Microplastic induces soil water repellency and limits capillary flow

Soils are considered the largest sink of microplastic (MP) in terrestrial ecosystems. However, little is known about the implications of MP on soil physical properties. We hypothesize that low wettability of MP induces soil water repellency, depending on MP content and size of MP and soil particles. We quantified wettability of mixtures of MP and sand. The sessile drop method (SDM) was applied to measure static contact angle (CA) of MP and glass beads at contents ranging from 0 to 100% (w/w). The results are extrapolated to varying combinations of MP and soil particle sizes based on specific surface area. Capillary rise was imaged with neutron radiography quantifying the effect of MP on dynamic CA, water imbibition, and water saturation distribution in sand. At 5% (w/w) MP content, static CA exhibited a steep increase to 80.2° for MP 20–75 μm and 59.7° for MP 75–125 μm. Dynamic CAs were approximately 40% lower than static CAs. Capillary rise experiments showed that MP 20–75 μm reduced water imbibition into sand columns (700–1,200 μm), with average dynamic CA of 40.3° at 0.35% (w/w) MP content and 51.8° at 1.05%. Decreased water saturation and increased tortuosity of flow paths were observed during imbibition peaking at 3.5% (w/w) local MP content. We conclude, in regions with high MP content. water infiltration and thus MP transport are hindered. Local low wettability induced by MP is expected to limit soil wettability and impede capillary rise.

Terrestrial soils are connected to groundwater, rivers, coastal zones, and oceans (Bank et al., 2021). Although there is limited evidence of MP transport from soil to groundwater, the effects of MP on transport and fate in soil, as well as the effect of MP on soil processes, need to be studied (Chia et al., 2021). Among anthropogenic pathways of MP entering the soil, agricultural managing practices cause a broad dispersal of MP (Rillig, Ingraffia, et al., 2017). For example, plastic mulching and subsequent tillage can intersperse larger polymer residues, which become brittle over time (Bläsing & Amelung, 2018). Abiotically, wetting-drying and freezingthawing cycles facilitate particle movement through porous media (Majdalani et al., 2008;Rillig, Ingraffia, et al., 2017). Following the biotic pathway of MP entering soil, bioturbation leads to incorporation of MP from the soil surface into the soil matrix . Earthworms can act as vectors for vertical and horizontal distribution (Huerta Lwanga et al., 2017;Rillig, Ziersch, & Hempel, 2017;Yu et al., 2019). Thus, MP becomes part of the plant-soil continuum and a study on plant reaction related to the presence of MP in soils has shown changes in productivity and community structure (Lozano & Rillig, 2020). Such effect can partly be explained by alterations of local soil water dynamics and soil structure (de Souza Machado et al., 2019).
However, the extent of and the conditions under which MP are transported through soils remain unclear. Additionally, their effect, if permanently or temporarily deposited, on soil hydraulic properties and soil moisture dynamics is unknown. Pristine particles, primary or secondary, not abiotically nor biologically weathered but mainly incorporated by anthropogenic activities, pose a hydrophobic surface addition to soils, possibly increasing soil water repellency. The

Core Ideas
• There is high awareness of microplastic contaminating environments raising concern worldwide. • Microplastic is found in terrestrial systems, but little is known about soil environments. • Microplastics are inherently hydrophobic and represent a hydrophobic surface addition to soil. • Consequently, they are potentially increasing soil water repellency. • We present an analysis of microplastic within porous media and show effects on water dynamics.
surface tension of water (γ = 72.8 mN m −1 ; Kalin & Polajnar, 2014) is higher than of , for example, PET (γ = 48.4 mN m −1 ; Papakonstantinou et al., 2007), making the main soil constituent quartz (γ = 280-1,180 mN m −1 ;) (Brace & Walsh, 1962) much more wettable than PET. For comparison, the surface energy of PP is 31.1 mN m −1 (Feinerman et al., 1997). Thus, the hydrophobicity of MP is inherent and can affect water infiltration on the pore scale with complex feedbacks on transport and retention of MP. We hypothesize that MP enhances soil water repellency in soil regions with high MP content. The magnitude of this effect depends on MP content and MP and soil particle size. We quantified the wettability of varying mixtures of MP ranging from 20 to 125 μm and quartz particles ranging from 40 to 1,200 μm. The sessile drop method (SDM) was applied to estimate the static contact angle (CA) of mixtures of MP and glass beads, with MP content ranging from 0 to 100%. Capillary rise experiments were performed to quantify the effect of MP on water imbibition and dynamic CA. Additionally, time-series neutron radiography was conducted to quantify MP and water content as well as distribution during capillary rise.

Investigated media
We mixed MP and porous media in varying combinations and quantities to estimate the effect of MP on soil wettability (Table 1). As hydrophobic polymer, we used PET, later referred to as MP. The raw material was provided by Veolia Deutschland GmbH. It has been processed by a centrifugal mill (ZM200, Retsch GmbH) and air-jet sieved into the fractions of 20-75 μm and 75-125 μm. Particle size analysis (Microtrac, Retsch GmbH) showed irregularly shaped particles, ranging from near spherical to fibrous resembling a realistic composition of MP found in the environment. Size distribution and shapes are reported in Supplemental Files S1-S4. Quartz sand (Raneem, Sand-Schulz GmbH) in the size range of 700-1,200 μm was used to create samples mixed with MP. The sand was treated with hydrogen peroxide and washed with deionized water to exclude the effect of organic matter on wettability. Additionally, magnetic particles were removed from the sand with a magnet.

Static contact angle
The SDM was used to estimate static CA (Goebel et al., 2013). Monolayers of glass beads mixed with MP in increasing contents were attached to a microscopy glass slide by applying a hydrophobic, double-sided adhesive tape (Bachmann et al., 2000(Bachmann et al., , 2003. Droplets of 2 μl deionized water (n = 11) were dosed on the prepared porous surface via an automated syringe and simultaneously recorded from a side angle with a highspeed, high resolution camera (device: drop shape analyzer DSA30S, Krüss GmbH). Recordings of each droplet geometry were subsequently analyzed and measured at the three-phase contact line to derive mean static CA 500 ms after detachment from the syringe by averaging the left-and right-side CA using an elliptic shape recognition mode (software: Advance, Krüss GmbH). To exclude size effects on static CA during SDM measurements (Bachmann et al., 2000(Bachmann et al., , 2003, MP was mixed with glass beads of the same size range. The MP 20-75 μm was mixed with glass beads of diameter 40-70 μm, and the MP 75-125 μm was mixed with glass beads of diameter 100 μm (Table 1).

Surface ratio extrapolation
To estimate the effect of MP on soil wettability, SDM measurements of static CA were extrapolated based on the geometric approach of Bachmann and McHale (2009). The purpose of this extrapolation is to obtain a first guess of the CA of MP-soil mixtures for varying MP concentrations. According to Bachmann and McHale (2009), the CA of a heterogenous rough surface can be predicted with a modified Cassie-Baxter equation: where CA net e is the observed equilibrium static CA (˚), φ is the Cassie solid fraction, and CA w is the Wenzel CA (˚). Here, surface protrusions into which the liquid completely penetrates are Wenzel-like and liquid bridging between those protrusions are Cassie-like. Furthermore, φ is described by a parameterized packing of spherical geometries predicting their wetted portions and their wetted planar projection subject to the Young's equilibrium CA (CA Y ) defined as the CA measured for a homogeneous, planar, and smooth surface. The Cassie solid fraction φ Z is defined according to Bachmann and McHale (2009) where the index Z is equal to S for the soil particles and MP for microplastic particles, and the ε i parameter represents the average distance between single particles being zero for connected particles. (Bachmann & McHale, 2009). The sensitivity of the extrapolation to this parameter is discussed in Bachmann and McHale (2009). In the case of MP and porous media, this equation is extended to a composite with two constituents, soil and MP, according to where CA net e (˚) is the static, fitted CA. Parameter f Z (kg m −2 ) is the respective surface fraction of MP and soil obtained by weight percentage (% i ), density (ρ i, kg m −3 ), average diameter of particles (d i , m), and shape factor (SF i ): The shape factor SF Z was set to 1 for spherical glass beads and to SF MP = 0.5 for MP (which have an irregular geometry; Supplemental Files S1-S4). The parameters ε S and ε MP were adjusted by manually fitting the measured static CA as a function of MP. For the extrapolation to varying MP and soil particle size, we kept SF S = 1 for sand and silt and we used ε MP = 0.85 fitted for the small MP size and ε S = 0.3 for sand and silt particles. The Young's CA (CA Y MP ) for PET was set to 74.3˚ (Zumstein et al., 2016). Based on SDM measurements of the used media, CA Y S was estimated as 30˚for soil particles, 38˚for 100-μm glass beads, and 50˚for 40-to-70μm glass beads. The parameters used for the calculations are shown in Table 2.

Dynamic contact angle
Capillary rise of water and ethanol into glass cylinders packed with MP-amended sand was captured with a DCAT 11EC tensiometer (DataPhysics Instruments GmbH, software: DCATS 32), which records mass increase over time by the imbibing liquid. Glass cylinders (0.8 × 5 cm) were filled with pure sand (static CA = 35˚, n = 5), sand mixed with MP (25-75 μm) of gravimetric contents of 0.35% (n = 3) and 1.05% (n = 3). Respective dynamic CA for each treatment was calculated by solving the Lucas-Washburn equation for water and ethanol according to the description of Liu et al. (2016). The modified Lucas-Washburn equation was used, resulting in a mass-gain fitting function (Liu et al., 2016, Method 3): where m is mass (kg), r is effective pore radius (m), γ is surface energy (N m −1 ), CA is dynamic CA (˚), ρ is density (kg m −3 ), N is the number of equivalent capillaries, g is the gravitational constant (m s −2 ), τ is tortuosity, W is Lambert function, η is dynamic viscosity (N s m −2 ) and t is time (s). The CA was estimated by fitting the water imbibition, with r and τ obtained by fitting the ethanol imbibition assuming a zero CA. The resolved equation for obtaining the dynamic CA states: The parameter κ summarizes the combination of constants B, C, and D used in Method 3 described by Liu et al. (2016), where index W refers to water and ETH to ethanol (Liu et al., 2016).

Neutron imaging
Time-series neutron radiography of capillary rise into rectangular aluminum containers (inner dimensions: 0.6 × 1.6 × 6 cm) was conducted at the ICON (Imaging with cold Neutrons) beamline at the Paul Scherrer Institute in Villingen, Switzerland (Kaestner et al., 2011). The amended gravimetric MP contents were 1.05% (mid) and 2.10% (high). Imbibition of water was captured for three replicates of different MP content with an acquisition time of 0.5 s and a pixel size of 58 μm. The sample containers were prefilled with sand and MP was added from the top followed by concussing the sample for MP to enter the sand packing resulting in a heterogenous distribution of MP over depth of the sample (Figure 1a). This heterogeneity was preferred to a wet packing to avoid residual water after drying affecting liquid imbibition. The procedure of determining MP in soils is highly labor intensive, and destructive to the soil structure, hence quantities reported in literature resemble bulk soil averages with no information about the locality of MP within the soil matrix. In contrast, neutron radiography is a nondestructive imaging method that is highly sensitive to hydrous materials and, thus, optimal to image water distribution in porous media (A. Carminati et al., 2007;Kaestner & Schulz, 2015;Pleinert & Lehmann, 1997). Differing neutron attenuation coefficients with respect to material compositions allow for detecting quantities and distributions of sample constituents of the sample (A. Carminati et al., 2007). Acquired gray value data were corrected for neutron scattering using a black body bias correction (Boillat et al., 2018;C. Carminati et al., 2019;Kaestner & Schulz, 2015). The images were subsequently normalized to the open beam signal, spot cleaned, noise reduced, and filtered (Boillat et al., 2018;Kaestner & Schulz, 2015). Subtracting gray value signals of the dry sample references from their wet counterparts allows us to quantify liquid configuration. (A. Carminati et al., 2007;Kaestner & Schulz, 2015). The attenuation coefficient of water was derived from images of a step-wedge with voids of defined thickness filled with the respective reference liquid. The attenuation coefficient was used to calculate water content and saturation based on porosity by subtracting gray value signals of the dry sample references from their wet counterparts (A. Carminati et al., 2007;Kaestner & Schulz, 2015). Local MP content distribution in dry samples was derived from the difference in the average signal of the control treatment (n = 3) and the signal of MP amended sand. The attenuation coefficient of MP was fitted to match the total mass of MP applied to each replicate (n = 3 for 1.05% [w/w] MP and n = 3 for 2.10% [w/w] MP).

Static contact angle and extrapolation
The SDM measurements of static CA showed a decrease in wettability upon the addition of MP to glass beads ( Figure 2). Mixtures of smaller and mixtures of larger MP particles followed the same trend with a steep increase followed by a steady gain in CA with accreted MP content. The smaller MP particle mixtures exhibited a consistently reduced wettability for all MP contents compared with the larger MP particles treatment. The offset is explained by the higher surface roughness of the smaller particles. Bachmann et al. (2000Bachmann et al. ( , 2003 and Bachmann and McHale (2009) showed that the SDM provides similar results when compared with other methods, such as the Wilhelmy-Plate-Method. We used the SDM to create a baseline for obtaining a preliminary characterization of MP effect on soil wettability with regard to particle size.
We used Equation 3 to extrapolate the results of Figure 2 to homogenous mixtures of soil constituents and MP with varying particle sizes. We found that hydrophobicity can be induced by low amounts of MP when MP particles are smaller than soil particles (Figure 3). The larger the ratio between soil particle size and MP particle size, the less MP is necessary to render soil water repellent. This is explained by the specific surface scaling with the inverse of the particle diameter, as assumed in Equation 3. This preliminary result based on the extrapolation in Figure 2 suggests that soil wettability can be highly susceptible with regards to MP. Potential increases of soil water repellency are not only due to MP content but, furthermore, depend on the relationship F I G U R E 2 Static contact angle (CA) of monolayers of glass beads mixed with microplastic (MP). Polyethylene terephthalate (PET) 20-75 μm and 75-125 μm in diameter was mixed with glass bead of similar size (Tables 1 and 2). The 90˚wettability threshold is indicated by the black dotted line, single measurements by small dots (n = 11) and mean of each treatment by large dots. Small and large size MP mixtures are shown in blue and magenta, respectively. Lines represent the CA derived according to Equation 3 (small MP: R 2 = .93, large MP: R 2 = .90). SDM, sessile drop method F I G U R E 3 Contact angle (CA) extrapolation of a coarse sand (1,000 μm, left) and a medium silt (diam. = 10 μm, right) combined with increasing gravimetric content of polyethylene terephthalate (PET) of different diameter. The blue line reflects the largest difference in grainsizes between the two constituents and the largest addition of water repellent specific surface area for a given microplastic (MP) content. The thick black line on the left represents the combination of substrates chosen for further analysis between soil texture and MP particle size. However, Figure 3 is an estimation assuming that the effective CA of composite materials scales with the specific surfaces of the respective components, which needs to be tested. To test the reliability of this estimation, the method is compared with the capillary experiments shown in the paragraphs below.

Dynamic contact angle
The effective dynamic CA was derived for capillary rise in sand columns of varying MP (20-75 μm) content according to the description of Liu et al. (2016). Microplastic induced an increase in dynamic CA. However, the CA was about 40% lower than the extrapolated static CA based on Equation 3 ( Figure 4). The discrepancy in dynamic and static CA can be explained by the errors in the extrapolation of Equation 3, which assume that the CA scales with the CA of the sample components scaled by the respective specific surfaces. Additionally, the discrepancy is caused by the difference in sample geometry of the two methods. While a droplet of water placed on a fixed, pseudo-two-dimensional surface of glass beads and MP particles cannot bypass low-wettability MP surfaces during with mean dynamic CA (crosses) and individual measurements (dots) measured with the capillary rise method. The control is in black, 0.35% gravimetric microplastic (MP) content is in blue, and 1.05% gravimetric MP content is in magenta. The amended MP particles were polyethylene terephthalate (PET) 20-75 μm in diameter. The static CAs are consistently higher than the dynamic CAs, which is explained by the difference in the sample geometry (two-vs. three-dimensional) and flow dynamics spreading, water can bypass and even redistribute MP particles during imbibition in a three-dimensional packing of sand (i.e., as can be observed by use of time-series neutron radiography; see next section). This would explain a lower CA for the capillary rise.

Neutron imaging
Neutron radiography of capillary rise in sand columns of varying MP content of 20-75 μm in diameter provides information on the spatial distribution of water in soil samples mixed with MP. The images show a decline in water infiltration rate and a decrease in the final water content for increasing MP content ( Figure 5). Analyzing the MP distribution in the samples revealed a heterogeneous distribution of MP particles across the sample profile (Figure 1a). This heterogeneity resulted from the packing method. However, although it was not intended, it allows exploring the effects of MP on soil water content over a larger range MP content and determine the effect on water distribution (Figure 1a).
The final water saturation decreased with increasing MP content showing marked differences above 2.2-cm elevation (Figure 1b). To a height of 2.2 cm, signals are not distinguishable, as all samples are saturated. At greater elevation, the signals start to deviate from each other as increasing local MP contents restrict local imbibition. Figure 6 shows the relation between local MP content and water saturation after 420 s of capillary rise for five ranges of matric potentials estimated based on elevation above the water table. At each matric potential, saturation decreased with increase in MP content. Furthermore, MP treatment increased the observed range of water saturation with comparably large variation of observed saturation at the most negative range of matric potential.
Subtracting the water saturation in every pixel of control treatments from spatially corresponding values of the MP treated sand shows the differences in water saturation induced by MP (Figure 7). At every matric potential, as the MP content increases, water saturation is decreased. For the chosen combination of soil and MP particle sizes at a critical value of approximately 3.5% MP content, the decline in saturation due to MP is diminished. Gravimetric MP contents above 3.5% make it increasingly difficult for water to infiltrate by (a) simply blocking the pore volume and (b) critically reducing local wettability and preventing capillary rise.

DISCUSSION
We found that pristine PET particles induce soil water repellency, and that such an effect is soil and MP particle size specific. We observed reduced water saturation for local gravimetric MP contents between 1 and 3.5% for the combination of particle sizes of 20-to-75-μm PET and 700to-1,200-μm sand. Results of time-series neutron radiography highlight the effect of a heterogeneously distributed MP particles on water flow. Water infiltration can be impeded, and air can be entrapped, resulting in decreased water saturation and decelerated water infiltration as flow paths become more tortuous. Thus, the mobility of MP particles in MP hotspot regions is likely decreased when compared with other parts of the soil pore space where wetting and drying cycles might cause the displacement of MP. Furthermore, degradation of MP in such regions with high MP content is likely to be reduced due to the damped microbial activity at low soil water content. Average MP contents above 3.5% are unlikely to occur at the scale of the soil profile, and we do not expect to observe MP-induced water repellency at this scale. However, similarly to our experiments in which MP was heterogeneously distributed, we expect heterogeneous distribution of MP in soils, with hotspots of high MP content. In such regions, MP might occur in amounts sufficient to limit water flow and reduce local water content, causing the extended retention of MP in such regions. Furthermore, low water content in these microregions might extend the period necessary for MP degradation by hydrolysis or enzymatic microbial activity F I G U R E 5 Water content during capillary rise in microplastic (MP, 20-75 μm) amended sand (700-1,200 μm). The control is in black, medium MP content is in blue, and high MP content is in magenta. Mean values of velocity (v), water saturation (Θ), and volumetric water content (θ) after capillary rise to a height of 4.65 cm are indicated by the dotted lines in color code. The shaded outlines resemble the 95% confidence interval of the mean. The higher the total MP content the lower are velocities, water contents, and saturation F I G U R E 6 Water saturation (Θ) in each pixel (dots) against local microplastic (MP, 20-75 μm) content (dots in color code) for different ranges of matric potential (Ψ = matric potential). Faces in the background represent the mean water saturation distribution of control treatments at the respective matric potential range (light version of respective color code legend) and delay coating of MP by soil abundant substances like iron or organic compounds.
With regards to transport of MP particles during imbibition, we cannot exclude displacement of MP during imbibition. Because of their inherent low wettability, pristine MP tend to accumulate at the air-water interface and can be displaced by water during wetting and drying cycles. Water flow through pores might occur until accumulated MP particles create a hydrophobic patch forcing water to bypass. Nevertheless, from observations as well as coinciding analysis of the MP content related water saturation, the effective transport is considered minor.
Our conclusions are based on observations made using pristine MP particles and artificial soils with no organic matter, Vadose Zone Journal F I G U R E 7 Difference in each pixel (small dots) between the water saturation values of the mean control treatment (Θ C ) and microplastic amended variants (Θ MP ) at the corresponding range of matric potential (Ψ) clay content, and soil structure. Although MP particles are mostly introduced uncoated, in natural soil systems, they can be coated over time with soil abundant compounds (i.e., dissolved organic substances, iron oxides or hydroxides, and/or microbial biofilms). These processes potentially increase MP wettability and facilitate degradation. Further studies should elucidate the temporal dynamics of MP wettability in natural, undisturbed soils under varying environmental conditions (soil moisture content, temperature, chemical composition of the soil solution, and soil structure).

A C K N O W L E D G M E N T S
This study was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)-Project Number 391977956-SFB 1357. We kindly acknowledge Subproject Z01 for providing microplastic particles.

C O N F L I C T O F I N T E R E S T
The authors declare no conflict of interest.