Green Enzymatic One-Pot Synthesis of Renewable and Biodegradable Surfactants in Supercritical Carbon Dioxide (scCO 2 ).

We seek to expand the opportunities to exploit glycerol, a largely untapped renewable feedstock, by exploiting enzymatic catalysis in supercritical carbon dioxide (scCO 2 ). This work highlights a promising and clean approach to bio-renewable amphiphilic polyester-based biodegradable surfactants. We have developed a low temperature (40, 50 and 60 °C), low energy melt processing route to biodegradable, renewable poly(glycerol succinate) (PGLSA) polymers that importantly have a low degree of branching (3% <DB< 11%). Our approach shows significant advantages over traditional melt polycondensation at 110-120  C, where the standard catalyst-free approach led only to highly branched (DB >85%) or insoluble crosslinked materials. We have exploited these linear PGLSA materials to create a library of ‘ green ’ surfactants by end-capping with lauric acid or poly(ethylene glycol). Our approach avoids pre-modification of the monomers and fewer synthetic steps are required. Finally, we evaluate the performance of these new surfactants, focussing upon surface tension, critical aggregation concentration (CAC) and water contact observed for the T the


Introduction
Our growing global population will lead to increased demand for energy and resources. 1, 2 In recent decades, the study of alternative and renewable feedstocks for the synthesis of polymeric materials has received significant attention [3][4][5][6][7] with applications ranging from medical and tissue engineering, food packaging, coatings, cosmetics, surfactants and more. 8 Surfactants are widely used as emulsifiers, detergents and foaming agents across our society, with applications across home-, personal-, health-and crop-care. Since their introduction in the early 20 th century, the production of surfactants from petrochemical sources has continuously increased 8 reaching 18.5 million tons per year, and is forecast to grow at a compound growth rate (CAGR) of 5% from 2018-2023. 9 There is now an awareness that we need more environmentally-friendly and economically viable surfactants 10 preferably derived from renewable resources 11,[12][13][14][15] Glycerol is an underexploited by-product of biodiesel production, in particular from the hydrolysis of biomass derived triglycerides (such as palm oil, sunflower oil or rapeseed oil) which results in valuable methyl esters. 16 Between 2007 and 2016, biodiesel production increased by 83% in the European Union. 17 Thus, glycerol is widely available and its price is inversely proportional to the increase in biodiesel production. 18 As a comonomer, we have focussed upon succinic acid (SA) which has been widely employed as a starting material for different applications in the surfactant, food and pharmaceutical industries. 19 Presently, SA is produced from petrochemical feedstocks, but there has been a trend towards production of bio-based SA from biomass (e.g. sucrose and glycerol). [20][21][22] The polycondensation of glycerol and succinic acid to form novel biodegradable materials has been studied previously, but has typically involved energy intensive processes (heat /vacuum), toxic solvents and/or catalysts. 21,[23][24][25][26] In addition, if not well-controlled, the polycondensation of glycerol and diacids (succinic acid, azelaic acid and glutaric acid) gives only low conversions and a plethora poorly controlled cross-linked or branched materials. [27][28][29] Biodegradable and bio-renewable polyesters such as poly(lactic acid) (PLA), poly(glycerol-succinate) (PGLSA) and poly(lactic-co-glycolic acid) (PLGA) have been combined with PEG, lauric acid (LA) and palmitic methyl ester to prepare surfactants.
However, high temperatures, toxic catalysts and solvents are always required. 25,27,[30][31][32][33][34] In this paper we show for the first time the production of a library of biodegradable surfactants derived from glycerol. To do this, we exploited scCO2 to facilitate mild reaction conditions and the allow the use of a lipase CaLB (Novozyme-435) as a chemo-and regio-selective catalyst to yield the linear and low molecular weight polymers desired to construct a range of surfactants. In order to tune the amphiphilic balance of the PLGSA backbone, PEG and lauric acid were then employed as hydrophile and hydrophobe respectively. PEG can be produced from biobased feedstock such as bagasse 35 and lauric acid is a naturally occurring fatty acid.
Thus, the entire surfactant molecule can be considered biorenewable and fully biodegradable.
Experimental Section and Materials are reported in the Supporting Information document.

Results and Discussion
The key focus of this work is the design and optimisation of an enzymatic synthesis of PGLSA exploiting scCO2 (Scheme 1) to develop a facile route to linear and low molecular weight polyesters. Avoidance of branching is important because as branching increases, the number of pendant hydroxyl groups on the polymer chain is decreased, and this compromises water solubility. 36,37 Control experiments were also performed in scCO2 without CaLB and under the more traditional melt polymerisation conditions at 120°C with and without CaLB. Scheme 1. Schematic representation of the synthesis of PGLSA from glycerol and succinic acid.

Enzymatic polycondensation of poly(glycerol succinate) under supercritical conditions
Previous studies have determined a very low solubility of succinic acid in scCO2 38 and others have investigated the phase equilibrium of the CO2/glycerol system. 39 In our experiments, the strong interaction and solubility of scCO2 in glycerol is clearly shown by the appearance of bubbles in the liquid glycerol upon depressurisation ( Figure S1, E). This interaction offers the opportunity that scCO2 could act as a processing aid in polymerisation, lowering viscosity and improving mass transfer of monomers to the catalyst or enzyme. 36,37 A series of polymerisations of succinic acid and glycerol were trialled using CaLB (25 wt.% wrt monomer including polymer support) to gauge the effect of temperature (40, 50 and 60 ºC) and molar ratio of the monomers (Table 1). Hb protons from the B0 and B1 structures, using 1 H-NMR and focussing on proton Hb (see Figure 1). b Measured by DSC.
* yield of recovery: actual amount of material physically recovered from the reactor after reaction.
The molar ratio of the monomers was found to only slightly affect the size and topology of the   Figure S2A)  was obtained. Increasing the diacid content or glycerol content (Table 1,    Melt polycondensation of poly(glycerol succinate) PGLSA was also synthesised without catalyst via the more traditional melt polycondensation at 120C (Table 2, entries 7, 8 and 9). The yield of polymer obtained was generally much lower than that obtained in scCO2. When performed with a 1:1 molar ratio in the absence of catalyst at 120C, a polymer with a molecular weight up to 3,900 Da (Đ > 2) was obtained with DB 18% and this was found to be only partially soluble in water. The increase in branching can clearly be seen ( Figure 3). Increasing the SA content to a G:SA molar ratio of 1:1.5 and 1:2, increases DB dramatically. The molecular weight of the polymer also increased, and these materials were found to be insoluble in water, in accordance with previous studies. 47 and B1 structures, using 1 H-NMR. The chemical shifts used related to Hb the proton are reported in Figure 1. b Measured by DSC.
* yield of recovery: actual amount of material physically recovered from the reactor after reaction. We also probed use of the enzyme under conventional melt conditions, reasoning that the inherent chemo-and regio-selectivity might yield the desired linear polymers (entries 10 and 11 in Table 2). Use of CaLB does indeed significantly lower branching, dropping the DB to 13 and 11%. The molar ratio of the monomers was found to only slightly affect the size and topology of the polymers in the presence of CaLB in scCO2 as well as in melt conditions. On the other hand, when no catalyst was present gelation was seen to occur earlier and at a lower ratio of Gly:SA ( it is clear that the increased molecular weight and higher DB that are obtained lead to higher Tg. Indeed, it is well known that Tg increases with molecular weight 48 and that branching also influences chain interactions 49,50,38 . The Tg's of the PGLSA chains synthesised via enzymatic melt polycondensation did not show a dependence upon molecular weight ( Table 2), but such Tg values in the range of -50 ºC are strongly indicative of other linear polyesters such as poly(butylene itaconate) and poly(1,5-pentylene adipate)). 38,51 The use of CaLB in the melt certainly leads to more linear polyesters containing the desired pendant hydroxy groups. However, the melt condensation process at 120C is not ideal since the high temperatures lead to degradation of the enzyme which cannot be recycled and tend towards high molecular weight and low yield. Higher molecular weights in general are problematic for developing surfactants because such materials show only low water solubility and cannot be utilised effectively. The use of scCO2 clearly facilitates the synthesis of linear low molecular weight poly(glycerol succinate) at lower temperatures and with reasonable yield. Moreover, it has been shown previously that such supported enzymes can be recycled and reused several times in scCO2. 36,52 MALDI_TOF analysis for PGLSA synthesised in melt conditions without any catalyst ( Figure   S3) showed different repeat unit patterns and much more branching compared to the PLGSA polymers obtained in scCO2 with CaLB ( Figure 2). This initial screening highlights the crucial combination of enzyme and low temperature in our use of scCO2 leading to linear, low molecular weight (water soluble) poly(glycerol succinate) under mild reaction conditions and with high yield. In the next section we go on to exploit these syntheses to create renewable and biodegradable surfactant from PGLSAs with the addition of lauryl or PEG moieties (Scheme 3).
We also looked carefully at performing the polymerisations in conventional solvent. Clearly the solvent must not be miscible with water to allow ofr removal of water during the polycondensation process. Reactions in toluene were optimised (see SI) and it was found  (Table Table 3 (Table 3 and Figure 4) were in good agreement with the values obtained through GPC (Table 3). Very positively, all the polymerisations of LA-PGLSA polymers show only low degrees of branching (DB) ( Table 3) which is ideal for the potential use as surfactant molecules  The calculated Mn NMR (  Table 3) gave PEG-PGLSA with a molecular weight of 1,500 Da (Mn GPC ) (Đ > 2, DB = 0%) and 94% yield was obtained.
It is well known that thermal properties of polymeric structures can be affected not only by the degree of polymerisation, but also by end-groups. 41 The Tg values for the samples synthesised at 40 ºC (entry PEG-PGLSA 4, Table 3) were higher than those produced at 50 and 60 ºC (-45 ºC vs ca. -75 C ) (entries 5 and 6, Table 3). This appears to show the important influence on the thermal properties of the degree of branching since at 40 ºC, PEG-PGLSA showed a DB of 13%, while at 50 and 60 ºC, no branching was detected. This increase of the degree of branching in the presence of PEG moieties might well increase the entanglements between the polymer chains and lead to higher Tg. No such correlation was observed for the Tg trends for the uncapped PGLSA.
We have demonstrated that the enzymatic synthesis of biorenewable and biodegradable surfactants can be controllable, with low branching values giving linear and water-soluble chains with a range of potential end group functionalities. Moreover, the syntheses with scCO2 are at lower temperatures, and provide a clean and efficient route to new surfactants. These data complement earlier studies that showed the synthesis of PEG-based surfactants under enzymatic scCO2 for azelaic acid and 1,6-hexanediol end-capped with hydrophilic methoxy poly(ethylene glycol) moieties 37 and those based upon sorbitol and lactide. 57 The replacement of conventional solvents by scCO2 is certainly viable since this introduces opportunities ofr lower temperature processing and the utilisation of enzymatic routes to polymer based surfactants. 58

Surfactant properties
Amphiphilic, biorenewable and biodegradable polymers can find application in formulations for wetting agents, emulsifiers and detergents, but only if they are able to sufficiently reduce the surface tension of water. The PGLSA polymers (with G:SA ratios of 1:2 and 2:1) that were not end-capped showed only a minimal reduction in the static and dynamic surface tensions and do not show any significant surface-active properties (Figure 6 and S5). By contrast, the end group modified LA and PEG polyesters show great promise as green biodegradable surfactants demonstrating a significant reduction of the dynamic and static surface tension values ( Figure 6 and S6) and these compare favourably with benchmark surfactants (Tween TM 20 and NatraGem TM E145) derived from petrochemical feedstocks.    Table 4) and, since the degree of branching at 5% is the lowest, this probably reflects the presence of more hydroxy pendant groups in a larger polar head.
The PEG-PGLSA surfactants gave higher CAC values, likely due to the high hydrophilicity of PEG as an end-group. The PEG-PGLSA at 60ºC showed a surprisingly high CAC value, >1000 mg/L (> 1 wt.%). This might be a consequence the presence of free M-PEG (unattached to the PGLSA) (entry 8, Table 4).
The sizes of the self-assembled structures are obtained from DLS and are in the range of 170-600 nm (entries 3-8 from Table 5). Personal care and cosmetics applications require selfassembled aggregates in a size range between 200-500 nm. 59 Whilst for drug delivery, selfassembled aggregates must typically be smaller than 200 nm; 60 to deliver efficient penetration through blood vessel walls. 61, 62 The sizes of the self-assembled structures will of course be influenced by the molecular weight of the building blocks, the length of the end-cappers, and the hydrophobicity of the non-polar block. 37 These data did not show any significant size differences between the two types of endcapped PGLSA-based polymers, but promisingly the size distributions were similar to the commercial surfactants NatraGem TM E145 and Tween TM 20 in the range 100-300 nm.
Water contact angle (Θw) is of pivotal importance for different applications, including cleaning, lubrication, coating and printing. 63 Θw values observed for PGLSA polymers not end-capped with LA or PEG were near 100º showing that the bare PGLSA backbones had minimal ability to reduce the interfacial tension between the water solution and the solid surface. On the other hand, the end-capped PGLSA surfactants show promise with contact angles in the same range as Tween TM 20 and NatraGem TM E145 (Table 6).  These data collectively demonstrate that our PGLSA-based surfactants are effective in reducing the surface tension of water compared with the commercial surfactants and deliver CAC values, aggregate sizes and water contact angles that show promise as biorenewable and biodegradable surfactants.

Conclusions
We have exploited the unique properties of scCO2 to allow melt synthesis of poly(glycerol succinate) PGLSA at very mild temperatures (<60 °C). This allows effective use of an enzyme providing the opportunity for application as wetting agents. These preliminary results clearly show that all the synthesised PGLSA-based surfactants that we have developed can form selfassembled aggregates with suitable size for personal care and cosmetic applications. 52