Styrene-free unsaturated polyester resins derived from itaconic acid curable by cobalt-free accelerators

A series of polyester prepolymers was synthesized from itaconic acid, phthalic anhydride, propane-1,2-diol and diethylene glycol by condensation polymerization. The use of itaconic acid as a source of unsaturation (instead of more common maleic anhydride giving fumarate moieties) enabled to replace styrene reactive diluent by methyl methacrylate. Room temperature curing of a model polyester resin was initiated by butanone peroxide in combination with several cobalt-, iron- and vanadium-based accelerators. Measurements of gelation time and exothermic behavior in thermally isolated installation revealed very promising catalytic properties for oxidovanadium(IV) dibutylphosphate. In follow-up tests, mechanical properties of the model unsaturated polyester resin were tuned by variation in propane-1,2-diol/diethylene glycol ratio and composition of acrylate/methacrylate reactive diluent. Mixtures of methyl methacrylate with secondary crosslinking agents (e.g., ethylene dimethacrylate, triethyleneglycol dimethacrylate and trimethylolpropane triacrylate) enabled to improve ultimate tensile strength, Young’s modulus, tensile toughness and impact toughness. Reported experimental data indicate that the described styrene- and cobalt-free system is very promising for reduction in health and ecological issues of currently used unsaturated polyester resins curable at room temperature.


Introduction
Petroleum-based unsaturated polyester resins (UPRs) are thermosetting systems commonly used in composite industry [1]. Due to low cost, low density and excellent mechanical properties, glass fiber reinforced UPRs have found application as a construction material for boat building, storage tanks and components of wind energy plants [2]. UPR concretes are formulated from UPRs, different aggregates and fillers [3]. Filler-and reinforcing agent-free formulations of UPRs are used as furniture lacquers, adhesives and paints [1].
Classical UPRs are solutions of unsaturated polyester (UP) prepolymers in a reactive diluent (RD). The backbone chain of UP prepolymer usually contains fumarate segments (F), as a source of unsaturation, formed from maleic anhydride by isomerization. So far, styrene (STY) is almost the exclusive RD of choice for UPRs due to high reactivity, wide availability, low cost, low viscosity and rigid structure [4]. Nevertheless, the high reactivity of STY and its thermal instability led to necessity of stabilization by inhibitors to avoid undesired reactions upon handling and storage [5].
STY exhibits a tendency to produce alternating copolymer with UP, documented by reactivity ratios lower than unity (r F = 0.005, r STY = 0.35) [6], which leads to a densely crosslinked polymer network necessary for assembly of durable materials. Ongoing efforts to replace STY in composite industry are motivated by health and environmental issues. European Chemical Agency (ECHA) classifies STY as a species suspected to be toxic to reproduction [7]. According to US Department of Health and Services, it is reasonably anticipated to be a human carcinogen. Moreover, US Environmental Protection Agency identified STY as a hazardous air pollutant [8][9][10]. Replacement of STY is further motivated by increasing concerns regarding to lowering green gas emissions and the fast depletion of fossil-based feedstock [11].
It should be noted that direct replacement of STY in composite industry is not a simple and straightforward process mainly owing to different polymerization behavior of alternative RDs and different mechanical properties of cured UPRs. Bio-based derivatives of styrene and allylbenzene are promising alternatives of STY for bio-based composites even though they show a lower dissolution capacity and give UPRs of higher viscosity [12][13][14]. They further suffer of pure availability and high price since they are in the early stage of development. Vinyl carboxylates, such as vinyl acetate (VAc) and vinyl levulinate (VL), seem to be a promising monomer capable to replace STY, as they have a tendency to produce alternating copolymers with fumarates (r F-= 0.32, r VAc = 0.03; r F = 0.81, r VL = 0.01) [6,15]. Nevertheless, application of vinyl carboxylates in composite industry is very limited by their low reactivity, which often led to partially cured UPR containing a lot of residual RD [15]. The situation is even worse in the case of diallyl phthalate (DAP) and ethyl vinyl ether (EVE). These monomers hardly polymerize through radical polymerization. Moreover, reactivity ratios of both systems (r F = 1.25; r DAP = 0.01; r F = 2.7; r EVE = 0) reveal a tendency to fumarate homopolymerization omitting presence of the RD (Fig. 1) [16,17]. It should be noted that allyl esters (e.g., DAP) are not compatible with peroxide initiators owing to a risk of acrolein formation [18]. Vinyl ethers (e.g., EVE) are sensitive to hydrolysis under acidic conditions, therefore, not suitable for common UPRs with terminal carboxylic groups [19].
Modification of the UP backbone is another strategy to improve copolymerization behavior of the UPmethacrylate pair. For example, replacement of fumarate building blocks by itaconates leads to a more frequent alteration of monomer units, as documented by reactivity ratios for dimethyl itaconate (DMI)-MMA pair (r DMI = 0.42; r MMA = 1.28) [30]. It should be noted that use of the itaconic acid (IA) instead of maleic anhydride increases the bio-based content of the UPR, as IA is readily available from carbohydrates on the industrial scale [31]. Several examples of itaconate-based UPs curable with bioderivable methacrylates have already been reported in literature [32][33][34].
Dialkyl itaconates are other bio-based monomers used as the alternative RDs for polyesters bearing itaconate building blocks [34][35][36][37][38]. In this case, an appearance of a random copolymer is expected, but its application is limited by a lower reactivity of itaconates compared to MMA [39]. (1) where f a is molar fraction of UP unsaturation (feed composition) and F a is molar fraction in copolymer [20]. Gray circles denote ranges of used concentrations.
Curing of UPRs proceeds through radical polymerization that can be initiated by a thermal decomposition of initiators or a photochemical production of initiating radicals. 1 However, the curing process of commercial systems is often initiated by redox systems those usually consist of two components (reductant-oxidant pair) producing initiating radicals upon mixing [40]. Main advantages of this approach, over thermal initiation or decay of photoinitiators by UV-light irradiation, are ability to produce composites under a low energetic consumption, as highly efficient redox systems can perform under mild reaction conditions [41]. Initiating redox system consisting of tertiary amine and benzoyl peroxide, developed in 1950s, is widespread used in composite industry due to sufficiently fast kinetics, tunable properties and facile implementation into various technological processes [42][43][44]. However, necessity of stoichiometric amounts of the components (amine vs. peroxide) and toxicity of the low molecular amines remain its main drawbacks [45] even though the latter issue could be solved by binding of the amine reductant to an acrylic oligomer that prevents its leaching from the polymer [46]. Another way to reach a highly efficient initiating system utilizes a decomposition of organic peroxides by redox-active transition metal compounds. Such ''initiator-accelerator'' systems are capable to produce initiating radicals catalytically even from low reactive peroxides (e.g., hydroperoxides or ketone peroxides). It is documented on cobalt-based accelerators that are commonly used for room temperature curing of UPRs (Scheme 1) [40,47].
Nevertheless, their future is questionable owing to health and ecological issues [48]. Soluble cobalt compounds, including carboxylates used as accelerators, are currently under in-depth evaluation by Cobalt REACH consortium and facing reclassification to class 1b carcinogens, which would tighten up the legislative restriction by ECHA considerably [49]. Furthermore, cobalt has been assessed by European Union as a critical element due to its high economic importance and high supply risk. Combination of these factors with its low recycling rate might lead to deficiency of cobalt on global market in near future [50].
Despite various manganese, iron and copper compounds in combination with peroxides are capable to initiate radical polymerization of UPRs [18]; our previous study has shown that only few of them are enough powerful to perform at room temperature in combination with ketone peroxides [47], which is crucial for their applications in the field of glass fiber reinforced composites [2]. The absence of sufficiently active and ecologically sustainable accelerator, which would effectively substitute the cobalt(II) carboxylates in the field of room temperature curable nonstyrene UPRs, opens the gap for this study.
First part of this study deals with a synthesis of itaconate-based UPR, in which styrene RD is replaced by MMA. This model UPR is then used for investigation of the catalytic activity of different accelerators. The most suitable one, for this type of UPRs, was selected and used at optimal concentration for following optimalization of the UPR composition. Variation in the reactive diluent composition enabled to yield different styrene-free polymers. Their physical properties were compared with itaconate-based UPR using styrene as the RD.

Preparation of UPR
A mixture of monomers IA (260 g, 2.0 mol), PA (296 g, 2.0 mol), PG (Table 1) and DEG (Table 1) and esterification catalyst FASCAT 4201 (Bu 2 SnO; 1 g per 1 kg of monomers) was heated under nitrogen inlet while stirred by anchor stirrer. The reaction mixture becomes homogenous at * 140°C. At 160°C, xylene (* 75 mL) was added, and nitrogen flow was stopped, as solvent vapors begin to produce inert atmosphere. The temperature of the reaction mixture was maintained at 160°C and produced water was distilled off using Dean-Stark apparatus. During the syntheses, samples were collected periodically to follow the acid value (AV) decrease. When AV dropped below 50 mg KOH/g, the reaction mixture was cooled down, and volatiles were vacuum evaporated at 100°C. After cooling to 80°C, the neat UP was diluted with MMA to reach a stock solution of 80 wt.% solid content. Final dilution to 70 wt.% solid content was done by given reactive diluent (or diluent mixture) at room temperature.
Sample diluted with STY was prepared in similar way. The itaconate-based UP was treated with hydroquinone (0.5 g per 1 kg) and cooled to 70°C before dilution. STY was treated with hydroquinone (0.25 g per 1 kg) and precooled to -30°C. After the addition, the mixture was stirred at laboratory temperature for 2 h. Appeared mixture was homogenized by stirring at 80°C and then cooled to room temperature.
Molar concentration of itaconate double bonds (f a ) was calculated as a ratio between number of unsaturations from the UP over number of unsaturations from both the UP and RD components.

Determination of gelation time
Given UPR (12 g) was treated by given accelerator and homogenized by stirring. Commercial solutions of accelerators (Co-C, Co-2EH and Fe-O) and synthesized semi-solid samples (VO-D and VO-PO) were weighted on analytical scales. Solid sample of Fe-B (17.2 mg) was dissolved in UPR (10 g) to give a stock solution (0.01 wt.% of Fe). 1 g of this solution was diluted in with 9 g of UPR to reach second stock solution (0.001 wt.% of Fe), which was used for preparation of final formulations. After accelerator dissolution, MEKP (50 lL) was added, and the mixture was stirred for 10 s. The gelation time was determined on a Techne Gelation Timer GT-5 (Cole-Parmer, UK) using a 16-mm stainless steel plunger according to ISO 2535 [54]. We note that the time measurement started at a point when MEKP was added to the formulation.

Exothermic behavior
Given UPR (12 g) was treated by given accelerator and homogenized by stirring. After that, MEKP (50 lL) was added, and the formulation was stirred for 10 s and inserted into a thermally isolated installation. Temperature development in the central part of the curing formulation was followed by a K-type thermocouple temperature sensor (t 99 = 7 s) using a multifunctional data logger Testo 435 (Testo, Lenzkirch, Germany). We note that zero at the time scale corresponds to a point when MEKP was added to the Scheme 2 Components of redox initiating systems.
formulation. Correction of the exothermic curves on adiabatic conditions and estimation of DT ad , (dT/ dt) max and t max was done as described elsewhere [47].

Preparation of specimen for mechanical testing
Specimen with sizes suited for different mechanical tests was prepared by curing of UPR formulations in silicone molds. Formulations of UPR were treated by VO-PO (12.8 mg per 12 g of UPR), homogenized by stirring, treated by MEKP (50 lL per 12 g of UPR) and stirred again. The initiated formulations were degassed in ultrasound bath (3 min. in degas mode), poured into silicone molds and cured for 24 h at room temperature. The samples in molds were then post-cured for 1 h at 50°C, 1 h at 70°C and 1 h at 90°C. After post-curing, they were slowly cooled down to room temperature. Gel content of post-cured and finely grinded samples was determined by extraction by tetrahydrofuran (THF).

Spectroscopic measurements
1 H NMR spectra were measured at 300 K on a Bruker 500 Avance spectrometer. The chemical shifts are given in ppm relatively to Me 4 Si. Vibration spectra were collected on a Nicolet iS50 FTIR spectrometer equipped with Raman module (Waltham, MA, USA). Infrared spectra were measured using build-in diamond attenuated total reflection crystal in region 4000-400 cm -1 (data spacing = 0.5 cm -1 ). Raman spectra were measured using Nd:YAG excitation laser (k = 1064 nm, power = 0.5 W, data spacing = 1 cm -1 ) in the region 3500-200 cm -1 .

Size exclusion chromatography
The experimental setup for size exclusion chromatography (SEC) consisted of a Waters liquid chromatograph Alliance e2695 with a refractive index detector 2414 and two Agilent Mixed-C columns 300 9 7.5 mm. THF at a flow rate of 1 mL/min was used as the mobile phase. The analyzed samples were prepared as solutions in THF at the concentration of & 3 mg/mL; the obtained solutions were filtered with 0.45-lm filters; the injected volume was 100 lL. The columns were calibrated by narrow polystyrene standards covering the molar mass range of 162 to 6 9 10 6 g/mol. The obtained molar mass averages are relative to polystyrene and cannot be considered to be true molar mass values. However, they can be used for mutual comparison of samples of similar chemical composition and molecular structure.

Dynamic mechanical analysis
The viscoelastic properties of cured samples were determined on a DX045 device (RMI, Czech Republic) in a single-fixed point configuration. Rectangular specimens (50 9 6 9 4 mm) were analyzed in the single cantilever bending geometry (d = 11 mm) using a deviation of -0.15 to 0.15 mm at 1 Hz frequency. The measurements were carried out in the -60 to 150°C temperature range using a heating rate of 3°C/min. Density of crosslinking (m e ) was calculated from DMA diagrams according to Eq. (2): where G 0 is sheer modulus at 140°C, R is gas constant and T is temperature [55,56].

Tensile stress-strain properties
Ultimate tensile strength (r max ), strain upon failure (e f ) and Young's modulus (E) were measured on an Instron Universal Testing Machine 1122 (Bluehill Universal) with a load cell of 5 kN, support span of 22 mm and head speed of 5 mm/min. The measurements were done with flat dog-bone-shaped specimen of 47.5 mm overall length and 14.7 mm width of the grip section (reduced section: 12.5 9 5 9 2 mm) according to ASTM D1708-18 [57].
For accuracy, at least eight specimens were tested for each UPR blend. Tensile toughness (U T ) was calculated from tensile stress-strain diagrams according to Eq. (3): where r is stress, e is strain and e f is strain upon failure.

Impact toughness
The Charpy impact test was carried out on a Plastic Pendulum Impact Tester 501 J-3 (Shenzhen WANCE Testing Machine Co.) using rectangular specimen of 50 9 6 9 4 mm size and a support span of 40 mm according to Č SN 64 0612 [58]. The measurements were done on unnotched specimens. For accuracy, at least eight specimens were tested for each UPR blend.   Table S1.

Synthesis of model UPR
Itaconate-based unsaturated polyester (UP1) was prepared by condensation polymerization of IA, PA, PG and DEG in 1/1/1.2/1 molar ratio (Scheme 3; Table 1). The excess of PG was used to ensure 1/1/1/ 1 stoichiometry of the building blocks, as this monomer is slowly distilled off the reaction mixture together with water, as verified by 1 H NMR spectroscopy (see below). The introduction of PA reduces achievable crosslinking density and forms rigid segments in the resin structure. Flexible aliphatic segments are introduced by DEG. It is noteworthy that the resin is not fully bio-based as only IA and PG are prepared from biomass on the industrial scale [31,34]. PA is bio-derivable via Diels-Alder addition of bio-based maleic anhydride (from furfural) and bio-based furan [59]. Several synthetic routes have been proposed for production of bio-based MMA reactive diluent, but they have not been extended to the industrial scale [25]. The condensation polymerization was catalyzed by [Bu 2 SnO] and done in xylene solution at 160°C until AV of 51 mg KOH/g. Such reaction conditions prevent radical polymerization of the binder even when no inhibitor is used. We note addition of the inhibitor could slow down curing process considerably and reduce activity of accelerators tested below.
Vibration spectra of neat UP1 show a pattern typical for unsaturated polyesters. C=O stretching of ester function appears in the infrared spectrum at 1716 cm -1 as a band of very strong intensity (band h in Fig. 2). Raman spectrum shows several vibration modes characteristic for aromatic rings. They involve C-H stretching at 3075 cm -1 , antisymmetric C=C stretching modes at 1601 and 1582 cm -1 and symmetric C=C stretching mode at 1042 cm -1 (bands c, j, k and q in Fig. 2). Aliphatic chains give bands in region 2995-2880 cm -1 (bands e-g in Fig. 2). Unsaturation of the polyester is evidenced by C=C stretching band appearing in the Raman spectrum at 1645 cm -1 (band i in Fig. 2). Such vibration mode is well resolved even in the infrared spectrum due to polarization of the double bond. OH stretching, appearing in the infrared spectrum (bands a and b in Fig. 2), evidence residual OH and COOH functions not converted to ester functions.
Size-exclusion chromatography (SEC) revealed number average molar mass (M n = 1170 g/mol) and polydispersity (2.47) for polyester UP1. These values imply that average UP molecule contains approximately 1.8 reactive C=C double bonds originating from IA building blocks. 1 H NMR spectroscopy of UP1 in CDCl 3 (Fig. 3) shows well separated signals of PA aromatic ring (d = 7.8 and 7.5 ppm), IA vinylidene group (d = 6.3 and 5.7 ppm) and IA methylene group (d = 3.3 ppm). Methyl group of PG moiety appears as a broad signal at 1.3 ppm. A set of signals at 3.9-3.5 ppm was assigned to methylene groups of DEG neighboring the ether function. The remaining protons of DEG and PG appear as a set of broad signals in the region 5.5-4.0 ppm. Molar ratio of PA/ PG/DEG segments (1/1.07/1.10), calculated from integral intensities of the signals at 7.5, 1.3 and 3.7 ppm, proves that the excess of PG is distilled off upon condensation polymerization together with produced water. The 1 H NMR spectroscopy further revealed a lower IA/PA molar ratio (0.66/1) in  polyester structure than the molar ratio of the feed monomers (1/1). It is ascribed to isomerization of IA to mesaconate (MES) and nucleophilic addition of hydroxyl group to the IA double bond, known as Ordelt reaction (Scheme 4) [60]. Degree of the isomerization was followed by a well separated signal of MES at 6.8 ppm [61]. In the case of UP1, approximately 9% of IA isomerized to lower reactive MES. Consumption of remaining part of IA (25%) is ascribed to the Ordelt reaction. Unfortunately, the extent of Ordelt reaction was hard to track by 1 H NMR spectroscopy owing to overlap of characteristic signals with DEG segments [61]. Therefore, its contribution was calculated as a complement of the determined contents of IA and MES units in feed content of IA.
Dissolution of UP1 in MMA gives a simple model resin UP1-MMA30. The solid content was adjusted to 70 wt.%. At this concentration, viscosity of the formulation is enough low (690 mPaÁs) for following tests of initiating systems. We note that the molar fraction of double bonds from IA segments (f a ) is 0.29 (see Fig. 1 for position in the copolymerization diagram).

Accelerators for model UPR
Effect of various metal-based accelerators (Scheme 2) on the room temperature curing of UP1-MMA30 by the action of butanone peroxide (MEKP) was studied. Initial experiments were done on commercial sample of cobalt(II) carboxylate (Co-C; mixture of carboxylates in ethanol) commonly used for curing of styrene-based UPRs and on a solution of cobalt(II) 2-ethylhexanoate in mineral spirits (Co-2EH).
Treatment of UP1-MMA30 with cobalt-based accelerators gives very long gelation times (Co-C: t gel = 80.9 min, Co-2EH: t gel = 96.1 min) at metal concentration recommended for styrene-based UPRs (0.01 wt.%). It reflects lower reactivity of MMA when compared with STY. Higher accelerator dosage enables to shorten gelation time considerably. At optimal metal concentration (0.09 wt.%), formulations treated by Co-C and Co-2EH give gel after 67.3 and 50.9 min, respectively. Higher dosage of Co-C leads to prolongation of the gelation process probably owing to a presence of protic solvent, which inhibits radical polymerization. In the case of Co-2EH (declared as a solution in mineral spirits), strong inhibition was not observed at similar conditions, but the higher dosage leads only to negligible t gel shortening (Fig. 4). Such experiments imply that variation in cobalt concentration is not capable to adjust t gel below 40 min, which was arbitrarily defined as upper limit of common use.
Exothermic data for UP1-MMA30 formulations were collected in a thermally isolated system and reported after adiabatic correction. Cobalt-based accelerator Co-2EH exhibits satisfactory exothermic behavior at metal concentration 0.09 wt.%, which was established as the optimal by the gelation time measurements (Fig. S1). High temperature rise, upon the polymerization (DT ad = 84.2°C), reflects a high degree of cure. Maximal rate of temperature rise (dT/ dt) max = 18.8°C/min) and time in which the maximum is reached (t max = 90.6 min) indicate a slow cure process, which is in line with long gelation time. Curing of formulation treated with Co-C is slower at 0.09 wt.% as documented by lower (dT/dt) max and longer t max , see Fig. S1.
Two iron-based accelerators were selected for this study; commercial sample of iron-bispidine complex, known under trademark Oxy-Coat (Fe-O; solution in PG), and its derivative with increased solubility in non-polar solvents (Fe-B; neat) [51]. Both compounds are applicable in a wide range of concentrations (Fe-O: 1.5 9 10 -5 -1 9 10 -3 wt.%; Fe-B: 2 9 10 -5 -1 9 10 -3 wt.%) where t gel is shorter than 40 min. Due to similar molecular structure, Fe-O and Fe-B show a very similar behavior at metal concentrations below 1 9 10 -4 wt.%. At higher dosage, protic solvent in Fe-O inhibits the curing process leading to prolongation of t gel . In the case of Fe-B, t gel stays at its lower limit (* 20 min) under similar condition (Fig. 4).
Effect of Fe-O and Fe-B on exothermic behavior of UP1-MMA30 was investigated at metal concentrations 1 9 10 -4 and 5 9 10 -4 wt.%. Although these formulations exhibit much shorter lag time, their curing proceeds very sluggishly as documented by low values of (dT/dt) max ( Figure S2). At 1 9 10 -4 wt.%, exothermic curves of Fe-O and Fe-B are developed in a very similar way as documented by similar values of (dT/dt) max * 0.5°C/min and t max * 110 min. Higher dosage of the iron-based accelerators leads to earlier rise of Fe-B as evident from lower value of t max , which well correlates with gelation time measurements. To our surprise; however, maximal rate of the temperature rise stays very   [49]. g Crosslinking density [55,56]. h Gel content. similar for both formulations ((dT/dt) max * 1.6°C/ min). Our scrutiny of vanadium-based accelerators was focused in oxidovanadium(IV) dodecylbenzenesulfonate (VO-D) and oxidovanadium(IV) dibutylphosphate (VO-PO). We note that oxidovanadium(IV) toluenesulfonate, recently established as a very promising accelerator for styrene-based UPRs [47], is not covered in this study owing to insufficient solubility in UP1-MMA30.
At the metal concentration range 0.03-0.1 wt.%, VO-D gives gelation times shorter than 5 min, which was arbitrarily defined as lower limit of common use. Surprisingly, the concentration dependence is very steep (Fig. 4). Hence, lowering of metal concentration below 0.02 wt.% leads to deterioration of the accelerator activity. Exothermic curve of VO-D, collected at 0.03 wt.%, shows a slow temperature rise already from beginning of the experiment. After 11.4 min of curing, rate of the temperature rise reaches the maximum; (dT/dt) max = 15.8°C/min.
To our surprise, the accelerator VO-PO exhibits a very different dependence of gelation time on concentration when compared with VO-D (Fig. 4). The curve has much less steep slope, which makes adjustment of the gelation time by variation in the accelerator dosage much easier. We imply metal concentration range 0.01-0.03 wt.% as optimal dosage for formulations of VO-PO, since t gel fits into afore established limits of common use (5-40 min).
Strong catalytic activity of VO-PO and its mild concentration dependence enable to measure exotherms in a wide range of metal concentrations (0.01-0.09 wt.%). These measurements have shown that (dT/dt) max rises with rising concentration up to 81.5°C/min at 0.07 wt.%, which occurs within 5.1 min of curing (Fig. 5). At higher concentration, (dT/dt) max starts to drop due to overdose effect. It is noteworthy that exothermic behavior of VO-PO at 0.01 wt.% is very similar to Co-2EH at 0.09 wt.%; both formulations show similar values of (dT/dt) max and t max .
In summary, each accelerator exhibits a different behavior in the formulation MEKP/UP1-MMA30. Their activity can be quantified by several different ways. Given accelerator can be regarded as more powerful when it is able to reach shorter t gel . Hence, each compounds reaches characteristic lower limit values for this parameter. According to this approach, activity of here presented accelerators rises in the order: A more convenient approach defines accelerator as more active when it provides a given value of t gel (usually optimal for given application) at lower concentration. According to this approach, the activity rises in a different order: Of course, catalytic activity of accelerators is also related with exothermic behavior if their formulations. In this case, it is necessary to determine a concentration, at which (dT/dt) max reaches maximal value. Such concentration and related (dT/dt) max and t max values can be then used for comparison of their activity. Although such parameters were determined are only for VO-PO, the collected exothermic curves imply a rise of catalytic activity in the order Co-C \ Co-2EH \ \ VO-D \ VO-PO. Based on these considerations, VO-PO at 0.01 wt.% was chosen for room-temperature curing of all specimens necessary for following study of mechanical testing of synthesized non-styrene UPRs. We note that all specimens were post-cured at elevated temperature to reach full conversion of the polymerization process.

Optimization of UPR composition
Cured model resin UP1-MMA30 contains 86.6% of gel, which implies high degree of crosslinking. The THF extract consists of UP with much lower degree of unsaturation than original UP1, as revealed by infrared and Raman spectroscopy (Fig. S5). It implies a high content of chains without IA units those, in principle, cannot be incorporated into the crosslinked network. DMA analysis of cured UP1-MMA30 verified occurrence of crosslinked network, enabled to determine temperature of glass transition (T g = 75.2°C) and quantified density of crosslinking (m e = 0.48 molÁm -3 ).
As expected, styrene analogue UP1-STY30, prepared by dissolution of UP1 in the same mass of STY, exhibits higher T g value (94.4°C) due to rigid aromatic segments. In contrary to UP1-STY30, the dumping factor peak is much broader and unsymmetric with a shoulder at * 70°C (Fig. 6), which implies formation of heterogenous polymer network [28]. Nevertheless, observation of fracture surfaces by SEM implies absence of large inhomogeneities in the sample (Fig. S8). We note that molar fraction of UP unsaturation is very similar for UP1-MMA30 (f a-= 0.29) and UP1-STY30 (f a = 0.29). Values for the other formulations under the study vary in narrow range 0.28-0.36 (Table S4).
To improve mechanical properties of the itaconatebased resin, MMA in UP1-MMA30 was partially replaced by several readily available acrylate/ methacrylate-based RDs (Scheme 5). Acronym of ternary blends contains name and weight content of the second RD only; overall content of reactive diluents stays 30 wt.% in all cases (for details see Table S4).
Plasticizing effect of MA is documented on the formulation UP1-MA10, containing MMA and MA in the 2/1 weight ratio. Dynamic mechanical analysis (DMA) reveals a drop of the glass transition temperate (T g ) from 75.2 to 66.0°C (Fig. 6). Similar intensity of damping factor (tan d) peaks is in line with similar density of crosslinking (m e ; Table 2). As expected, EHMA has much stronger effect on T g value, as it contains a long and benched alkyl tail. Hence, a formulation containing MMA and EHMA in the 1/5 weight ratio (UP1-EHMA5) exhibits similar T g value (68.3°C) as UP1-MA10. We note that gel content of cured formulations UP1-MMA30, UP1-MA10 and UP1-EHMA5 is very similar; it fits into the Scheme 5 Reactive diluents used in this study. range 85.5-87.0% and nears the values determined for styrene-based formulation UP1-STY30 (86.5%).
Lower T g value of UP1-MA10 has only a minor effect on the tensile stress-strain curve (Fig. 7). Difference in r max and e f values nears the experimental error. On the contrary, the use of EHMA leads to a considerably lower tensile toughness (U T ) due to much lower ultimate tensile strength (r max ). Elongation at break (e f ) is not influenced in this case. Charpy impact test revealed that the modification with MA improved impact resistance of cured UPR considerably (a cU = 4.1 ± 0.6 kJÁm -2 ) while effect of EHMA is negligible ( Table 2).
Monomers bearing two methacrylate functions (EDMA and TEGDMA) and three acrylate functions (TMPTA) were chosen as a modifying RD due to ability to perform as secondary crosslinking agents (Scheme 5).
Increasing content of these secondary RD leads to higher gel content in the cured samples and well correlates with higher degree of crosslinking ( Table 2). The DMA analysis documents a strong effect of EDMA on viscoelastic properties of cured samples. Very short spacer between methacrylate functions causes a considerable decrease in dumping factor peak (tan d) and increase in storage modulus (E') at temperatures above T g already at 2 wt.% EDMA content (UP1-EDMA2). Such effect is even more distinct at higher EDMA contents. This behavior proves that the polymer network becomes more compact and tighter, as documented by higher number of crosslinks calculated from storage modulus at rubbery plateau (see, m e in Table 2) [55,56].
The blends containing EDMA exhibit considerably higher ultimate tensile strength (r max ) than UP1-MMA30, but elongation at break (e f ) decreases already at 2 wt.% content. High tensile toughness is observed only for blends UP1-EDMA2 and UP1-EDMA5 (see U T in Table 2). Increased content of EDMA leads to more brittle materials as documented by absence of necking area in stress-strain diagrams of UP1-EDMA5 and UP1-EDMA10 (Fig. 7).
Cured blends containing TMPTA, a monomer with three acrylate functions and short rigid spacer, exhibit similar development of DMA curves as those treated with EDMA at the same dosage (Fig. 6), which indicates comparable viscoelastic properties. At 2 wt.% dosage, TMPTA has similar effect on stress-strain curves as EDMA. At higher TMPTA content, however, the material becomes more brittle as documented by lower r max and e f values (Fig. 7, Table 2).
Introduction of flexible spacer into structure of crosslinking RD enables to improve material toughness, as documented on formulation UP1-TEGDMA5 (U T = 4.5 ± 0.2 MJÁm -3 ). The cured material breaks at a higher strain (e f = 12.9 ± 0.5%) than UPMMA30; r max value falls between UP1-MMA30 and UP1-EDMA5. Such behavior is in line with a lower effect of TEGDMA concentration on density of crosslinking ( Table 2).
All secondary crosslinking agents mentioned afore enhance impact resistance of the cured UP1-MMA30 (Table 2). Strong effect is observed mainly for the blends treated with EDMA. At 2 wt.% dosage, the impact toughness reaches the maximum (a cU = 5.0 ± 0.6 kJÁm -2 ).
Properties of the cured UPR were further tuned by variation in DEG/PG ratio. We decided to lower the content of flexible DEG segments to increase tensile strength of the cured polymer. As documented by SEC, partial replacement of DEG segments by PG has only a minor effect on polymerization degree. Number-average molar mass of modified polyesters UP2-UP4 varies between 1120 and 1310 g/mol. 1 H NMR spectra of these prepolymers reveal only minor variation in the IA/MES/PA ratio. The content of intact IA double bonds slowly decreases with increasing PG content, but the IA/PA ratio does not drop below 0.62/1 (Fig. 3, Table1).
As expected, cured blends with a lower content of flexible DEG segments exhibit higher T g value than UP1-MMA30, but it they did not reach a value observed to styrene-based resin UP1-STY30 (T g-= 94.4°C). Stronger tan d peaks (Fig. 8) are due to looser polymer network in the modified polyesters (see m e in Table 2), which is ascribed to lower content of intact IA double bonds. Ultimate tensile strength (r max ) of the cured formulations rises in the order UP1-MMA30 \ UP2-MMA30 \ UP3-MMA30. Further decrease in DEG content leads to a more brittle material, as documented by absence of necking area in stress-strain diagram (Fig. 8) and lower values of r max , e f and U T ( Table 2). Minor effect of DEG/PG ratio on impact toughness is documented by Charpy test ( Table 2).
Modification of the formulations of UP2 and UP3 by the secondary crosslinking agents enabled to improve their mechanical properties considerably. DMA analysis proved increased T g value and appearance of considerably tighter network in blends containing 2 wt.% of EDMA and 2 wt.% of TMPTA (Table 2). Blends containing 5 wt.% of TEGMA show only a minor decrease in dumping factor peak and similar T g value as parent blend without the secondary crosslinking agent ( Table 2, Fig. S4).

Conclusions
This study has clearly shown that STY can be easily replaced in common UPRs by ecologically more acceptable MMA when fumarate segments in UP prepolymer backbone are replaced by itaconates. Superior properties of cured itaconate-based UPRs are ascribed to favorable copolymerization parameters compared to F/MMA pair.
Mechanical properties of the cured UPRs can be easily tuned for given application, as exemplified on UP prepolymers with modified backbone and on variation in acrylate/methacrylate-based RD composition. Improved ultimate tensile strength and impact toughness were gained in the case of MMA mixtures with a low content of secondary crosslinking agent (e.g., EDMA, TMPTA and TEGDMA), which provides tighter and more compact polymer network.
Ability of the UPR system to be cured at room temperature was verified on initiating system consisting of MEKP and several transition metal-based accelerators. Our experiments imply VO-PO as preferred accelerator due to optimal exothermic behavior, lower dosage than in case of cobalt-based accelerators and its ability to reach shorter gelation times than iron-based accelerators.
In summary, here described styrene-and cobaltfree system has a great potential to reduce health and ecological issues of currently used unsaturated polyester resins including those curable at room temperature.

Data availability
Data will be made available on request.

Declarations
Conflict of interest Authors declare no conflict of interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licen ses/by/4.0/.