Introduction

Plastic waste is a significant environmental challenge and its incineration generates carbon dioxide and other toxic gases that pollute the air1. A large amount of non-biodegradable plastics also ends up in the ocean, creating microplastics that are dangerous for the marine ecosystem and can enter the foodchain2. There is a large range of possible actions addressing this problem (which should be pursued simultaneously), ranging from reduced consumption to reuse, enhanced collection and processing infrastructure, as well as chemical recycling. Another way is the development of alternative materials including biodegradable bioplastics generated from biomass. Ideally, they would be compostable or even break down and degenerate in the general natural environment.

Biodegradation tests are essential to evaluate the impact of bioplastics on the environment, as bioplastics are developed with the intention of reducing the negative effect of nondegradable petroleum-based plastics2,3,4. However, most of the biodegradation studies only focus on the time required for the material to degrade and the level of degradation. There is still limited research to study physical changes associated with the biodegradation process to better understand the degradation mechanism. A key to further optimise the biodegradation process is an understanding of the morphological development of biodegradable bioplastics over time. In general it is expected that bioplastics in the natural environment absorb moisture, which might lead to swelling, plasticization, hydrolysis and weakening of intermolecular forces5. This may generate or increase internal porosity due to locally different swelling rates with associated internal stresses, leaching of water-soluble components, and the formation of microcracks within the material6. It is logically to expect that the rate of degradation increases with increasing external access to internal porosity due to initial size exclusion of microorganisms like bacteria or fungi. This also plays into the pathways for uptake of disintegrating plastics due to their coverage by biofilms7. Quantifying this access for microorganisms with size of the order of microns8 in 3D is a challenging task.

X-ray computed tomography (XCT) is a nondestructive characterisation technique that records the internal structure of a material with high resolution. The technique has found use in many applications ranging from the characterisation of delicate materials such as artefacts and meteorites to quantifying the composition of rocks and minerals, analysing the distribution of ingredients in food products9to elucidating the in-growth of bone into bone scaffolds10,11. The latter studies considered the percolation of new bone into the scaffold, which was related to a critical radius of percolation on the order of 100 \(\mu\)m. Here, we are interested in a similar percolation argument in the context of plasticiser movement and the evolution of physical access of microorganisms to the internal porosity of bioplastics.

Banana pseudostem, as the biggest part of a banana plant, is an underutilised agricultural waste12. The quantity of the pseudostem available globally is huge as every time the banana fruit is harvested, banana pseudostem is chopped down13,14. This waste is also available throughout the year as the banana plant is not a seasonal plant15. As the dry part of banana pseudostem mostly consist of cellulose16,17,18, high production and continuous availability of banana pseudostem throughout the year makes this waste a potential source of nanocellulose for bioplastic. We note that the pseudostem of bananas is an excellent raw material for bioplastics due to its high content of cellulose. In previous work19,20, we tested PEG 400, 1000, 2000, 4000, and 8000; a combination of banana pseudostem nanocellulose and polyethylene glycol (PEG) with a molecular weight of 1000 g/mol (PEG 1000) resulted in a bioplastic with the most desirable properties. It had approximately 4% percent elongation and 30 MPa tensile strength.

Given the scale of the environmental issues caused by non-degradable bioplastics, it is important to develop methods to understand their formation including disintegration and biodegradation. Our hypothesis is that X-ray computed tomography can explain the structural changes in bioplastics due to the biodegradation process. Furthermore, we expect that it will find applications in identifying the cause of changes in degradation rates due to morphological arguments of access to internal porosity. Here we introduce the method and basic approach for a biodegradable bioplastic film, which challenges the limits of the micro-CT approach in terms of resolution and field of view. Nanocellulose bioplastic plasticised with PEG 1000, namely PEG film, is used as the specimen. Since the disintegration and/or biodegradation of bioplastics is a multi-scale process, characteristics of PEG film before and after degradation were evaluated using both 2D scanning electron microscopy (SEM) and 3D X-ray computed tomography (XCT) to observe morphological changes. SEM, offering a resolution of a few nano metres, is an excellent tool for observing the surface change of bioplastics, while XCT provides insights into the internal 3D mesoscale structural change of the film, identifying the degradation process. The remainder of this paper is organized as follows: First, we describe the films utilized in this study and recall shortly the relevant film preparation methods. This is followed by a description of standard biodegradability tests, and SEM and XCT techniques relevant to this study. Finally, we present the results followed by a discussion and conclusions.

Materials and method

Materials

Samples for plastic disintegration tests were (1) certified commercial compostable plastic (BioPak that was made from Ecopond®), (2) filter paper (Advantec Toyo 131), (3) unplasticized banana pseudostem film (banana film), and (4) PEG plasticised banana pseudostem film (PEG film). Commercial compostable certified plastic and filter paper were used as positive controls. The filter paper is made of cotton cellulose with thickness 0.25 mm. Banana film and PEG film were prepared from banana pseudostem nanocellulose with diameter of 10-25 nm, crystallinity index of 66%, and carboxylate content of 0.57 mmol/g21.

Preparation of banana film and PEG film

The preparation method for the banana film and PEG film has been published in21and22 and is shortly summarized here. Firstly, cellulose from banana pseudostem was isolated by treating the biomass with potassium hydroxide solution and sodium chlorite. The cellulose (9 g) was then oxidised in a mixture of 0.2 g NaBr, 0.04 g 2,2,6,6-Tetramethyl-1-piperidinyloxy (TEMPO), 22.5 mL of 12% NaClO, and 100 mL Milli-Q water. The oxidation was carried out for 3 h at room temperature under constant stirring. The reaction was quenched by adding 5 mL ethanol. The mixture was then centrifuged, and the residue collected as nanocellulose. The nanocellulose was then mixed with Milli-Q water to form a suspension. The suspension was poured into Petri dishes and dried in an oven at \(\hbox {40}\,^{\circ }\) C to form banana films. PEG films were made in the same manner as banana films, except that PEG with molecular weight of 1000 g/mol was added to the nanocellulose suspension. The amount of added PEG was 60% DM of nanocellulose.

Plastic disintegration test

The plastic disintegration test follows the standard method of ISO 20200:2004(E) (ISO, 2015). We note that this standard was developed for the testing of the disintegration of plastic under composting conditions. It assesses the physical fragmentation of the material into smaller pieces. Biodegradation, the complete breakdown of the organic matter by microorganisms into constituent components like CO2, water and biomass, is not a necessary part of the plastic disintegration process. Auxiliary additional measurements are required to confirm the complete breakdown of organic matter by microorganisms, which are not carried out in this work. Test media were prepared by mixing compost (10%), sawdust (40%), rabbit feed (30%), corn starch (10%), saccharose (5%), corn seed oil (4%), and urea (1%) (ISO, 2015). Commercial bioplastic, filter paper, banana film, and PEG film were cut into small film pieces with a size of \(25\,\textrm{mm}\times 25\,\textrm{mm}\). The film pieces were dried in a vacuum oven (\(\hbox {40}\,^{\circ }\) C, overnight) and weighed after drying. The films were inserted into a reactor after they were immersed in Milli-Q water for 30 s. The reactor was a plastic container equipped with a screw cap and had two holes at 6.5 cm from the bottom of the container with a diameter of 5 mm for air circulation. The prepared media were added to the reactor with the ratio of dried film samples to media being 1:100. The reactors were then incubated in the oven with temperatures \(58\pm 2\) \(^{\circ }\) C for 90 days and \(25\pm 2\) \(^{\circ }\) C for another 90 days. Details of operations during the six months incubation are listed in the Supplementary material.

Sample analysis with scanning electron microscopy (SEM)

Samples of the films were prepared for SEM analysis by cutting the film into small pieces. The pieces were mounted on SEM stubs using carbon tape to ensure proper adhesion. After mounting, the samples were dried in an oven at \(\hbox {40}^{\circ }\)C for 24 hours to remove any residual moisture. Following drying, the samples were coated with a thin layer of chromium using a sputter coater to enhance conductivity and prevent charging during SEM imaging. The SEM analysis was conducted using a FEI NOVA NanoSEM 230 instrument. The operating conditions were set to a high vacuum environment with an accelerating voltage (HV) of 5.00 kV. Images were captured at magnifications ranging from 300x to 5000x using an Everhart-Thornley Detector (ETD) for secondary electrons. The spot size was maintained at 3.0, and the working distance (WD) was at the range of 5.1-6.6 mm.

Three-dimensional sample analysis with X-ray micro-computed tomography (XCT)

The XCT workflow consists of three steps (Fig. 1), namely (I) the acquisition of projection data and 3D reconstruction of the tomogram of the material23,24, (II) the material identification step also called segmentation25, and (III) the multi-scale analysis of morphological changes.

Fig. 1
figure 1

An illustration of the XCT workflow as applied in this work: the acquisition and reconstruction of grey-scale tomograms is followed by signal correction, sample boundary identification, further filtering and segmentation, opening the way for the extraction of morphological and physical properties (aspect ratio not to scale).

Full size image

Step I, image acquisition: Samples analyzed with XCT were PEG film before the plastic disintegration test (PEG_before) and PEG film after the plastic disintegration test (PEG_after). The sample films were mounted inside a 1.5 mm diameter capillary glass tube. Special care was taken to mount the tiny samples rigidly to avoid sample movements during stage turning. An acquisition time of 10.22 hours was used to collect 3600 projections at 50 kV with a 0.7 mm Al filter in circular acquisition mode. The reconstructed 3D images of these two films have a voxel size (volumetric pixel) of 683 nm.

Step II, Material identification: Analysing the three-dimensional macroscopic sample shape requires the definition of the sample boundary. Tomogram voxels outside the sample boundary are masked (orange voxels in Fig. 1). This mask is the result of the closing operation of the image A (defined as union of high-intensity voxels) by the spherical structuring element B according to mathematical morphology26. The closing operation consists of a dilation of A by B followed by an erosion operation and can be written as \(A \bullet B = (A \oplus B) \ominus B,\) , where \(\oplus\) notes Minkowski addition and \(\ominus\) stands for Minkowski subtraction. The remaining grey-scale voxels constitute the 3D reconstructed images of the films. For sample PEG_after the image resolution is sufficient to separate the microstructure into individual phases, also called segmentation. This is achieved with an active contour method operating on tomogram intensity and gradient25. Three material phases, namely air (low density), solid (high density), and an intermediate density phase are identified.

Step III, image-based properties: Given the macroscopic sample shape, a three-dimensional local material size of interest can be defined based on 3D distance maps. First, each voxel in the phase of interest (or the sample) is assigned the 3D shortest distance to the boundary masked voxel. This Euclidean distance map (EDM) represents the radius of the largest sphere that can be centered on each voxel of the phase of interest without intruding into other phases. Assigning each voxel in the phase of interest the largest radius of a sphere which can be placed fully within that same phase and covers that voxel defines the covering radius map (CRM). This three-dimensional CRM represents a 3D measure of local thickness, roughness and shape. Efficient algorithms for the derivation of the exact EDM are given in27and have been adapted for the calculation of the CRM in28.

Since the connectivity of the resolved pore space to the outside of the sample may play a role in the enhanced disintegration of the sample by microorganisms, an additional morphological analysis is carried out. Consider again a sphere as structuring element: Like for the covering radius map a radius is assigned to each voxel in the phase of interest, with one additional constraint: the sphere must be able to reach the respective location in the sample from outside the sample; a percolating path of at least diameter of the sphere must therefore exist. This closely simulates the injection of non-wetting mercury into a sample28,29, which is a capillary drainage process; consequently, we term the corresponding distance map a capillary drainage map (CDM). This three-dimensional CDM then identifies the connected pore spaces and removes isolated pores (shown in Fig. 9). Finally, the surface area density was calculated according to the method of30.

Results

In this section we first report the plastic disintegration test results of all samples, illustrating the relative performance of the PEG films. In the second section we then target the potential mechanism of disintegration of the PEG film specifically, utilizing micro-CT imaging technology and 3D morphological analysis.

Fig. 2
figure 2

Recovered films after six-month burial in solid waste.

Full size image

Plastic disintegration tests

Plastic disintegration tests were conducted on four specimens, namely commercial biodegradable plastic, filter paper, banana film, and PEG film. The specimens were buried and incubated for six months in plastic disintegration media containing compost, sawdust, rabbit feed, corn starch, saccharose, corn seed oil, and urea. Images of the recovered films after six months burial are presented in Fig. 2. The aluminium foil pieces at the top left side of the image illustrate the initial size of the samples. It is evident from the image that there was a size reduction after soil burial. Filter paper, which contained 100% cellulose, however, did not show a significant size reduction. This could be because the filter paper had a thickness of 0.25 mm, which was about 8-10 times thicker than the other samples, so that the rectangle shape of the filter was still preserved. Unfortunately, we were not able to measure the weight of the samples after the disintegration test because the bioplastics were hydrophilic and very fragile, and the dirt was strongly attached to them. When attempting to wash some of them, they further disintegrated, and it was almost impossible to retrieve them and obtain a reliable measurement.

The commercial biodegradable plastic sample was a certified compostable bioplastic from a synthetic biodegradable bioplastic combined with plant-starch – Ecopond with a thickness of 0.02 mm. Significant plastic disintegration was observed and only three small fragments were left after six months from five pieces of the films (Fig. 2). During the process, the film surface was cracking (Fig. 3), which increased the surface area for enzymatic or microbial attack.

Fig. 3
figure 3

SEM images of original and degraded samples.

Full size image

The higher degree of disintegration of the commercial biodegradable plastic relative to the filter paper is attributed to two possible reasons. First, the thickness of the commercial control film was less than 1/10th of the filter paper, resulting in a significantly lower surface-to-volume ratio. Second, the solid waste ecosystem might be richer in microorganisms that digest starch rather than cellulose. The banana film, which only contained nanocellulose, disintegrated into five small fractions as can be seen in Fig. 2. Some microorganisms with rod and coccus shapes were observed on the surface of the buried banana film (Fig. 3). As for the commercial biodegradable plastic, cracking was also observed in the banana film after six months. However, the observed cracks are less frequent and have apertures more than a factor 10 smaller. There were also micropores at the area where microorganisms were spotted. Size reduction in the banana film after soil burial is an indication that the banana pseudostem nanocellulose prepared in this research is biodegradable. The presence of film fragments at the end of the test did not necessarily mean that the nanocellulose could not be completely degraded since the filter paper, which was composed of cellulose, also did not fully degrade. The degree of size reduction of the PEG film was higher than the banana film (Fig. 2). Only four small pieces of the films were recovered after six months burial. The size of the remaining films of the PEG film was also relatively smaller than the size of the remaining films of the commercial biodegradable plastic. Again, cracking was observed after 6 months with apertures comparable to the banana film.

Micro-CT analysis of PEG films

To further elucidate the observed changes for PEG film, the original film (PEG_before) and the (partially) disintegrated film (PEG_after) were scanned using XCT. The reconstructed 3D images of PEG_before and PEG_after are visualised in Fig. 4.

Fig. 4
figure 4

Micro-CT images of original and degraded samples.

Full size image

While PEG_before appears homogeneous at the given resolution with little internal structure, PEG_after displays significant internal heterogeneity as well as macroscopic surface roughness. It can also be seen that the PEG_after sample is more porous than PEG_before. For further illustration, a video of the 3D image of the PEG_before and PEG_after is available as supplementary material. As in Fig. 4, the video shows that PEG_after exhibits larger pores than the original film. The pores seem to be connected along the axis of the film. The acquired tomograms (Fig. 4) display heterogeneity at multiple scales defining two analysis goals for morphological analysis. In the following, sample shape, volume, and local thickness at the macro-scale, as well as interior material distribution and connectivity at the micro-scale are analysed.

Consider first the evolution of the macroscopic sample shape. The boundaries of the sample are determined by morphological closing operations as introduced earlier. The remaining grey-scale voxels are the 3D reconstructed tomogram image films (Fig. 5a, b) and corresponding macroscopic sample shape.

Fig. 5
figure 5

Definition of sample shape and volume by application of morphological closing operations to the high-intensity phase. Slices view through 3D tomogram intensity with a voxel size of 683 nm.

Full size image

The orange-colored boundary of the sample is excluded in all further calculations such as intensity histograms or morphological analysis. Given the definition of the macroscopic sample shape, we calculate the EDM and CRM of PEG_before and PEG_after over the full sample domain to estimate 3D sample roughness. Visualisations for the EDM and CRM of PEG_before and PEG_after are given in Fig. 6, highlighting the heterogeneity in sample thickness for PEG_after in Fig. 6d.

Fig. 6
figure 6

2D slices through the 3D distance maps of the whole samples (union of all phases within the samples). Top: Euclidean distance map (EDM) of (a) PEG_before and (b) PEG_after. Bottom: Covering radius map (CRM) of (c) PEG_before and (d) PEG_after. Bright colors correspond to large distance values.

Full size image

The change in macroscopic sample shape can be quantified by extracting the statistics of regional inscribed radii (CRM) (Fig. 6c, d), which are given in Fig. 7. The distribution is volume weighted and each voxel within the sample contributes equally through its assigned inscribed radius to the thickness distribution. We note that this distribution is based purely on the sample shape and ignores internal porosity or material distribution. The mode of the regional film thickness distribution changes from about \(13\,\mu\)m for PEG_before to \(24\,\mu\)m for PEG_after. The PEG_after distribution function is also much broader indicating a significant increase in surface roughness with degradation. The latter can be seen by noting that the smaller radii are not associated with the thickness of the sample itself, but are inscribed radii for the near-surface voxels representing crevices. Some discrete entries can be seen for the largest radii in both distributions. These correspond to inscribed spheres representing maximal local film thickness; the volume weighting emphasizes these inscribed spheres.

Fig. 7
figure 7

Sample thickness distribution of the full sample as derived in 3 dimensional space by the covering radius map (CRM) (see corresponding Fig. 6c and Fig. 6d ).

Full size image

The change in material density over the sample can be quantified by comparing histograms of tomogram intensity. Intensity histograms and intensity profiles along the z-direction (averaging intensity over z-slices) are shown in Fig. 8. From the intensity histogram one notes that the lowest intensity of the material (air) coincides for both PEG_before and PEG_after in Fig. 8a. Both Fig. 8a and Fig. 8b show a reduced density of the film after degradation. The high-intensity tails seen in Fig. 8 have similar shape and correspond to bright spots (see Fig. 5a, b). They remain distinct after the disintegration process with respect to the surrounding material, albeit material density is reduced. This may be an indication of the effect of swelling of the sample due to water adsorption.

Fig. 8
figure 8

PEG_before, depicted with red, shows higher than PEG_after, depicted with black in the intensity history of the whole sample and the difference shows strongly on the average intensity along axis.

Full size image

A further analysis of the internal microstructure is feasible for sample PEG_after, where significant internal features are resolved (Fig. 9a). The film is significantly more heterogeneous after degradation and shows ring-like structures in the 2D cross-section; most rings are open and show the absence of any light-grey filling material. The resultant segmentation into three phases, namely solid (red), intermediate intensity voids (green) and resolved pore space (black) utilizing the active contour method is given in Fig. 9b. From visual inspection large “tunnels” show in PEG_after and the intermediate phase is often contained in larger voids, which are not necessarily connected to the outer film surface at the resolution of the tomogram (683 nm). Since the connectivity of the resolved pore space to the outside of the sample is an important consideration for determining the mechanism of potential biodegradation of the sample, percolation analysis utilizing the CDM is carried out on the resolved pore space (in 3D), shown in Fig. 9c as 2D cross-section. Note that closed pores appear as black as the corresponding distance is zero (not connected to the outside). The disconnected pore space is then merged into the solid (red in Fig. 9d). A comparison of the pore-size distribution of CRM and CDM for the resolved pore space is given in Fig. 10, demonstrating that many large pores are only accessible by smaller openings. Furthermore, since some of the pore space is disconnected to the outside, the corresponding CDM entry shows a large contribution at zero distance. Many pores are barely connected to the outer film boundary at image resolution. The percolation type connectivity analysis shows that 14.33% of the pore space was isolated from the total pore volume (48.24%) for the material PEG_after. In addition, the PEG_after shows 7% of resolved intermediate intensity material for the achieved voxel size of 682.6 nm, with intermediate material partially filling the isolated pore space. The specific surface area also increased dramatically after degradation from 0.012 \(mm^2\) to 5.01 \(mm^2\).

Fig. 9
figure 9

Partitioning of PEG_after for morphological analysis of the interior of the sample. Slices through the (a) tomogram, (b) three-phase segmentation (green shows the intermediate intensity material), (c) capillary drainage map (CDM), and (d) disconnected pore space and intermediate phase merged into the solid. The blue arrows indicate the merging of intermediate phase and disconnected resolved pore space into the solid.

Full size image
Fig. 10
figure 10

Total 3D pore size distribution (red) and the connected pore size distribution (black) for PEG_after based on morphological operations of CRM and CDM.

Full size image

Discussion

The tendency of PEG to migrate to the surface of a film has been reported by31. In this work we observed that the PEG sample increased in thickness by almost a factor of two. Furthermore, this increase in thickness was accompanied by the generation of large voids. Many of these larger voids gained a tunnel-like connection to the surface. However, a significant number remained closed at image resolution. This suggests a process where microbial activity evolved inside the sample leading to a process of void growth. The most logical explanation is that this process generated gas, and thus was accompanied by a build-up of internal pressure, with some of these gas bubbles eventually bursting through the sample surface, increasing surface roughness and surface area in the process. The results are in line with work by32and33, who previously found that PEG accelerated the biodegradation process of bioplastic.

The actual process of invasion of microorganisms into the sample is less clear. The microbial population on the surface of the PEG film was denser than the other samples and controls. Fig. 3shows how the microorganism covers the surface of the PEG film. Considering the hydrophilic nature of PEG and the observations of31 regarding the tendency of PEG to migrate to the surface, a likely mechanisms is that the movement of PEG was accompanied by the penetration of water and microorganisms to the inner gallery of the film, thus accelerating the biodegradation process. The PEG that migrated to the surface of the film may have attracted these microorganisms.

Conclusions

Banana pseudostem nanocellulose based film disintegrated and likely is biodegradable. The addition of plasticiser (polyethylene glycol) sped up the disintegration process. Based on the X-ray CT analysis for the PEG films, it was found that local thickness and roughness of the films increased due to disintegration. Material density decreased as shown by comparisons of intensity, surface area and surface density between the original material and the material after disintegration. Enlarged tunnels formed during the disintegration process, while no resolved porosity was observed before disintegration at image resolution. Some tunnels were connected to the outside and some were not (found from CDM map) and an intermediate density material often existed inside in internal porosity not connected to the outer surface of the material. The formation of these tunnels is likely a two-stage process due to the migration of the plasticiser to the surface of the film, followed by a build-up of gas from microbial activity. We suggest that the latter enabled the build-up of sufficient pressure to generate larger voids, which partially penetrated the sample of the surface, thus destroying its integrity.

More generally, this work showed the capacity of micro-CT methods to work with rather “flimsy” bioplastic films, achieving sufficient contrast, resolution and field of view to study the evolution of porosity over time including its connectivity. Future studies may address the variation of blend ratios, additives, and/or the design of different heterogeneities on the surface of bioplastics, which could work as initial attack vectors to disintegrate bioplastics more quickly. Furthermore, while we followed a disintegration protocol (ISO20200:2004) here, future studies may consider time-lapse 3D imaging for in-situ conditions. This should yield significantly more insights to the relative importance of disintegration/degradation pathways as function of time, while also coupling to the eventual bioplastics uptake by microorganisms, which may be associated with indigestion and accumulation rather than disintegration of micro plastics.