A Bio‐Mimetic Leaf Wetness Sensor from Replica Molding of Leaves

Plant diseases cause hundreds of billions of dollars in global crop production each year. Many plant diseases develop when pathogens germinate and proliferate in the fertile environment of excess water on the leaf surfaces. Accurate measurement of how long a leaf stays wet is important to assess the risk of pathogen infestation and decide on appropriate management tactics. Commercial leaf wetness sensors estimate the duration of leaf wetness by monitoring the water accumulated on the sensor surface. However, this one‐size‐fits‐all sensor surface does not replicate the wetting behavior of specific leaves resulting in an imprecise measurement of leaf wetness. Here it is developed a leaf wetness sensor with replica molded surfaces that closely mimic the leaf of interest and provide a more accurate measurement of leaf wetness duration compared to commercial sensors. This simple approach allows for better prediction of leaf wetness duration for each plant species, affording individualized and more effective disease forecasting for the management of plant diseases.


Introduction
Plant diseases cause an estimated 19% annual loss of crop production, at an annual cost of hundreds of billions of dollars. [1] Plant diseases on wild and agricultural species are favored by excess water on leaf surface, which provides pathogens with the environmental conditions needed to germinate and infect the host plant. [2] The longer a leaf stays wet, the higher the risk that disease will develop, because many plant pathogen propagules require several hours of continuous moisture to germinate and initiate infection. [3] For this reason, estimates of how long leaves DOI: 10.1002/adsr.202200033 stay wet after precipitation or irrigation (leaf wetness duration; LWD) are a critical component of many disease forecasting systems that allow for effective integrated pest management. [4] Such disease forecasting systems provide growers with guidance about the likelihood of development of specific diseases on particular crops, allowing appropriate management actions (e.g., agrochemical application) when needed to avoid crop loss, while also allowing growers to avoid unnecessary agrochemical applications that are environmentally and economically costly. Commercially available leaf wetness sensors commonly use changes in resistance or capacitance to measure the amount and duration of wetness that accumulates on a sensor surface. [5] However, existing sensors estimate leaf wetness duration from one-size-fits-all sensors of a particular shape, texture, and hydrophobicity; these traits strongly affect surface wettability and vary widely among plant species. [6] Leaf wetness sensors able to measure leaf wetness duration as actually experienced on leaves of diverse plant species would open novel opportunities for fieldbased research on the ecophysiology of plant-pathogen interactions and help improve disease forecasting systems.
The surface structure and surface chemistry of a leaf affects its wettability (hydrophilicity), which is quantified using the contact angle ( ) of a water droplet on the surface. [7] For example, micro-scale papillae covered with nano-scale wax crystals in the lotus leaf (Nelumbro nucifera) make it super-hydrophobic with a contact angle greater than 160°. [8] Such superhydrophobicity gives the leaf self-cleaning properties that has attracted attention as the subject of biomimicry and bioinspiration for engineering of superhydrophobic self-cleaning surfaces. [9] A variety of replica molding strategies are used to reproduce the hydrophic or hydrophilic surfaces of leaves, including both microscale and nanoscale features: nickel electro-forming and UV-nanoimprint lithography; [10] sol-gel, soft lithography, and hydrothermal imprinting; and a hybrid soft lithography with polydimethylsiloxane combined with polycaprolactone (PCL) and polyurethane acrylate (PUA). [11] Here, we introduce a new concept of leaf wetness sensors with a surface that is the exact replica of a leaf with the same wetting properties (Figure 1). The sensor consists of a circuit board with a printed interdigitated capacitative element made of copper, which is covered by two layers of polydimethylsiloxane (PDMS). The first layer provides insulation between the copper and the outside   The leaf of interest is mounted to a Petri dish using double-sided Kapton tape. PDMS (20:1 weight ratio) is poured over the leaf and left to cure at room temperature. B,F) Once fully cured, the negative leaf replica mold is cut and peeled from the Petri dish. C,G) The negative leaf replica mold is then pressed face-down (adaxial down) onto a PCB sensor that has been coated with uncured PDMS (10:1 weight ratio). For the PDMS imprint, we use a base to curing agent ratio of 20:1, whereas procedures typically use a PDMS ratio of 10:1 for both the negative replica and imprinted surface. [17] Previous work has shown that the complex venation patterns of natural leaves can be replicated in greater detail using a ratio of 20:1 versus templates synthesized in a ratio of 10:1. [18] To avoid excessive heating, which results in the leaves shriveling during curing, and consequently loss of fidelity of the imprint, the PDMS imprints were cured at 23°C for 48 h. The PCB sensor is then placed in an oven to cure at 60°C. D,H) Negative leaf replica mold is peeled away from PCB sensor exposing the imprinted surface.
atmosphere avoiding shorting of the interdigitated electrodes by the accumulated water. The second layer is the PDMS replica [12] of a leaf of a plant species of interest that gives the sensor surface the same wettability as the leaf. This approach allows the sensor to estimate leaf wetness duration more closely reflective of the real leaf. We have made sensors with replica of leaves from three different woody plants with different leaf-wetting properties, growing in close proximity in in coastal California: California bay (Umbellularia californica), Western Sycamore (Platanus racemosa), and a horticultural variety of escallonia (Escallonia x Iveyi). We compared the performance of the sensors with a commercially available sensor (PHYTOS 31; Meter Group, USA). We demonstrate that our biomimetic leaf wetness sensor provides accurate, species-specific measurements of the leaf wetness duration of plants over a broad range of leaf traits.

Leaf Replica Fabrication, Sensor Assembly, and Characterization
The first step in the sensor fabrication creates surfaces that replicate and have the same wettability of the leaf of interest (Figure 2). We use a simple PDMS double casting method. [12] We fabricate the negative template mold by covering the leaves of interest with uncured PDMS while they are attached to the bottom of a Petri dish, which collects the excess polymer (Figure 2A,E). After curing, we detach the negative PDMS mold, cut it to the shape of the leaf, and functionalize it with a fluorinated trichlorosilane to minimize adhesion and facilitate the final peel-off process (Figure 2B,F). [13] The negative PDMS mold forms the replica of the leaf surface on top of the interdigitated capacitative sensor that is coated with a thin layer (180 m) of uncured PDMS (Figure 2C,G). After curing, the final peel off process reveals the leaf wetness sensor with a sensing surface that is the exact replica of the chosen leaf ( Figure 2D,H). If needed, we then perform an oxygen plasma step to adjust the hydrophilicity of the PDMS surface to more closely match the wettability of the leaves of interest. [14] Oxygen plasma creates -OH groups on the surface of the PDMS to increase surface hydrophilicity. [14] OH groups on the PDMS surface have a lifetime of hours to days depending on conditions and the surface contact angle tends to plateau to an intermediate hydrophilicity. [15] We have not observed significant changes in hydrophilicity during our work, nonetheless, this is a weakness of this approach that can be corrected by creating more stable self-assembled monolayers on the PDMS to provide a long-term solution. [16] We fabricated replica surfaces for our biomimetic leaf wetness sensor using leaves from three co-occurring plant species (California bay, sycamore, and escallonia) that represent a range of leaf wettability (Figure 3A-C) and compared the replica surfaces to the surface of the PHYTOS 31 (METER GROUP, USA) commercial sensor ( Figure 3D). We quantified the wettability of the leaf surfaces by measuring the contact angle ( ) of a sessile water droplet imaged with a VHX-5000 Digital Microscope (Keyence, USA) and analyzed using Image J. [12] Measuring the contact angle of a sessile water droplet is a good predictor of how water will interact with and wet a surface [12] and we decided to use this measure to compare the leaves and the replica. We selected species with leaves that represented with a range of hydrophobicity depending on their leaf surface microstructure and surface chemistry. [19] While none of the leaves that we selected had extreme superhydrophobic or superhydrophilic contact angles, [20] they ranged from the hydrophobic California bay ( = 101°) (Figure 3E), followed by the less hydrophobic escallonia ( = 80°) (Figure 3F), to the somewhat hydrophilic West Sycamore ( = 66°) ( Figure 3G). For comparison, the surface of the commercial sensor is more hydrophilic than any leaves tested ( = 57°) (Figure 3H). This means that when exposed to atmospheric moisture, the commercial sensor will likely accumulate more water than the plant leaves and retains the water longer. To confirm www.advancedsciencenews.com www.advsensorres.com that the replica molding process creates a faithful replica of the target leaves, we measured the contact angles of the leaf replica ( Figure 3I-K) to ensure that they were within 3°of the contact angle of the original leaves ( Figure 3E-G). We did not find it useful to replicate the surface of the commercial sensor. To further characterize the results of the replica molding process, we acquired Scanning Electron Microscope (SEM) images ( Figure 3L-R) of the original leaves and the PDMS replica. The replicas appear to capture the important morphological traits of the leaf surfaces such as spacings of leaf cells and venation patterns. These patterns range from ≈10 um in the California bay leaf and replica ( Figure 3L,O) to ≈40 um in the Sycamore ( Figure 3N,Q). A detailed characterization of the leaves using SEM is beyond the scope of this work and is already present in the literature for many leaves. [21] Ultimately, replicas were considered to be high quality and appropriate for further characterization as sensors because their contact angle and wettability were the same as the contact angle and wettability of the original leaves. We then place the leaf replica on top of a custom-made printed circuit board that contains the capacitate part of the sensor. The sensor design and optimization is described in Figures S1 and S2 (Supporting Information).

Wetting Measurements
To test whether the replica of the leaves would mimic the wetting behavior of the real leaves, we examined the condensation behavior of water droplets generated in a custom-made humidity and dew-controlled chamber (Figure 4 and Figure S2, Supporting Information). After 30 min in the chamber, the same wetting patterns appeared on the leaves ( Figure 4A-C) and their corresponding replicas ( Figure 4E-G). The commercial sensor ( Figure 4D) showed more surface condensation than any of the leaves, as expected based on its greater hydrophilicity (contact angle 57°). These patterns are as expected because the amount of water condensation and retention depends on the physiochemical characteristics of the condenser surface of the leaves and the replica. Therefore, leaves and replica with lower contact angle are expected to condense and accumulate larger quantities of water than their more hydrophobic counterparts with higher contact angle. Measurements of the sensor capacitance acquired over 70 min ( Figure 4H) mirror the visible patterns of wetting in the images ( Figure 4A-G). This is consistent with expectations because, the change in sensor capacitance is a function of the amount of water condensed and accumulated on the sensor. Therefore, sensors with more wettable surfaces should also record greater change in capacitance (ΔC). As expected, the hydrophilic commercial sensor ( = 57°) captured a film of condensed water and yielded the greatest change in capacitance. After 25 min, the commercial sensor ( = 57°) exhibited a ΔC of 55%, higher than the moderately hydrophilic sycamorereplica sensor ( = 69°; ΔC = 52%), the moderately hydrophobic escallonia-sensor ( = 83°; ΔC = 32%), and the hydrophobic bay-sensor ( = 99°; ΔC = 9%). These trends are reflected in the close correlation between (a measure of the surface relative surface free energy) and sensor capacitance response (ΔC) (Pearson's r = −0.974). With both the commercial sensor and sycamore leaves being hydrophylic, the commercial sensor could provide appropriate estimates of leaf wetness for that species; however, the commercial sensor would overestimate leaf wetness for the more hydrophobic species by as much as 50% (Figure 4). The sensors appeared to be stable to prolonged wetting; we did not observe any swelling or change in transparency of the replica that would indicate excessive water absorption and an associated change in surface properties. These measurements indicate that hydrophilicity and hydrophobicity of the leaf and sensor surfaces affect both water condensation and accumulation as well as recorded change in capacitance of the sensor. They confirm that a sensor with surface properties closer to the leaf of interest will more accurately record water condensation and accumulation than a generic sensor.

Conclusion
Accurate measurement of leaf wetness duration is essential to plant disease forecasting systems used to directly inform management practices, including the use of agrochemicals and timing of irrigation systems. Commercial leaf wetness sensors estimate the amount of surface water and leaf wetness duration by measuring the change in capacitance of a surface that accumulates condensed water. However, the one-size-fits-all commercial sensors do not accurately reflect the variation in leaf traits among species that dramatically affects leaf wetness duration. Here, we have demonstrated that biomimetic leaf wetness sensors designed to closely replicate the physical and wettability properties of leaves of individual plant species can provide a more accurate measurement of leaf wetness duration than do currently available hydrophilic commercial sensors. Overestimating leaf wetness duration would erroneously elevate perceived risk of disease development, potentially leading to unnecessary application of fungicides with negative environmental and economic impacts. For crops with more hydrophilic leaves, underestimated leaf wetness could lead to missed opportunities to control disease. Custom-built sensors are inexpensive to producerequired materials cost about US $5 per sensor. The replica molding strategy we present is applicable to any type of leaf, and therefore is applicable to all crops, horticultural plants, or wild species that are the focus of monitoring or research. The resulting sensor more accurately measures leaf wetness and can therefore improve disease forecasting and facilitate integrated pest management. Future challenges include capturing additional leaf features such as angle an curvature that can affect water retention, ensuring a durable lifetime of the sensor to make sure its effectiveness does not fade after prolonged exposure to the environment, and potentially exploring the use of biomaterials for the replica molding process to aid in disposing the sensor. [22] These important aspects will be explored in future work.

Experimental Section
Printed Circuit Board Sensor Design: The PCB sensor interdigitated electrode design was printed on a 1.6 mm thick fiberglass substrate and with the sensor head area constrained to a 20 × 45 mm 2 area. Each electrode had a finger width of 1 mm and a gap space of 250 m. The PCB electrode side was covered with a 25 m protective coating of PDMS (10:1 weight ratio) by spin coating at 1000 rpm for 10 min. The coated PCB sensor was cured in an oven at a temperature of 60°C for 48 h.
Leaf Replica Fabrication: To create the negative mold, leaves were first removed from plants, rinsed under running water for 1 min, and dried with nitrogen gas. Within a few hours of collection, the leaf was attached to a disposable Petri dish using double-sided Kapton tape. PDMS (SYLGARD 184; Dow Corning, USA) was prepared by mixing the PDMS prepolymer and cross linker in a 20:1 weight ratio and degassed for 1 h in a vacuum desiccator to remove air bubbles. The 20:1 ratio PDMS was poured into the Petri dish containing the leaf; the Petri dish was then placed back into the vacuum desiccator to remove any further bubbles formed during the pouring process. The PDMS was cured in an oven in ambient air at a temperature of 25°C for 72 h. Once fully cured, the PDMS negative mold replica was cut using a box cutter and peeled from the template.
Sensor Assembly: To imprint the PCB sensor with the leaf pattern, the final step requires pressing the negative leaf replica mold into the PCB sensor. A second layer of PDMS was spun at 250 rpm for 60 s to yield a 200 m layer on the PCB sensor. Next, the negative mold leaf replica was treated with trichloro(1H,1H,2H,2H-perfluorooctyl)silane (PFOCTS) (97% Sigma Aldrich, USA) to act as a non-stick agent. Deposition of PF-COCTS was performed via gas-phase evaporation in a desiccator for 2 h. The treated negative mold leaf replica was then firmly pressed face-down into the coated PCB sensor and was cured in an oven at a temperature of 60°C for 48 h. Once fully cured, the negative leaf replica mold was peeled from the PCB sensor.
Plasma Treatment: Replica PDMS leaf surfaces required further treatment following sensor assembly. After coating, replica LWS were exposed to oxygen plasma treatment at a base pressure of 40 mTorr, oxygen flow rate of 17.6 sccm, and RF plasma power of 45 W for exposure times varying between 10 and 60 s.
SEM Imaging: All SEM images were taken using a FEI Quanta 3D field emission microscope (FEI, USA). Surfaces were observed at a power of 10 kV and spot size of 3.5 nm.
Contact Angle Measurements: Contact angle measurements were obtained using a VHX-5000 Digital Microscope (Keyence, USA) and analyzed using Image J (imagej.nih.gov). Deionized water was used to determine the surface energy of leaves and PDMS leaf replicas. Contact angle were measured on water droplets with a volume of 5 L.
Experimental Leaf Wetting Setup: Artificial dew formation was created in a controlled dew environmental chamber where meteorological factors could be regulated including ambient temperature and humidity. Within the chamber, a relative humidity of 80% was maintained via a closed-loop system involving an ultrasonic humidifier and humidity sensor. Each sensor was mounted at a 45°angle using a custom 3-D printed stage. To induce condensation, the sensor surface was set to a temperature of 36°C and controlled using a closed-loop system involving an infrared temperature sensor and Peltier cooler. Recordings of the fabricated LWSs based on replicating the surface of its corresponding leaf species were performed for a total of 30 min to reach full surface water saturation. Capacitance measurements were taken using the PCAP02 Capacitance-to-Digital Converter (ScioSense, The Netherlands).

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.