Design of Polyhydroxyalkanoate (PHA) Microbeads with Tunable Functional Properties and High Biodegradability in Seawater

Poly(3-hydroxybutyrate-co-3-hydroxyvalerate) and poly(3-hydroxybutyrate-co-3-hydroxyhexanoate) were used to prepare microbeads, with diameter ranging from 50 to 100 µm, by an emulsion-evaporation process. The emulsification-evaporation process enables the formation of spherical polyhydroxyalcanoate beads, with crystalline rates similar to the ones of the former polymer and with important surface roughness as compared to amorphous polylactic acid smooth beads. The mechanical properties of the different PHA beads are also found to be intimately linked with their crystalline content, with modulus varying between 1 and 7 GPa. The degradation behavior of these PHA microbeads was tested under marine environment and revealed a rapid degradation, similar to cellulose, and a degradation rate correlated with the crystalline content. These results emphasize the possibility and interest in developing PHA materials with tunable functions and degradation properties.


Introduction
Microplastics are defined as solid plastic objects smaller than 5 mm in size, insoluble in water and not biodegradable [1]. Personal care products (PCPs) have been identified as a potential source of environmental pollution due to their high content in primary microplastics, with typical number and mass content about 2162 particles/g or 0.04 g g −1 [2][3][4]. According to the PCPs' consumption, approximatively 1500 tons/year of micro plastics leach into the global aquatic environments, which account for 0.1-0.8% of the annual global release of primary microplastics in the world oceans [2]. Until recently, PCPs products are mainly formulated with conventional microplastics such as polyethylene, polypropylene, poly(ethylene terephthalate) or poly(methylmethacrylate) [5][6][7][8][9]. In order to limit their impact on the marine environment, many countries have attempted to ban the presence of these microplastics in rinseoff cosmetic products (United States in 2015, France and South Korea in 2016, Canada and New Zealand in 2017) [10][11][12]. These laws and decrees have led manufacturers to develop sustainable solutions with the use of particles or beads, for example in exfoliation products, such as natural organic ingredients including plant or fruit hulls, kernels seeds, microcrystalline cellulose or the use of mineral particles such as silica or pumice stone [13,14]. If these additives have the benefit to being natural and biodegradable materials compared to conventional polymers [15,16], they nonetheless also present drawbacks because of their irregular shapes and sizes. They are also generally colored, not or poorly stable in aqueous medium, with an inadequate hardness and their commercial quantity and availability are rather limited [17]. The cosmetics industry is therefore interested in developing biodegradable particles, in the micrometer size range, which are gentler in action than ground natural ingredients [18,19] for applications such as facial cleansing or scrubbing [5-8, 13, 18, 19].
Polyhydroxyalkanoates (PHA) which are bacterial polyesters especially biodegradable in marine environment [20] are natural candidates for applications in cosmetic meeting environmental constraints. Depending on the bacteria, stress conditions and substrates, a broad variety of PHA can be produced with different monomer units, leading to a variety of chemical and physical properties [21]. These properties can be modulated by the chemical composition or length of the lateral chain, the proportion of the monomer units and their distribution all along the polymer chain [22]. For example, Lemechko et al. [23] obtained poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBHV) macromolecules by optimizing and choosing the nature of carbon source (agroressources effluents) and the amount of valeric acid to produce a range of polymers with a controlled proportion of each monomer. Thermal analysis showed a decrease of the fusion enthalpy as the hydroxyvalerate unit (HV) content increased. By increasing the proportion of HV monomer (from 5 up to 20%), the PHBHV polymer becomes more ductile, its glass transition temperature (T g ) and crystalline rate [24] are also decreasing. Poly(3-hydroxybutyrate-co-3-hydroxyhexanoate) (PHBHHx) is another kind of commercially available PHA. The substitution of the hydroxyvalerate unit by an hexanoate unit (HHx) also modifies the polymer properties. This polymeric material becomes more ductile than PHB [25,26]. The T g temperature is also affected by a variation of the chemical unit of the copolymer [26] or the rate of HHx units within the copolymer. For example, increasing the proportion of HHx from 7 to 18% slightly decreases the glass transition temperature from − 5 to − 8 °C [27][28][29]. Consequently, it has been reported that the degree of crystallinity of PHBHHX is lower than that of PHBHV, explained by a greater steric hindrance generated by the HHx unit [29]. The amount of HHx unit within a PHA is also influencing the crystalline rate as highlighted by some authors who have measured a decrease of about 20% of the crystalline rate when the percentage of HHx unit increases from 4 to 20% [29]. Several methods have been described for the production of biopolymers microparticles [30,31] such as emulsification, gelation, drying, coacervation or precipitation. The choice of the method and then of the proceeding and formulation parameters influence the physico-chemical characteristics of the particles such as the porosity, sphericity, size, dispersity, surface appearance or shape [32][33][34][35][36][37]. The type of polymer, in particular its botanical origin, affects the size of the so-formed microparticles. For example this has been demonstrated for starch-based particles obtained by nanoprecipitation [32]. The process of elaboration of micro and nanoparticles is also a fundamental parameter that governs their final properties. For example, it has been shown that ultrasound process allows to obtain small PHB microbeads in comparison with a stirring process, with diameter of 0.14 µm versus 32 µm respectively [33]. The organic solvent used also strongly affects the final shapes and sizes of the particles [31]. It has been shown that, dichloromethane allows to obtain more spherical PHBHV microbeads than chloroform [34]. When dichloromethane is used as solvent, particles with a wide range of diameters, between 0.8 and 7 µm, can be obtained whereas a narrow distribution, between 0.1 and 0.4 µm, is observed when chloroform is used. Other authors have shown that the size of PHBHV microbeads can be tuned by adjusting the surfactant concentration, obtaining a range of microbeads from 389 to 39 µm for 0.5 to 4% of poly vinyl alcohol surfactant [36]. The porosity of PHB-based microbeads is also found to be dependent on the amount of aqueous phase, used in an emulsification process [33]. The final process of microbeads fabrication is the solvent evaporation process which is also an important step in determining the final properties of the particles [33]. The biomedical domain is certainly the one that has contributed to the most significant results in the elaboration of PHA microbeads [38]. In a majority of studies on PHA particles, the emulsification process was developed to elaborate micrometer size particles made from PHBHV or PHBHHx polymer. Table 1 sums up some of the main parameters used in these processes.
The objective of this work is to synthetize PHA microbeads for cosmetic applications with the following goals: size of approximately 100 µm, spherical shape particle with tunable surface aspects and mechanical properties and biodegradable in the marine environment. Surprisingly, to the best of our knowledge, the development of PHA microbeads for such application is poorly described in the literature. We have used three commercial PHA, of different chemical structures, to prepare such microbeads by an emulsificationevaporation process. The physico-chemical properties of the different beads were then characterized, in terms of shape, crystallinity, surface, mechanical properties and biodegradability and compared with PLA microbeads, used as control. This latter polymer was chosen due to the large amount of studies on the elaboration and the characterization of PLAbased microbeads in the literature.
Among the three selected PHA, one (PHBHV) is readily commercially available while the two others (both PHB-HHx) are only sparingly distributed over the world due to their low available quantities. In this context, this prospective study will make it possible to consider expanding the range of applications of these PHA, particularly for the PHBHHx polymers.

Materials
Poly(3-hydroxybutyrate-co-hydroxyvalerate) (PHBHV) with 3 mol% in HV was supplied by Tianan Biological Materials Co. Ltd. (China), under the trade name ENMAT Y1000P. Poly(3-hydroxybutyrate-co-hydroxyhexanoate) (PHBHHx) at 6 and 11 mol% in HHx were supplied by Kaneka Corporation (Japan), under the trade name Aonilex X131A and Aonilex X151A, respectively. Polylactic acid (PLA) with 4% D-lactide was purchased from NatureWorks under the trade name Ingeo 7001D. The physico-chemical properties of the investigated polymers are given in Table 2. For clarification in the following text, the notation PHBHV, PHBHHx (6%) and PHBHHx (11%) and PLA will be used. All chemicals products and reagents used in these experiments were analytical grades and were purchased from Sigma-Aldrich. In addition to polymer materials, extra-pur powder of microcrystalline cellulose (also named micronized cellulose), purchased from acros organics (chromatographic grade, with diameter around 50 µm), was used as a reference polymer in the degradation test.

Synthesis of Biopolymers Microbeads
Microbeads of PHBHV, PHBHHx (6 and 11%) and PLA were elaborated by emulsification process. PHBHV was dissolved at 50 g L −1 in chloroform under reflux conditions (50 °C), PHBHHx and PLA were dissolved at the same concentration but in dichloromethane under reflux conditions

Morphological Analysis
The morphology of the microbeads was analyzed by a scanning electron microscope (JSM-IT500HR from JEOL). SEM observations were carried out with secondary electron detector at an acceleration voltage of 3 kV. The particles were stuck on an adhesive carbon tape and gold-coated with a sputter coater (Scancoat6 from Edwards). The determination of the microbeads diameter and their shape was extracted from SEM images by image analysis. For each sample, a minimum of 500 microbeads were analyzed using ImageJ software (version 1.52, NIH). The circularity factor (CF) was determined using Eq. (1): The surface topographies of the microbeads were also measured by atomic force microscopy (AFM). AFM images were obtained with a multimode 8 atomic force microscope (Bruker, Santa Barbara, CA) operated on the scanasyst@ mode (Bruker) under ambient conditions (23 °C, RH = 50%). Standard scanasyst tips (Bruker), with a resonance frequency of 70 kHz and a spring constant of 0.4 N/m were used. Images were analyzed using the Nanoscope analysis software (V1.80). To ensure a good reproducibility in the measurements, for each sample, a minimum of three areas were investigated for each microbead, and for each sample a minimum of five different microbeads were observed.

Thermal Properties
Thermal properties were determined by Differential Scanning Calorimetry (DSC). About 5-8 mg were introduced in

Infra-Red Spectroscopy
Attenuated Total Reflection Fourier Transform Infrared (ATR-FTIR) spectroscopy (Vertex70v, Bruker) was operated in the range from 4000 to 600 cm −1 with 4 cm −1 resolution. Dried microbeads were deposited directly on to the diamond crystal. For each sample, a minimum of 16 scans were performed to ensure a good reproducibility in the signals.

Nanomechanical Properties
Nanoindentation experiments were performed with a Nanoindenter XP (MTS Nano Instruments) equipped with a three-side pyramid indenter (Berkovich) [57]. All Experiments were conducted under ambient conditions (23 °C, RH 50%), using the continuous stiffness measurement (CSM) method with the following parameters: 3 nm amplitude and 45 Hz oscillations using a 0.05 s −1 loading rate. Measurements were taken at depths to 1500 nm. A Poisson's ratio of 0.35 was used in all modulus calculations as suggested by Young et al. [58] for polymers where the Poisson's coefficient is unknown. For each sample, around 75 indents were performed with 5 × 5 matrix on different locations; average values of both elastic modulus and hardness were then calculated from curves according to the method of Oliver and Pharr [59]. Experiments were performed on polymer pellets embedded in an epoxy resin (Epolam 2020), dried at 50 °C overnight and then submitted to a careful polishing process (until paper grade 4000). Since the polymer pellets are agglomerated in dense particles, they appear in white in contrast with the transparent resin. Prior the indentation experiments, the sample is observed with an optical microscope (OlympusAX70), equipped with a 5 times BD objective (NA 0.15) to get a large image of the whole surface necessary to distinguish between the two materials and to choose an indentation area of interest.
The AFM technique was also used to probe the nanomechanical properties of the polymer microbeads using the peakforce quantitative nanomechanical measurements (PFQNM). In PFQNM, the piezo of the AFM is vertically oscillating at a frequency of 2 kHz, with an amplitude of 150 nm. While the piezo moves the sample beneath the tip in the X and Y direction, a force curve is recorded in every coordinate, allowing to extract mechanical properties from the materials. The spring constant of RTESP-525 commercial tips were evaluated using the Sader method (https:// sader method. org) and the tip radius was determined using the relative calibration method. A polystyrene film of 2.7 GPa was used as a standard of calibration (PFQNM SPM kit-12M, Bruker). The peak force setpoint was set at 200 nN.
The indentation modulus was calculated using the Dejarguin-Muller-Toporov (DMT) model using the nanoscope image analysis software (V1.80). For each sample, a minimum of three areas were recorded and for each polymer, a minimum of three different microbeads were analyzed. The microbeads were simply fixed using double-sided tape to prevent them from moving during the measurement.

Biodegradability
Stability tests were carried out in aqueous solution, as adapted from personal care tests, since aqueous solution is the least stable medium for exfoliating particles studied in cosmetic formulation. Test solutions were prepared by introducing 2% (w/w) of microbeads in purified water containing 1% of phenoxyethanol. The stability of the dispersion was followed for solutions stored under three different conditions, ageing at room temperature in the absence of light, at room temperature in the presence of light, or in a solution maintained at 45 °C to mimic an accelerated ageing. The variations in pH of the different solutions were followed over a period of three months.
The biodegradability of the different microbeads was determined with the NF EN ISO 19679 test that measures aerobic biodegradation of non-floating plastics at the seawater/sediment interface. The amount of total organic carbon contained in each sample, m CO 2,theorical , was determined by elemental analysis and is respectively of 44.44 ± 0.02% for micronized cellulose, 55.18 ± 0.02% for PHBHV, 56.52 ± 0.25% for PHBHHx (6%), 57.08 ± 0.18% for PHB-HHx (11%) and 49.59 ± 0.17% for the PLA. These theoretical masses of carbon were then used to calculate precisely the mass of product (approximatively 5 g) needed to get the same amount of carbon per polymer in the degradation test.
Sample preparation was previously described for solid/ liquid state, as an adaptation of the Sturm test [20]. Under continuous oxygenation in a sealed system, 5 g of microbeads were introduced in a mixture of seawater (100 mL) and sediment (30 g), placed in a first compartment, and the second compartment was containing an absorbing solution with a NaOH solution (20 mL, 0.2 mol L −1 ) and distilled water (20 mL). A blank flask with no sample was also included in each test set-up to determine the blank respiration of the seawater, as well as a standard composed of micronized cellulose to validate the results of the test. The test was carried out for 250 days in a water bath at 25 °C.
Since the produced carbon only comes from the biodegraded sample, the CO 2 trapped by the absorbing solution is proportional to the amount of carbon consumed. The product Na 2 CO 3 , by the reaction of CO 2 and NaOH is precipitated by a BaCl 2 solution. The remaining NaOH is titrated with hydrochloric acid (0.1 mol L −1 ) to determine the CO 2 trapped by the absorbing solution. Rate of biodegradation (%CO 2 ) was determined from the Eq. (2): where m CO 2,sample is the amount of CO 2 produced in the sample test, m CO 2,control is the amount of CO 2 produced in the blank test. (2)

Morphology of Microbeads
SEM images of the different microbeads are presented in Fig. 1. All the beads are spherical, as revealed by a circularity factor close to 1 ( Table 3). The diameter of the beads (Table 3) is found to be dependent on the polymer nature, thus indicating that, for a similar elaboration process, using  identical parameters, the chemical nature of the polymer controls the final size of the bead. PHBHV and PLA microbeads are the largest particles, with diameter around 100 µm, whereas PHBHHx particles have smaller diameters, that also decrease with a lower content in HHx. The large standard deviation measured on our samples may be associated with the elaboration process but also to the limited the number of particles measured by image analysis. These diameters are nonetheless consistent with other types of microparticles obtained using a similar process and determined by SEM or optical microscopy [16]. From the SEM images, the surface roughness can be qualitatively observed. The SEM images at different resolution ( Fig. 1) reveals that the PLA bead is rather smooth in comparison with the rough surfaces of PHBHHx or PHBHV beads. The roughness was quantitatively estimated by means of the AFM, on area of approximatively 25 µm 2 . It should be noted here that stable AFM images were difficult to obtain on a larger scale (> 25 µm 2 ) due large z deviation generated by the topography of the beads and by the radius of curvature of the beads. Figures 2 and 3 presents AFM images obtained of the top of the different beads and their corresponding surface analysis, confirming the smooth topography of the PLA beads and the important roughness of the different PHA beads. At such scan size, a roughness between ~ 130 and 210 nm is obtained for the three PHA particles whereas a roughness of only 25 nm is obtained for the PLA. This observation is well correlated with the maximum height (Rmax parameter) measured on these images, showing a large distribution in height for the PHA particles (up to 1000 nm) and a smaller height distribution for the PLA particle (Rmax up to 200 nm). The roughness of the PHA particles can be related with the semi-crystalline character of the polymer. Finally, it should also be mentioned, despite the difference in crystallinity rate between PHBHV and PHB-HHx samples, no difference in RMS (considering scan area of 25 µm 2 ) were measured between the different PHA beads.

Thermal Properties
The thermal properties of four types of microbeads are reported in Table 4 and the thermograms presented in Fig. 4. Characteristic values are determined from the first heating to observe the effect of the elaboration process on the final properties of the beads. PHA microbeads have a glass transition temperature around 0 °C, with the lowest Tg observed for the 11% HHx content PHA beads. PHBHV exhibits a melting temperature at about 173 °C. In the case of the PHB-HHX beads, two melting peaks were detected, at 125 and 144 °C for PHBHHx (6%) and 109 and 137 °C for the PHB-HHx (11%) respectively. These two melting peaks could be attributed to the melting temperature of the PHB segment and the PHHx segment, as reported in other works [26,27,40,49]. Nevertheless, melting phenomenon could be very complex; an apparent double peak could also be the result of an exotherm peak superimposed on an endotherm peak (crystalline perfection through melting/recrystallization) works. The proposed interpretation of the double melting peaks is not straightforward and can also be explained by other complex phenomenon occurring in semi-crystalline polymer such as crystal thickening, crystal perfecting or recrystallisation that occur simultaneously with fusion and influenced by the heating rate [60,61]. Crystalline reorganization upon heating can only be excluded by comparing DSC curves recorded on the same sample (same polymer with the same microstructure) with sufficiently different heating rate. This crystallization study would require much more work to be more assertive about the mechanics involved.
PLA microbeads exhibits higher glass transition temperature, around 60-63 °C, and a melting temperature of 147 °C. These results are also in agreement with those described for amorphous PLA microbeads obtained by emulsion-evaporation in the presence of dichloromethane [34].
From these data, it appears that PHBHV sample gives the most crystalline beads. A difference in crystalline content can also be observed between the two PHBHHx samples, with the most crystalline bead being the 6% in HHx content. These data emphasize the important role of the chemical nature of lateral chain (HV vs. HHx) in the thermal properties of the polymer material. The increase of the lateral chain size also contributes to slightly decrease the glass transition temperature from 2 °C for PHBHV to − 1 °C for PHBHHx (11%). This is in agreement with the literature data since it has been demonstrated that increasing the fraction of HV monomers in PHBHV leads to a decrease in the glass transition temperature proportionally [62]. Similarly, it has been shown that, increasing the percentage of HHx from 7 to 18% results in a decrease in the crystallization rate and suppression of the spherulitic growth rate [27].
FTIR spectroscopy analysis was performed in order to get access to the surface crystallinity of the different beads. This technique is highly complementary with the DSC technique that gives an overall estimation of the crystallinity, independently of the structural organization of the polymer within the bead and from the center of mass. FTIR is sensitive to the polymer organization on its periphery and within a thickness of a one micron approximatively. FTIR spectra of PHA microbeads are presented in Fig. 5. The bands at 1230, 1380 and 1724 cm −1 are assigned to the crystalline part of the PHA sample whereas those at 1186 and 1741 cm −1 are representative of the amorphous part (see Table 5). By comparing the intensity ratios of some of the characteristic bands of PHA, some authors were able to evaluate a crystallinity index and established a correlation between FTIR and DSC measurements [63,64]. For example, in the work of Xu et al. [63], the crystallinity index was calculated as the ratio , with a section profile for each bead. The mean roughness (Rq value) and the maximum height (Rmax) are also given to provide a comparison between the surface topographies of each bead, for scan areas of 25 µm 2 of the crystalline band at 1380 cm −1 , assigned as the conformational band of helical chains in the crystalline phase, over the amorphous reference peak at 1453 cm −1 , assigned as methyl (CH 3 ) or methylene (CH 2 ) deformations [64]. In other studies, the crystallinity index of PHA samples was also evaluated by calculating the ratio of the absorbance peak at 1227 cm −1 over the absorbance peak at 1184 cm −1 [65], between the absorbance peak at 1724 cm −1 over the absorbance peak 1453 cm −1 [64], or between the absorbance peaks at 1380 cm −1 over the amorphous band at 1186 cm −1 [66].
Intensities of these ratio were calculated and reported in Table 6. It can be observed that the highest intensity ratios are obtained for the PHBHV beads, followed by the PHB-HHx at 6 and 11% respectively. These results are in agreement with the previous DSC results and confirms that the crystallinity on the surface of the particles is well correlated with their crystallinity content. The roughness of the different PHA beads is also well correlated with the crystalline aspect of the periphery of the beads, as probed by FTIR.

Mechanical Properties
Nanoindentation measurements directly on the microbeads, with the MTS equipment were not successful since the beads were not clearly identified within the epoxy resin. Thus, nanoindentation test have been performed here directly on aggregates of pellets, more distinguishable within the  Table 4 Thermal characteristics of the different microbeads determined by DSC (scanning rate of 10 °C min −1 under nitrogen flow) a Values of melting enthalpy, ∆Hm, are given as raw data and not converted into crystalline content because of the absence of reference for a PHA with 100% crystallinity

Microbeads
T g (°C) T m (°C) ∆H m (J/g) a PHBHV 3 ± 2 173 ± 1 97 ± 2 PHBHHx (6%) 2 ± 1 125 and 144 ± 1 58 ± 2 PHBHHx (11%) − 1 ± 1 109 and 137 ± 1 40 ± 6 PLA 61 ± 2 147 ± 1 22 ± 1 epoxy resin due to their large size (few millimeter) and white color. In parallel, PFQNM experiments allow a more direct analysis of the mechanical properties of the beads since indentation experiments are conducted directly on the top of the different microbeads. Nanoindentation measurements on pellets were performed in order to have a reference  1 3 modulus given the absence of PFQNM measurements on PHA polymers in the literature, to the best of our knowledge. Table 7 reports the values of the modulus for the different polymers pellets, as measured by nanoindentation. At first, the Young's modulus for PLA is found to be around 5.5 GPa, in agreement with the literature value [70]. If we compare the three PHA samples, the two PHBHHx samples give a lower Young's modulus as compared to the PHBHV. For the PHBHV polymer, a Young's modulus of 7.5 GPa has been found, similar to the results of Chick et al. on injected PHBHV polymer [48]. The highest modulus found for the PHBHV sample can be related with its high crystalline content. In contrast, for the PHBHHX samples, a lower Young's modulus was found for the 11% content in HHXmonomer, in comparison with the 6% content. If this observation is opposite with the crystalline content of both samples, it gets along with the work of Voyiadjis et al. [71], who studied the mechanical properties of PEEK semi-crystalline polymers. They observed a softer response of the amorphous polymer part as compared to the crystalline regions. Figure 6 represents the distribution in the indentation modulus for the different beads as measured by PFQNM. Results obtained are consistent and following the same trend as the moduli obtained by nanoindentation, but with a slight increase in the PFQNM modulus values. However, a precise comparison between both techniques should be precautious. At first, the calibration of the tip radius was performed on a smooth Polystyrene film in comparison with the rough surface topographies of the different beads. As a consequence, an increase of the tip contact area is likely to occur, especially on rough surfaces as it is the case here, which affects the indentation modulus calculated from the force curve analysis. It has also been recently postulated that PFQNM experiments may provide slightly larger modulus as compared to other mechanical techniques, since the force curve may be partially impacted by the presence of adsorbed water that may increase the adhesion force between the tip and sample [72]. When the microbeads are not crystallized, thus with a low roughness like PLA microbeads, the Young's modulus is close to that found in the literature [73]. The variation in the mechanical properties, as observed by the two techniques confirms that the bead present different mechanical properties, in correlation with their crystalline content and their chemical structure.
Finally, the hardness of the different samples, extracted from the nanoindentation measurements using Oliver and Pharr [59] theory, are shown in Table 7. The results reveal that both PHBHHx samples have the lowest hardness in contrast with the PHBHV or the PLA samples. The observed trend for PHA samples is also well correlated with the crystalline state of the different samples, the more crystalline being the tougher. For the PLA sample, which is less crystalline than the PHA sample, the mechanical properties can be explained by its high glass transition in comparison with the PHA samples. It has been reported that the intrinsic stiffness of glassy polymers below the T g may lead to microhardness values larger than those obtained for semi-crystalline polymers [74].

Stability in Aqueous Medium and Biodegradability in Marine Environment
The stability of these particles in aqueous medium, with various conditions to mimic ageing conditions for cosmetic products (room temperature, room temperature in the presence of light and accelerated ageing at 45 °C) was observed by measuring the pH of the solution during a period of 3-months storage. Measurements of pH versus times may provide information on the stability of biodegradable polymers in aqueous media. The hydrolysis of polyesters (PLA, PHA for example) results in the formation of carboxylic acid molecules and alcohol. A decrease of pH (acidification), due to the formation of carboxylic acid functions, can thus be a direct measurement to follow the hydrolytic degradation of these polyester chains. For a stable polyester in an aqueous medium, the pH should not change over time. A blank sample, without beads, is used as a control to ensure a stable pH over the period and conditions of investigation.
The comparison between the PHA and PLA microbeads allows to observe different trends. On the one hand, the pH variation for the solution containing the PHA microbeads was only slightly decreasing, with a pH variation inferior to 1 pH unit. For example, after 3 months at room temperature, a pH of 3.9, 3.7 and 3.8 for PHBHV, for PHBHHx (6%), for PHBHHx (11%) were measured respectively while the starting pH of all solutions was measured at 4.2. Similar results were obtained when the PHA microbeads suspensions were  Figure 7 shows the biodegradability of the different microbeads in seawater. The biodegradation has been performed on marine environment (seawater + sediments) at 25 °C, according to the NF EN ISO 19679 standard. As expected, PLA microbeads are poorly biodegradable in these conditions since their biodegradation degree is only about 20% after 250 incubation days. It has been previously demonstrated that PLA is relatively stable, due to its glassy state at 25 °C, as long as the medium temperature does not exceed the PLA glass transition temperature, i.e. about 55 °C [75].
Concerning the PHA series, significant differences on the biodegradability can be noted depending on the chemical PHA structure. The PHBHV turns out to be the most biodegradable as its biodegradability percentage reaches 90% after 250 days of immersion. For the PHBHHx, the composition of monomer units also influences the biodegradability since the PHBHHx with 11% in HHx biodegrades faster than the PHBHHx with 6% in HHx. The biodegradation rate is 80% for the first and 62% for the second. It is very interesting to note in Fig. 7 that all these PHA present a biodegradability greater than, or close to cellulose, the reference sample. The beginning of the biodegradation is even faster for the 3 PHA, as compared to cellulose.
This spectacular PHA biodegradability is explained by the action of some marine microorganisms such as bacteria which excrete extracellular PHA degrading enzymes, i.e. PHA depolymerases, that hydrolyze water-insoluble PHA chains into water-soluble forms [76]. The resulting products are finally metabolized into the cells and utilized as nutriments [20]. At this temperature, the microbead degradation is therefore managed by an enzymatic degradation which is a heterogeneous surface reaction. Previous studies have revealed the presence of two PHA degradation mechanisms occurring in parallel (enzymatic degradation and chain scission by hydrolysis) but the enzymatic degradation is largely predominant at 25 °C in marine ageing conditions [75]. This process, whatever the sample shape, takes place in the presence of PHA depolymerase involving two steps: the first step involves the adsorption of the enzymes on the surface by the binding domain of the enzymes and the second step involves the enzymatic cleavage of polymer chains by the active sites of the enzymes [77].
The slight differences in terms of biodegradation between the three PHA studied are more complicated to explain since the biodegradation is a combination of physical, chemical and biological phenomena leading to the material dissolution by enzymatic action of microorganisms. More experiments are needed in order to understand the role of the extrinsic (correlated to the medium) and intrinsic (relative to the polymer) parameters influencing the biodegradation process. It might be relevant to identity and quantify the microorganism population which specifically colonizes the microbeads surface during the test [78]. The diversity of microorganisms associated with these different stages of biodegradation is not yet characterized but this is being further explored in order to better understand the mechanisms. Likewise, the surface morphology of the different PHA has also to be studied in order to correlate the surface properties with the colonization and then the biodegradation. In this study, the PHBHV with 3% in HV is the most biodegradable while being the most crystalline compared to the two other PHB-HHx. This could a priori constitute a surprising result even if some bibliographic data show that the substrate binding domain of the enzyme is capable to bind to the crystalline PHA material. PHA depolymerase being an enzyme made up of a catalytic domain and a substrate-binding domain, both these domains are connected by a linker domain. Subsequently, the catalytic domain starts to cleave the single crystals which can be enzymatically hydrolyzed [21]. Other parameters relative to PHA will have to be explored as the hydrophilic/hydrophobic balance of the surface as well as the surface porosity. Understanding the mechanism of PHA degradation and the factors that affect its degradation will help the researcher in designing suitable material for the specific needs.

Conclusion
The emulsification-evaporation method allowed the preparation of spherical micrometer PHA beads with tunable materials properties, surface morphologies and related degradation behavior. Using different chemical PHA structures, we showed that the different beads have properties and surface morphologies that are governed by the crystalline organization of the polymer chains within the beads, thus able to provide suitable abrasive and mechanical properties for cosmetics applications. Since pollution of aquatics systems by microplastics should be stopped, the degradation behavior of these PHA microbeads were further tested in marine environment. Biodegradation experiments reveal that the degradation rate and kinetic were faster than those of cellulose polymer, considered as the most biodegradable polymer materials. They also suggest the crucial role of the crystalline content in the degradation process of PHA beads. Very interestingly, these PHA particles are stable in aqueous media commonly used in cosmetics while being rapidly biodegradable in the marine environment. By combining these two behaviors, they thus offer ideal characteristics for the development of microbeads in cosmetics. In addition, the use of different PHA structures allows to tune the surface morphologies, the mechanical and biodegradable properties of PHA beads. This possibility provides an effective and promising approach to replace conventional plastic beads from formulation in cosmetic products.