymerizations, the molar masses grow linearly with monomer conversion, which underlines the living character of the systems and demonstrates the possibility to synthesize polymers with precisely specifiable block lengths up to very high molar masses. The degree of control, however, seems to be higher for the polymerizations of styrene. Here, products with lower PDIs were obtained. Within the individual polymerizations, the PDIs decrease to a rather constant value after the initial period of the polymerization. Exemplarily, a plot of M̅_n and the PDI (RI signal) as a function of monomer conversion along with the corresponding SEC curves are shown for polymerization 4 in Figure 3.

The general tendency can be observed that RAFT agents with more trithiocarbonate groups lead to polymers with higher PDIs. This finding, however, does not mean that these RAFT agents exert a lower control over the polymerizations, but is just what one would expect theoretically as we could show elsewhere based on the ideally expected distributions of the generated polymers [37]. The absolute PDI values are higher than those usually obtained in RAFT polymerizations because the molar mass distributions consist of several overlaying peaks with lower dispersity for each number of blocks. These individual peaks could partially be resolved when RAFT agents with low numbers of functional groups were employed (see below).

For the second polymerization step of the examined synthesis strategy, several prepared styrene polymers were employed as macromolecular RAFT agents in radical polymerizations of BA. The experimental conditions were comparable to those of the first polymerization step, except for the fact that the polymerizations were conducted in toluene because the solubility of the macromolecular styrene polymers in BA is rather low. The experimental results of all copolymerizations are listed in Table 4. It has to be noted that, for lack of appropriate SEC calibration data for the produced multiblock homopolymers, the calibration for linear polystyrene was used and the SEC curves and the data gained therefrom cannot be seen at all as fully correct. (In some cases even molar masses lower than those of the employed macromolecular RAFT agents were obtained.) The given mass ratios r_S/BA of styrene and BA units in the polymers were calculated on the basis of the gravimetrically determined monomer conversions and are therefore not affected by this error. Figure 4 shows the SEC curves for all samples of copolymerization III. It can be seen that the molar masses clearly grow with monomer conversion, independently of the number of trithiocarbonate groups in the macromolecular RAFT agent. It is especially remarkable here that no tailing can be seen in the SEC curves, which, due to incomplete reinitiation, is a typical feature in the SEC curves of RAFT-prepared copolymers with only two segments. This finding may be accounted to the fact that most polymer chains comprise several trithiocarbonate groups which makes it very likely that the reinitiation takes place at least at one of them. Accordingly, the dispersities of all produced multiblock copolymers are relatively low.

The multiblock polymers and copolymers were cleaved at the trithiocarbonate groups in order to compare the molar masses of the polymers prior to (M̅_n,0) and after (M̅_n,1) the cleavage. Ideally, the average number of trithiocarbonate groups would then be calculable by (3) g ¯ = M ¯ n , 0 M ¯ n , 1 – 1

The here used approach for the cleavage was the aminolysis reaction with n-butyl amine which leads to thiol-capped polymer chains. However, when cleaving multiblock polymers of only styrene, the oxidative coupling to longer polymers with disulfide bridges along their chain [45] could be observed (non-UV-active peak at double molar masses). To prevent this unwanted reaction, Lima et al. [46] presented a two-step strategy which was subsequently improved to a one-pot reaction by Qiu and Winnik [42]. It involves the “sealing” of the thiol groups by the Michael addition to a BA molecule directly after the aminolysis in the presence of the reducing agent tris(2-carboxyethyl) phosphine hydrochloride. With this procedure the coupling could indeed be prevented. However, the coupling reaction did not occur when polymers with BA segments next to the trithiocarbonate groups were subjected to the aminolysis so that the procedure of Qui and Winnik was not necessary in these cases. The same finding was made by others when cleaving polymethylmethacrylate polymers [46,47]. Xu et al. proposed an intramolecular cyclization reaction of the thiols with the ester groups of the adjacent polymer chain that is faster than the intermolecular coupling as an explanation for this phenomenon [48]. Because of the structural similarity of BA it seems reasonable to suppose that the analogous reaction takes place here.

Figure 5 shows the SEC curves of three styrene multiblock polymers that were prepared using polyfunctional RAFT agents with different average number of trithiocarbonate groups and the SEC curves of the corresponding cleavage products. The numbers in the diagrams indicate the positions of the maxima in the number-weighted molar mass distributions. Polymer 1d (Figure 5(a)) was prepared from trithiocarbonate A with the lowest number of functional groups (1.34, measured via SEC). Individual peaks for molecules with two, three and four blocks with maxima at 45,000, 94,000 and 150,000 g mol⁻¹ are resolved here. The SEC curve of the cleaved polymers shows that they have clearly lower molar masses but still features two individual peaks at 27,000 and 50,000 g mol^{− 1}. This is due to the fact that not all blocks within the multiblock polymers have the same length. Two types of blocks can be distinguished that cannot be interconverted into each other during the polymerization: those in the middle of the polymer chains, connected to trithiocarbonate groups on both sides, and those at the end of the polymer chains, which are connected to only one trithiocarbonate group. The first will be denoted “middle blocks” and the latter “end blocks”. In our theoretical study [37], we could demonstrate that the ratio of the average masses of middle and end blocks lies between 1 and 2, depending on the occurring mechanism, that is, the middle blocks are expected to have higher masses. Thus, the peak at higher molar masses can be assigned to the middle blocks and the one at lower masses to the end blocks. With that assignment, the peak positions of the maxima add up very well, having in mind that every block polymer consists of two end blocks and all other blocks are middle blocks. When cleaving polymers with a higher average number of blocks, the SEC curve of the cleavage products should show a more intense peak at higher molar masses for the middle blocks and a less intense peak for the end blocks. Figure 5(b), which shows the aminolysis of polymer 2d (5.46 blocks in average according to the SEC determined value), shows that this is indeed the case. Again, the positions of the maxima of the number-weighted distributions add up very well here. In the case that polymers with even more blocks are cleaved, the peak for the end blocks is no longer visible. This is demonstrated in Figure 5(c), showing the aminolysis of polymer 5d (18.3 average blocks according to the SEC value), which had a PDI of 2.86. The cleaved polymers by contrast have a PDI of only 1.18, which is lower than that of the other two cleavage products, obviously because there are less end groups which increase the dispersity. These results emphasize that the dispersity of the multiblock polymers is due to the distribution of blocks, rather than the broadness of the mass distributions of the individual blocks. Generally, a correlation of the average number of blocks obtained by aminolysis and the number of blocks in the utilized RAFT agents could be observed, although the first number was always lower than the latter (for low average block numbers even lower than the minimum value of 2). This held also true for the BA multiblock homopolymers.

The aminolysis of the prepared multiblock copolymers leads to the formation of polymers with three segments (the former “middle blocks”) and of polymers with two segments (the former “end blocks”). These reactions also yielded polymers with clearly lower molar masses and the decrease of the masses correlated to the number of trithiocarbonate groups in the initial RAFT agent.

Selected multiblock copolymer samples were preliminarily analyzed via DSC. Figure 6 shows the DSC curves of the polymer samples 6c and Va-Vd. In the curve of the styrene homopolymer 6c (13 average blocks according to the SEC analysis of the employed RAFT agent), a baseline step can be observed that is caused by the glass transition of the polymer. It can be seen that the glass transition temperature (T_g,6c = 344K) is lowered with respect to the one of pure linear polystyrene (T_g,PS = 373 K) [49], which could speculatively be explained by the easier rotatability of the components of the employed RAFT agent that are incorporated along the main chain. With increasing content of poly(n-butyl acrylate), which has a lower glass transition temperature (T_g,PBA = 219 K) [49], the signal in the DSC curves is shifted to lower temperatures. This drop of T_g implies that the material is increasingly plasticized by the increasing amounts of poly(n-butyl acrylate), which is not expected for clean microphase separated polymer domains, for which two constant T_g values should be observable. It should be noted here that the polymer material was not annealed specifically for prolonged times and microphase separation may thus not have evolved in the material. It is well anticipated that the phase separation process of these multiblock copolymers is very slow, due to the multitude of individual segments sitting along a well-entangled polymer chain, which clearly hampers the diffusion of individual segments in the polymer melt.