Technological aspects of synthesis of poly(ethylene glycol) mono-1-propenyl ether monomers

Abstract For the first time, the technological aspects of the highly productive and selective synthesis of UV-reactive poly(ethylene glycol) mono-1-propenyl ether monomers was developed. The solvent-free isomerization of model commercial available 2-allyloxyethanol and allyloxypoly(ethylene glycol) derivatives, type Allyl–[OCH2CH2]n–OH, n = 1–5, into a 1-propenyl derivative under the homogeneous catalysis conditions using the ruthenium complexes were evaluated. The effect of a various reaction conditions (i.e. the concentration of [Ru] complex, the reaction temperature, reaction gas atmosphere) together with trace amounts of allyl hydroperoxides formed via autoxidation reaction of allyl substrates on the productivity of catalyst was examined in detail. Moreover, the significant role of the allyl substrate structures on the catalytic activity of ruthenium catalysts were also recognized. The optimal parameters of the scaled-up synthesis together with productivity of catalyst were first established.


INTRODUCTION
Due to an increase of the environmental restrictions, the technology of UV-initiated radical or cationic polymerizations have attracted an increasing attention from years. These solvent-free and low-energy processes have employed in many thin fi lm industrial applications such as coatings, adhesives, printing ink and many others because they could essentially eliminate air and water pollution 1-3 . The cationic photopolymerization process presents some industrial important advantages over the radical one, however, the growth of popularity of this method is hindered by the lack of easily available and highly reactive monomers, especially multifunctional ones. The most popular cationic UV-polymerizable resins are cycloaliphatic epoxides, vinyl ethers, oxetanes, oxazolines and others 1-3 . However, recently the structurally similar compounds, namely 1-propenyl ether monomers have been developed as a promising alternative to the vinyl monomer especially due to their high reactivity in photocurable applications 1-4 and very attractive from technological point of view methodology of the synthesis via isomerization of corresponding allyl ethers catalysed by ruthenium complexes under homogeneous conditions (Scheme 1) 5-7 . This strategy with using readily available of allyl ethers of alcohols, diols and polyols from both natural and synthetic sources and [RuCl 2 (PPh 3 ) 3 ] was originally initiated and investigated by Crivello group 4 and postulated as easy and versatile method of synthesis of the UV-reactive mono-, di-and multifunctional 1-propenyl ether monomers. Our work in this area focused on wide and comprehensive studies of the reactions of multi-functional O-allyl systems with [Ru] complexes together with technological aspects of this processes [8][9][10][11][12][13][14] and also application of 1-propenyl ether monomers in UV curing polymeric systems 15-19 . The 1-propenyl ethers bearing free hydroxyl groups of CH 3 CH=CH-O-A-OH type (hydroxyalkyl 1-propenyl ethers, 1-propenyloxyalcohols) are the most desire as the reactive diluents for photopolymerizable systems with enhanced reactivity 15 or intermediates for the synthesis of hybrid monomers for special applications 1, 3 . It is well-known that the structure of A linker has decisive infl uence on its physicochemical properties, the course and rate photopolymerization, and also the structure and resulting properties of the cross-linked polymers 1, 3 . In this work, the low-molecular weight poly(ethylene glycol) (PEG) with precisely-defi ned numbers of PEG units were used as A linker. Generally, PEG is an industrially produced, relatively inexpensive, low or high molecular weight hydrophilic linear oligomer or polymer with a widely technical and medical applications 20-23 . Due to its excellent solubility in aqueous and organic media, fl exibility of the main chain, non-toxicity and biocompatibility, the incorporation of PGE into 1-propenyl ether molecule give way for the commercial utility of these product as valuable fi ne chemicals for special applications.
As was mentioned above, the catalytic isomerization of allyloxyalcohol substrates is recognized as the best method of 1-propenyloxyalcohol synthesis. The low-molecularweight poly(ethylene glycol) monoallyl ethers (APEG) with defi ned-length PGE are commercially produced via ethoxylation of allyl alcohol by ethylene oxide at the presence of alkaline catalysts such as NaOH KOH or Na 2 CO 3 20 . Despite 2-allyloxyethanol is the simplest example of an ethoxylated alcohol containing an allyl ether group, only a few reports of its isomerization have been found in the literature. For example the using Pd/C (5%) as the heterogeneous palladium catalyst gave only 41% yield of the

The general methodology of isomerization reaction
The isomerization reactions were conducted under solvent-free conditions and under the air or argon atmosphere using the Schlenk method according to the procedure described by us previously 12-14 . The allyl substrate (5 mmole) was placed into screw-capped glass ampoules together with ruthenium complex and stirred for given period time (the reaction conditions are given in footnotes of Table 1 and 2). In turn, all experiments in preparative scale were conducted using 0.3 mole (31 g) of 2-allyloxyethanol in a round-bottom fl ask with a red rubber septa (Sigma-Aldrich) and a magnetic stirrer under the argon atmosphere. The samples were taken at various times during the experiment using a glass syringe and analyzed.
The isomerization reactions were monitored by gas chromatography (Trace Ultra GC (Thermo Scientifi c), column Rtx-1 with length: 30 m, diameter: 0.53 mm, thickness of the fi lm: 1,5 μm (Restek) and FID detector) or/and 1 H NMR (Bruker DPX-400 spectrometer in CDCl 3 containing TMS as the chemical shift standard, and K 2 CO 3 as the HCl scavenger) or/and GC-MS analyses (optional) (gas chromatograph Hewlett Packard 6890 coupled to a mass spectrometer HP 5973 with electron impact ionization). The main 1-propenyloxyalcohol products obtained as an mixture of Z and E isomers were purifi ed via vacuum distillation above ruthenium catalyst (for spectroscopic data see reference 14 ).
The isomerization progress was evaluated in terms of the allyl group conversion (α AE ), the selectivity of the reaction to 1-propenyloxyalcohol (S PA ,) together with the stereoselectivity given as the percentage of E-1-propenyl group (%E) and the selectivity of the reaction to products of intra-and/or intermolecular addition of an OH group to the double bond (S AP ). Moreover, the activity of [Ru] pre-catalyst, determined as the turnover number (TON) and the turnover frequency (TOF, h -1 ) was calculated according to the following formulas: TON = n p /n [Ru] 3 ], {[RuCl 2 (p-cymene)] 2 }) as homogeneous pre-catalysts, under solvent-free oxidative or non--oxidative conditions. The isomerization reactions were conducted at the concentration of [Ru] complex of 0.1 mol% and at moderate temperature of 80 o C or 120 o C under argon or air atmosphere. Moreover, the possibility of using allyl substrate containing allyl hydroperoxides (AHP) and without additional purifi cation was also tested as one of the key technological aspects of this process. The results for Allyl-PEG 1 -OH and Allyl-PEG 2 -OH were collected in Table 1 and Table 2, respectively.
Firstly in should be pointed that the reaction of 2-allyloxyethanol and 2-[2-(allyloxy)etoxy]ethanol with [Ru] complexes run exclusively towards the isomerization products i.e. 1-propenyloxyalcohols given as mixture of Z and E isomers of 2-(1-propenyloxy)ethanol and 2-[2-(1-propenyloxy)ethoxy]ethanol) and/or towards the intramolecular addition of an OH group to the double bond (cyclization) products given as 2-ethylacetals i.e. 2-ethyl-1,3-dioxolane and 2-ethyl [1,3,6]trioxane, respectively (Scheme 3). Secondly, the results revealed in Table 1 and Table 2 have clearly shown that the catalytic activity of [Ru] complexes varied dramatically depending on the allyl substrate structure, the pre-catalyst used and also the reaction conditions, especially the reaction temperature. Moreover, the presence of the oxidative factors i.e. the allyl hydroperoxide (the allyl substrate impurities) or the oxygen from the air have also a signifi cant effect on the activity of the catalyst but they no change essentially the reaction chemoselectivity towards 1-propenyl ethers or acetals. And fi nally, the comparison the results obtained catalyst; y p -the yield of 1-propenyloxyalcohol product; τ -the reaction time.

RESULTS AND DISCUSION
Generally, the diffi culty of the conducting high selective and yielded the isomerization of the allyloxyalcohols at the presence of [Ru] complexes is connected with the fact that this reaction could leads to 1-propenyl derivatives and also products of the intra-and/or intermolecul ar addition of an OH group to the double bond, depending on kind of [Ru] catalyst and the temperature of reaction 10-12 (Scheme 2). Moreover, in the case of the 2-allyloxyethanol and its allyloxypoly(ethylene glycol) derivatives the activity of ruthenium catalysts and the relative rates of isomerization reactions could be strongly affected both by temporary chelation of ruthenium by the oxygen atoms from ethoxy groups (or/and C=C of η 3 -allylic bond) (postulated for the diallyl ethers in the literature previously 26 , including our work 9 ) and also by the availability of the OH-group to formation (or reproduction) of the hydride or dihydride active species of the ruthenium catalyst. On the other hand, the allyloxypoly(ethylene glycol) derivatives type Allyl-[OCH 2 CH 2 ] n -OH, n = 1-5 are very good allyl substrates for the [Ru]-catalyzed isomerization due to fact, that they undergo the autoxidation reaction very slowly or they do not practically react with oxygen during storage of a pure substrate. Therefore, the concentration of allyl hydroperoxides (AHP), which often cause decrease or even deactivate of the [Ru] catalyst is ultra-traces. For example, in the case of Allyl-PEG 3 -OH only 5-10 ppm  3 ] were selected for the isomerization of other tri-, tetra and penta(ethylene glycol) monoallyl ethers (Allyl-PEG n -OH, n = 3,4,5) under the same reaction conditions. However, it should be noted that the autoxidation reaction of those allyloxyalcohols with forming allyl hydroperoxides at the desired concentration of 25 ppm proceeded very slow during storage of a pure substrate. As was mentioned above, in the case of Allyl-PEG 3 -OH only 5-10 ppm of AHP were detected within 10 months of storage under refrigeration and air atmosphere. For accelerating the autoxidation reaction and obtaining the Allyl-PEG 3 -OH containing 25 ppm of allyl hydroperoxides, the sonification method at frequencies of 40 kHz within 3 days was successfully used. However, Allyl-PEG 4 -OH and Allyl-PEG 5 -OH practically did not undergo this reaction even under ultrasound conditions within 10 days. So, the reaction of both Allyl-PEG 4 -OH and Allyl-PEG 5 3 ] exhibited very high activity (the allyl group conversions was 100%) but low selectivity toward 1-propenyl products. However, it is also important to noted that in the presence of showed also that in general 2-[2-(allyloxy)etoxy]ethanol is a much more reactive as 2-allyloxyethanol.
So, in the case of Allyl-PEG 1 -OH and Allyl-PEG 2 -OH at the reaction temperature of 80 o C only [RuClH(CO) (PPh 3 ) 3 ] appeared to be the most active and selective pre-catalyst -the reaction proceeded with a quantitative yield of 1-propenyl ether derivatives within a very short reaction time of 0.5 h ( Table 1 and Table 2). What is important, regardless of the oxidative reaction conditions the activity and the chemoselectivity of reaction was maintained for prolonged time of 2 h. However, at the higher temperature of 120 o C, reaction of Allyl-PEG 1 -OH with [RuClH(CO)(PPh 3 ) 3 ] was stopped on the isomerization step in contrast to Allyl-PEG 2 -OH, where the 2-ethyl [1,3,6]trioxane with the relatively high yields from 2 to 13% was observed. Remarkably, the very fast and selective isomerization towards the 1-propenyl derivative was also detected for [RuCl 2 (PPh 3 ) 3 ] -at the temperature of 80 o C the reaction time of 0.5 h was suffi cient to obtain the yield of 100% of 2-(1-propenyloxy)ethanol under non-oxidative reaction conditions (Table 1). However, at the higher temperature of 120 o C and in the case of 2-[2-(allyloxy)etoxy]ethanol and the reaction selectivity was unacceptable -the mixture of 1-propenylalcohol and cyclic acetal was observed ( Table 2). The use of the oxidative conditions practically did not affect the selectivity values. Moreover, the most signifi cant differences in the reactivity of allyl substrates tested were revealed for less active ruthenium complexes such as hydride [RuClH(CO) (AsPh 3

) 3 ] or dihydride [RuH 2 (CO)(PPh 3 ) 3 ] and [Ru-H 2 (PPh 3 ) 4 ]
. Surprisingly, these pre-catalysts exhibited an unexpectedly low activity in the 2-(1-propenyloxy) ethanol synthesis, especially at lower temperature of 80 o C (Table 1). Interestingly, the introduction into hydride ruthenium molecule of [RuClH(CO)L 3 ] of triphenylarsine ligand instead triphenylphosphine caused not only lower yield of 1-propenyl product up to ca. 73% but also a substantial changes in the stereoselectivity of isomerization (the thermodynamically favored E-isomer was the major product, %E amounted ca. 80-90% at temperature of 80 o C) and also in the selectivity of cyclization (the yield of 2-ethyl-1,3-dioxolane equaled even 65% at temperature of 120 o C and at the presence of AHP at the concentration as low as 25 ppm) ( Table 1). The similar trend was observed for Allyl-PEG 2 -OH/[RuClH(CO)(AsPh 3 ) 3 ] reaction system, however the practically quantitative conversion values were noted (Table 2). In turn, the using dihydride [Ru] complexes, especially [RuH 2 (CO) (PPh 3 ) 3 ] as extremely selective isomerization pre-catalysts are the more technologically attractive catalytic systems of the 1-propenyloxyalcohols synthesis but they require the application of the higher temperatures of reaction ased with the length of PEG chain under both argon and air atmosphere. However, the prolongated reaction time of 2 h was enough to obtain a 100% yield values of 1-propenyloxyalcohols. On the other hand, at 120 o C the activity of [RuClH(CO)(PPh 3 ) 3 ] was practically the same, independently of the allyl substrate and the reaction gas atmosphere. The quantitative values of allyl group conversion and the yield of 1-propenyl products were determined but after 2 h of processes the values of selectivity of isomerization dropped from 100% to ca. 93-98% (Fig. 2).
The selected results of catalytic activity of [RuH 2 (CO) (PPh 3 ) 3 ] in the isomerization of allyloxy tri-, tetra and penta(ethylene glycol) derivatives under non-oxidative conditions at the reaction temperature of 80 o C and 120 o C were presented on Figure 3 (for clarifi cation, the values of yield of 1-propenyloxyalcohols under air atmosphere were slightly lower than under argon atmosphere, especially during the initial reaction period, so, for this reason these results were disregarded on the graph). The comparison of 1-propenyloxyalcohols yield values shows [RuCl 2 (PPh 3 ) 3 ] the oligomeric linear acetal products of intermolecular addition of an OH group to the double bond together with oligomerization reaction were formed even at the lower temperature of 80 o C. Thus, the popular ruthenium complex i.e. neat [RuCl 2 (PPh 3 ) 3 ] can be not recommended for the reaction of these high molecular weight allyloxyalcohol substrates. Figure 1 and Figure 2 present the values of 1-propenyoxyalcohols yield formed during isomerization reaction of AHP-free allyl substrates type Allyl-PEG n -OH, where n = 3,4,5 catalyzed by 0.1 mol% [RuClH(CO)(PPh 3 ) 3 ] under argon or air atmosphere at temperature of 80 o C or 120 o C, respectively (the results for the Allyl-PEG 1 -OH and Allyl-PEG 2 -OH given in Table 1 and Table 2 were also added for comparison). As can be seen in Figure 1, the hydride complex [RuClH(CO)(PPh 3 ) 3 ] exhibits very high activity and selectivity in the isomerization of all tested allyloxy tri-, tetra and penta(ethylene glycol) derivatives, but in the fi rst step of the reaction at the temperature of 80 o C after a short reaction time of 0.5 h the values of allyl substrate conversion gradually decre-  ethoxy]ethanol undergoes very fast isomerization reaction with a quantitative yield of corresponding 1-propenyloxyalcohol. Therefore, it is clear that the structure of allyl substrate (here − the number of PEG groups in the molecule only) has a crucial infl uence on the reaction rate, especially in the fi rst step of the reaction up to 30 min. and at lower temperature of 80 o C. Moreover, it is reasonable to suggest that the temporary decreasing of isomerization rate is connected with the chelation of ruthenium by the oxygen atoms from ether and the OH-group and also by the double bond carbon atoms. To confi rm our hypothesis the minimum energy structure of tested poly(ethylene glycol) mono allyl ethers can be determined from MM calculation using the MM2 program 27 (Fig. 4) together with models of plausible ruthenium complex with them temporary formed during reaction 28 (Fig. 5).
As can be seen, the low reactivity of 2-allyloxyethanol could be fi rst explain by relatively strong coordination of Ru by oxygen atoms with forming temporarily stable fi ve-member ring, additionally stabilized by Ru-(C=C) bond from allyl or 1-propenyl group (3-dentate structure) (Fig. 5). In turn, in the case of 2-[2-(allyloxy)ethoxy] ethanol simultaneous oxygen atoms chelation together with C=C bond to form 3-or 4-dentate structure of eight-member ring has relatively low probability and in consequence the isomerization rate substantially increase   26 and also in our papers 9, 12-14 that the double bond migration reaction catalysed by hydride transition metals complexes dominantly runs via the hydride addition -β elimination mechanism (1,2-hydrogen shift) with the preferential Z isomer formation. Additionally, the hydroxyl group of the allyloxyalcohol substrate (functional primary alkyl alcohol) might play the key role in the formation (or reproduction) of the hydride or dihydride active species of the catalyst in situ. Thus, in contrast to other tested allyloxyalcohols, the very easily availability of the -OH group to Ru atom in 2-[2-(allyloxy)ethoxy]ethanol molecule maintains or increases the activity of ruthenium hydride or dihydride catalyst and, in this way, contribute to acceleration of isomerization (Fig. 5).
Considering the technological aspects of the effi cient high-yielded synthesis of 1-propenyloxyalcohols functionalized by linear poly(ethylene glycol) (PEG) chain some experiments were undertaken including the deter-mination of the optimized reaction conditions with the high effectiveness of Ru catalyst, the possibility of using allyl substrate without additional purifi cation (containing AHP) and also the scaling up of process. For this purpose, the solvent-free isomerization of the 2-allyloxyethanol catalyzed by a particularly active [RuClH(CO)(PPh 3 ) 3 ] was selected as a model reaction system. All experiments were conducted using 0.3 mole of 2-allyloxyethanol. The reaction was investigated in the range of minimal pre--catalyst concentration of 0.01-0.05 mol% under non--oxidative reaction conditions or when the allyl substrate contained the traces of allyl hydroperoxides at the level of ca. 10 ppm (we had established that only a trace amount ca. 5-10 ppm of the AHP was detected during storage of a pure substrate during ca. 3 months). The reaction was conducted in a reasonable time to achieve the highest possible yield of 1-propenyl product. The selected results are collected in Table 3.
As can be seen, [RuClH(CO)(PPh 3 ) 3 ] showed an excellent selectivity and a relatively high catalytic activity in the studied reaction under non-oxidative reaction conditions and at minimal loading of 0.01 mol%. A high TON values of approximately 10 000 were obtained both at the moderate temperature of 80 o C and also at the higher temperature of 120 o C (entries 1 and 2 in Table 3, respectively). However, the time required to complete the reaction was three times longer at the lower temperature, than at the higher one, giving the highest TOF values of 6 667 h -1 at 80 o C vs. 20 000 h -1 at 120 o C. Moreover, the presence of AHP traces dramatically reduced the progress of the isomerization regardless of the reaction temperature (entries 3 and 4). Depending on the temperature of 80 o C or 120 o C, the isomerization practically stopped after 4 h or 5 h when only ca. 55% or 80% of the allyl substrate was converted. Nevertheless, it was found that a relatively low pre-catalyst loading of 0.025 mol% at lower temperature of 80 o C was suffi cient to eliminate completely the effect of allyl hydroperoxides on the isomerization rate and the same values of TON and TOF (4 000 and 4 000 h -1 , respectively) were obtained (entry 6). In the case of AHP-free allyl substrate the reaction occurs very fast with high TOF values of 8 000 h -1 (entry 5). Although the higher activity of [RuClH(CO)(PPh 3 ) 3 ] noted it is important to emphasized that further raising of the pre-catalyst amount to 0.05 mol% is unreasonable from the technological point of view (entries 7 and 8).
On the other hand, it was interestingly to verify the catalytic activity of cheaper and well-known complex [RuCl 2 (PPh 3 ) 3 ] in the isomerization of 2-allyloxyethanol. As was demonstrated in the screening research given in Table 1, at the reaction temperature of 80 o C under non--oxidative conditions the 100% yield of 1-propenylethanol was noted after only 30 min. (TON = 1 000, TOF = 2 000 h -1 ). Surprisingly, contrary to [RuClH(CO)(PPh 3 ) 3 ], [RuCl 2 (PPh 3 ) 3 ] pre-catalyst using a 2-fold lower amount below 0.1 mol% exhibited an unexpectedly low activity in the 1-propenylethanol synthesis (entries 9-12 in Table 3). The best results with the quantitative conversion of allyl substrate were obtained at the [Ru] concentration of 0.05 mol% at 80 o C within 2.5 h but the product yield was slightly lower (ca. 99%) due to the traces of cyclic acetal (2-ethyl-1,3-dioxolane) formed during prolonged reaction times. The TON value equaled 1 980, but TOF value was only 792 h -1 (entry 10). In turn, in the presence of a lower [Ru] loading of 0.025 mol% the prolongation of the reaction time was required to obtain the high values of allyl group conversion but simultaneously contributed to considerable decrease of the isomerization selectivity (entry 9). Thus, according to our   3 ] exhibited a high catalytic productivity with TON of 10 000 and TOF of 20 000 h -1 in the solvent-free isomerization of low reactive 2-allyloxyethanol. Due to fact that for large-scale production of medium-value fi ne chemicals the TON values between 10 000 and 50 000 and TOF > 10 000 h -1 are required 29-30 , [RuClH(CO)(PPh 3 ) 3 ] can be recommended for an effi cient and selective scaled up synthesis of 2-(1-propenyloxy)ethanol.

CONCLUSIONS
The catalytic double bond isomerization is an important atom-effi cient, one-pot and solvent-less reaction fulfi lling both sustainability criteria and a widespread application in the industry. Particularly effective catalysts for this reaction are soluble ruthenium complexes. However, the information regarding the optimal and technologically attractive reaction conditions are still very scarce in the literature. In this work, we have chosen commercially available poly(ethylene glycol) monoallyl ethers with well-defi ned PGE units from 1 to 5 as allyloxyalcohol substrates for obtaining 1-propenyloxyalcohols type CH 3 -CH=CH-[OCH 2 CH 2 ] n -OH, n = 1-5, as valuable fi ne chemicals with unique properties for special applications. We have fi rst demonstrated extensive screening experiments allowing to determine the importance of individual parameters of a highly productive synthesis of title 1-propenyloxyalcohols via the solvent-less [Ru]-catalyzed isomerization of appropriate allyl substrates. It was confi rmed that previously postulated by us the hydride ruthenium complexes such as [RuClH(CO)(PPh 3 ) 3 ] 7-13 is an extremely versatile, high active and selective catalyst also for the this reaction. However, theoretical considerations and the experimental evidence suggest that the structure of allyl substrate i.e. number of PGE spacer is a key factor affecting the reaction rate. The most reactive substrate was 2-[2-(allyloxy)etoxy]ethanol in the contrary to 2-allyloxyethanol, for which the isomerization was the slowest process. Moreover, the loading of [Ru] pre-catalyst, the presence and the concentration of allyl hydroperoxides and also the temperature of reaction are signifi cant limiting factors in the selectivity, productivity and reproducibility of the reaction. Finally, it was also demonstrated that under optimized conditions, neat [RuClH(CO)(PPh 3 ) 3 ] selectively and effectively catalyzes the isomerization of low reactive 2-allyloxyethanol with relatively high TON of 10 000 and high TOF value of 20 000 h -1 , what is suffi cient for large-scale production of valuable fi ne chemicals. This results opens the door for further studies of its technological effi ciency using other much more reactive poly(ethylene glycol) monoallyl ether substrates dedicated to large-scale process development.