Keywords

1 Introduction

Bioinspired deployable sensor networks for spatio-temporal monitoring of environmental parameters are gaining much attention and study lately [1,2,3]. Among several biological models, flying plant seeds are becoming more and more attractive for this application scenario. These plant seeds constitute a model of morphological computation found in nature [4]. It is, in fact, their morphology and structural features that give them the ability to passively fly and be dispersed by the wind, without the need of any energy input from inside [5].

Acer samara seeds were taken as inspiration to develop deployable auto-rotating sensors for wildfire detection [6], or airborne sensors for gathering atmospheric parameters, as temperature, air pressure, relative humidity and wind speed along their descent [7]. These solutions strongly rely on electronics, which generate e-waste; moreover, they are heavy and energy consumptive. Taraxacum seeds were chosen as a model by Iyer et al. to build battery-free sensing devices [1]. Although they are lightweight and solar powered, they are based on electronics and not biodegradable materials. Tristellateia seeds acted as a source of bioinspiration for Kim et al. to develop battery-free electronic and colorimetric microfliers for ultraviolet exposure and pH sensing [2], that were not biodegradable. Yoon et al. developed colorimetric fliers inspired by Tristellateia and Taraxacum seeds for sensing pH, heavy metal concentrations, ultraviolet exposure, humidity and temperature [3]. These fliers are biodegradable, yet some reagents used for sensing are not biodegradable and may be even toxic (e.g., S-8028, containing Cobalt dichloride). Alsomitra macrocarpa seed was taken as a model for developing fully biodegradable artificial seeds for visual pH sensing of rainwater [8].

The aim of the present work is to develop biomimetic, deployable, lightweight, fully biodegradable and 3D printed artificial seeds for the sensing of an environmental parameter, such as humidity, via optical readout. The artificial seeds are bioinspired by Ailanthus altissima seeds and they are made of porous cellulose acetate (CA). A sensor layer, made of photonic cellulose nanocrystals (CNC), is applied on one side of the artificial seed for humidity sensing of topsoil.

Porous tissues, i.e., aerenchyma, are common in plants, and they are also present in flying seeds [9], likely with the aim of enhancing lightness. We think it is important to reproduce this feature artificially and to tailor the positioning of porous structures by means of 3D printing. Cellulose acetate was dissolved in acetone [10] and mixed with lignin particles, which were later removed by water (pores formation by leaching technique). The degree of porosity was then characterized. Both cellulose acetate and lignin are biodegradable, as reported in literature [11,12,13].

Next, we employed the porous material for the 3D printing of an artificial seed bioinspired by Ailanthus altissima seed, also referred as samara. Differently from an Acer samara seed, which falls by autorotating around its vertical axis, Ailanthus altissima samara rotates both around its longitudinal axis and its vertical axis [14]. Moreover, Ailanthus altissima seed has another important flight mode, that is tumbling, that allows it to travel on the xy plane without the need for the wind [15, 16]. Ailanthus altissima seed is constituted of an actual seed part, placed at the center of an eye shaped wing. The wing has a twisted structure, with one side being twisted, to help lateral transport even if on the ground [14]. We performed a morphometric and aerodynamic analysis on natural Ailanthus altissima seeds. Based on the extracted data, we designed and 3D printed biomimetic Ailanthus altissima seeds using the previously developed material. We, then, analyzed the morphometry and aerodynamics of the artificial seeds to check their compliance with the natural seeds. The process of designing, 3D printing and testing took place iteratively until a behavior similar to the natural seeds was reached.

We developed a photonic crystal based on Cellulose Nanocrystals (CNCs) [17] for colorimetric sensing of humidity [18]. The sensor was added on the artificial Ailanthus altissima seed for the monitoring of humidity oscillation by analyzing the changes of the reflected color through a spectrometer-based and/or colorimetric analysis.

2 Experimental

2.1 Preparation and Characterization of the Porous Cellulose-Based Material for 3D Printing

A number of four solutions made of distinct ratios of cellulose acetate (30000 MW) (Sigma-Aldrich, Germany), and lignin (Alkaline) (TCI Europe N.V., Japan), were prepared.

First, a cellulose acetate batch solution was prepared by mixing in a beaker cellulose acetate in acetone at 30% w/w. The beaker was closed with Parafilm and aluminum foil and was put on a magnetic stirrer (Thermo Fisher Scientific Inc., USA) at 50 ℃ at 30 rpm for 1 h. The batch solution was split into 4 parts in which different amounts of lignin were added: 0, 33.3, 50, 66.6% of the cellulose acetate weight. Each new solution, unless the first, was closed and stirred at 50 ℃ at 30 rpm for 1 h.

The four solutions were used as 3D printing material for the Direct Ink Writing (DIW) process on a 3D-Bioplotter (EnvisionTEC, USA and Germany). Printing temperature and build plate temperature were set at 20 ℃, printing speed was set at 25 mm/s and the diameter of the used nozzle was 0.4 mm. The four compositions were printed at different pressures: 1.2, 1.4, 1.8, 2.2 bar for the 0, 33.3, 50, 66.6% solutions, respectively, because of the increasing viscosity of the cellulose acetate-lignin solutions. For brevity, we will refer to these solutions as 0-lig, 33-lig, 50-lig and 66-lig.

Square specimens (10 × 10 mm base) with three different thicknesses, 0.1, 0.5, 1 mm, (5 specimens per thickness), were printed for each of the 4 compositions, using the preset printing parameters; for a total of 60 specimens. They were then dried in oven (Vacutherm, Thermo Electron LED GmbH, Germany) at 70 ℃ for 30 min, to let all the acetone evaporate. Then, they were weighted with an analytical balance (Practum, Sartorius AG, Germany). The thickness of the square specimens was measured with a digital caliper (RS PRO 150 mm Digital Caliper, RS Components Ltd., UK) to derive the volume. Consequently, each specimen was put in a plastic petri dish filled with deionized water and left at rest for intervals of 1 h, 4 times, and for one last interval of the duration of 16 h, to allow the release of lignin in water, (5 intervals for a total of 20 h). After each interval, the specimens were dried, first with adsorbent paper, then in oven (Vacutherm, Thermo Electron LED GmbH, Germany) at 70 ℃ for 30 min, and their weight was measured with an analytical balance (Practum, Sartorius AG, Germany).

The evolution of weight in time was recorded and computed. Moreover, with the measured weights and volumes, porosity of the specimens was derived by the following relationship:

$$P=1- \rho /{\rho }_{AC}$$
(1)

where P is the porosity of the specimen, ρ is its density and ρAC is the density of cellulose acetate as reported by the manufacturer (1.3 g/cm3).

2.2 From Natural Flying Seeds to 3D Printed Artificial Flying Seeds

Ailanthus altissima seeds were collected from a tree in Morego (Genoa, Italy). The morphometric analysis was performed on n. 8 seeds. Dimensions of the seeds, i.e., longitudinal length (LL), transversal length (LT), wing thickness (Thw), seed capsule thickness (Ths) and diameter (D), were measured using a digital caliper (RS PRO 150 mm Digital Caliper, RS Components Ltd., UK). Mass of the seeds (M) was assessed with an analytical balance (Practum, Sartorius AG, Germany). Wing surface (S) was estimated with ImageJ from pictures of the Ailanthus altissima seeds taken with a camera (1280 × 800 pixels) of a Samsung A40 (South Korea) smartphone (Fig. 2c). The wing loading (W/S) was calculated from the weight value (W) of the seed and the wing surface (S). Microscope images of the wing were taken with a digital microscope (KH-8700, Hirox, Japan).

An aerodynamic analysis in laboratory conditions was performed to determine the descent speed (vd) in the spirally twisted flight mode of n. 8 Ailanthus altissima seeds. The seed was released from rest, in a laboratory without active ventilation, from a height of 2.95 m and allowed to fall freely. The flight of the seed was recorded by a camera of a Samsung A40 (South Korea) with a resolution 1280 × 800 pixels. The mean \({v}_{d}\) was calculated considering the time elapsed between the frame of the release and the frame in which the seed touches the ground. Each individual seed was tested 3 times, giving a total of 24 drops.

For the biomimetic design of the Ailanthus altissima artificial seed, a top view picture of a natural seed was taken with the camera (1280 × 800 pixels) of a Samsung A40 (South Korea) smartphone. The picture was used for the extraction of a vector file of the contours, with the free and open-source vector graphics editor Inkscape. The vector file of the contours was consequently imported in the 3D CAD modeling software Siemens NX, and the design of the artificial samara seed was developed taking into consideration the morphometric analysis performed on the natural seed. For the 3D printing of the artificial seed, a Direct Ink Writing (DIW) process was employed with the 3D printer Bioplotter (EnvisionTEC, USA). The CAD model was converted in STL format, sliced with Perfactory RP software and imported in the 3D printer software VisualMachine, where the build instructions were set. The material used for the printing of the artificial Ailanthus altissima seeds was the 50-lig solution and the set printing parameters were those above-mentioned for that composition.

The same morphometric and aerodynamic analysis conducted on the natural Ailanthus altissima seeds were conducted also on the 3D printed 50-lig artificial seed. Subsequently, the same seed was put in deionized water for 20 h for lignin removal, dried with adsorbent paper and in oven (Vacutherm, Thermo Electron LED GmbH, Germany) at 70 ℃ for 30 min. Again, morphometrics and aerodynamics were conducted on the artificial seeds. In addition, an estimation of porosity was performed by considering the nominal value of the volume derived from the CAD model and the mass derived from measurements.

2.3 Cellulose Nanocrystals (CNCs) Photonic Crystal Humidity Sensor Fabrication and Characterization

Cellulose nanocrystals (CNCs) were purchased from Cellulose Lab (Canada). CNC were prepared by hydrolysis with sulfuric acid and they had the following features: width 5–20 nm and length 100–250 nm. CNCs were dissolved in deionized water (7.5% w/w) and magnetically stirred (1000 rpm for 72 h). Then Glycerol (5% w/w) was added to the CNCs solution and magnetically stirred (1000 rpm for 72 h). The solution (3 ml) was then casted in circular plastic Petri (3 cm in diameter). After evaporation at room temperature (20 ℃) for 2–3 days, free-standing films with iridescent and photonic colors were obtained.

A square section of the CNC photonic crystal film (4 × 4 mm) was cut using a razor blade and attached onto the artificial seed using a 1 µL of Cellulose Acetate in acetone (30% w/w) as glue.

The reflectance and the shift of the photonic bandgap over different humidity variations (30%, 60% and 90%) were analyzed through spectrometry using an optical setup previously described [19] and consisting of: spectrometer FLAME-S-XR1-ES (200–1025 nm) (Ocean Insight, USA), light source SciSun 300 Solar Simulator (Sciencetech, Canada) working at 0.5 Sun (0.5 mW/cm2); and a bifurcated fiber-optic probe (diameter = 600 µm) (Ocean Optics, USA). Light exiting from the light source is fed through one arm of the bifurcated fiber-optic probe orthogonally to CNC surface and the reflected light was collected through the other arm of the bifurcated fiber-optic probe into the spectrometer that yields the reflectance spectra. A protected silver mirror (Thorlabs, USA) was used for normalization. Relative humidity was changed (30%, 60% and 90%) using a climatic chamber (CTC256, Memmert GmbH, Germany) with the temperature fixed at 30 ℃.

The colorimetric changes of the photonic bandgap over different humidity value (30%, 60% and 90%) were also captured by a Logitech Brio Stream, Logitech (Swiss), orthogonally placed onto the artificial Ailanthus altissima seed, and elaborated with Editor Video (Windows). CIE 1931 chromaticity diagram was plotted from spectral data using an online tool [20]. Colorimetric analysis of the CNC photonic crystal over the explored humidity variations was carried using ImageJ software analysis [21].

3 Results

3.1 Characterization of the Porous Cellulose Material

Weight loss due to lignin release was recorded over a period of 20 h for the four cellulose acetate-lignin compositions (0, 33.3, 50, 66.6% lignin) and for three different thicknesses (0.1, 0.5, 1 mm), as shown in Fig. 1.

For the 0-lig specimens (Fig. 1a) a very slight decrease in mass, constant in time, is observed, due to degradation of the cellulose acetate network. A mean mass loss of 3.8, 5.3 and 4.7% is recorded for thicknesses of 0.1, 0.5 and 1 mm, respectively. Porosity of the 0-lig specimens went from a mean initial value of 29.1, 47.7, 48.4% to a mean final value of 31.8, 50.5, 50.9%, for thicknesses of 0.1, 0.5 and 1 mm, respectively, as shown in Table 1.

In the 33-lig specimens (Fig. 1b) a greater loss was measured, due to the removal of most of the lignin, in addition to the degradation of CA. The mean recorded loss was 36.1, 32.9 and 30.5% for 0.1, 0.5 and 1 mm thicknesses respectively. If we correct these values with the percentages found for the pure CA (0-lig), we obtain a mean loss of 32.3, 27.5 and 25.7% respectively, caused by lignin release only. This indicates a residual fraction of lignin, 1, 5.8 and 7.6% respectively, still trapped in the CA network, as can be seen visually in Fig. 1e. The final mass of the 0.1 mm specimens is reached within the first hour of lignin release in water, due to the high surface to volume ratio. While for the 0.5 and 1 mm specimens, 90 and 81% of the total loss, respectively, happened in the first hour, indicating a slower release for lower surface to volume ratios. Final mean porosities for the 33-lig specimens were 68.5, 66.2, 58.9% for 0.1, 0.5 and 1 mm thicknesses respectively, as shown in Table 1.

As regards the 50-lig specimens (Fig. 1c), the recorded mean loss in mass was 53.5, 54.2 and 53.7% of the initial masses for 0.1, 0.5 and 1 mm thicknesses, respectively. Correcting with the degradation percentage of pure CA, the mean mass loss provided only by lignin releases were 49.7, 48.8 and 49.0% respectively. Only 0.3, 1.2 and 1% of lignin, respectively, stayed inside the cellulose acetate network, indicating a greater release of lignin due to a less packed CA network. Within the first hour, the mean mass loss of the 0.1 mm specimens was complete, while for the 0.5 mm specimens was 91% of the total mass loss. For the 1 mm specimens it was just 69% in the first hour and went over 90% after only 3 h. Porosity reached the values of 79.8, 73.3, 76.3 for 0.1, 0.5, 1 mm thickness, respectively, as shown in Table 1.

The 66-lignin specimens exhibited the major loss in weight: 69.6, 68.7 and 69.1% for 0.1, 0.5 and 1 mm thicknesses, respectively. The corrected mean values were 65.8, 63.3 and 64.4% respectively, with a fraction of residual entrapped lignin of 0.8, 3.3 and 2.2%. For the first hour, 99% of the total mass loss was reached for the 0.1 mm specimen, 93% for the 0.5 mm specimen and only 59% for the 1 mm specimen. Porosity increased to a maximum value of 89.1, 85.5 and 84.5 for 0.1, 0.5 and 1 mm thickness specimens, respectively, as shown in Table 1.

Fig. 1.
figure 1

Evolution of weight over time for square specimens of different cellulose acetate-lignin composition. (A) Mass loss over 4 h of 0% lignin-cellulose acetate specimens. (B) Mass loss over 4 h of 33.3% lignin-cellulose acetate specimens. (C) Mass loss over 4 h of 50% lignin-cellulose acetate specimens. (D) Mass loss over 4 h of 66.7% lignin-cellulose acetate specimens. (E) Evolution over a period of 20 h of 4 specimens (1 mm thick) with the following initial lignin percentage, from left to right: 0, 33.3, 50, 66.7%.

Full size image
Table 1. Initial and final mean values of mass and porosity before and after leaching for specimens with 3 different thicknesses (0.1, 0.5, 1 mm), and 4 different percentages in weight of lignin: 0, 33.3, 50, 66.6%, (0-lig, 33-lig, 50-lig, 66-lig, respectively. N = 5 for each value of mass and porosity)
Full size table

3.2 From Natural Flying Seeds to Artificial Flying Seeds

The morphometric analysis performed on n. 8 natural Ailanthus altissima seeds led to dimensions (Fig. 2a), that were used for the design and fabrication of the artificial seed. The longitudinal length (LL) was 51.3 ± 1.6 mm, the transversal length (LT) was 9.0 ± 0.7 mm, the wing thickness (Thw) was 0.18 ± 0.02 mm, the seed capsule thickness (Ths) was 1.7 ± 0.1 and its diameter (D) was 5.8 ± 0.2 The mean mass (M) measured for the natural seeds was 22.6 ± 2.2 mg. The mean of the measured wing surfaces (S) resulted in 319 ± 17 mm2 and led to a wing loading (W/S) of 0.69 ± 0.08 N/m2.

The aerodynamic analysis was intended to establish the descent speed of Ailanthus altissima seeds derived from the spirally twisted flight mode because this was the most frequent descent mode. The mean descent speed was 0.64 ± 0.12 m/s. Other flight modes were observed in addition to twisted spiral: tumbling, simple rotation and dead weight.

The morphometric and aerodynamic data acquired from Ailanthus altissima seed were used to design an artificial seed that could have the same characteristics. The design and fabrication workflow is shown in Fig. 2e. The vector file of the contour of a photographed model seed was slightly modified in the 3D CAD software to fit the mean dimensions extracted by the morphometric analysis of the natural seeds. The curls of the wing were not realized. A number of 8 artificial Ailanthus altissima seeds were printed using the 4 cellulose acetate-lignin solutions previously tested and their weights were measured. The 50-lig composite was the chosen material for the development of the final artificial seed. That is because the mass of the seed printed with it was nearly double the mass of the natural seed, and upon lignin removal in water half of the mass was lost, giving a final mass similar to that of the natural seed. The printed Ailanthus altissima seeds resulted in a mean mass of 43.0 ± 1.8 mg. The wing surface (S) was 320 ± 11 mm2, right as the wing surface of the natural seed, while the wing loading (W/S) resulted in 1.32 ± 0.08 N/m2, a doubled value, as expected. Porosity was calculated to be 33.5%.

The aerodynamics of n. 8 50-lig Ailanthus altissima seeds was studied. The main occurring flight mode was the twisted spiral, as in the natural seed. The measured mean descent speed (vd) was 1.07 ± 0.11 m/s, which is 66% more than the natural samara seeds.

The same seed was tested after lignin removal in water and drying in oven. The resulting mass (M) was 22.4 ± 1.1 mg, showing a loss in weight of nearly half (48%), and it was identical to the mass of the natural Ailanthus altissima seed. The mean wing surface (S) did not change (320 ± 11 mm2), while the mean wing loading (W/S) became 0.69 ± 0.04 N/m2, in quite good agreement with the natural seed. Porosity of the artificial seeds was greatly enhanced, reaching the value of 65.4%. Porosity can be viewed under an optical microscope as shown in Figs. 2f and 2g, where a natural and an artificial porous wing are compared.

The mean descent speed (vd) of the twisted spiral flight mode of the porous artificial samara seeds was measured to be 0.64 ± 0.03 m/s, the same as the natural. A comparison of the mean morphometric and aerodynamic data between the 3 seeds (natural, 50-lig artificial, porous artificial), can be found in Table 2.

Fig. 2.
figure 2

Development of the artificial seed from the natural Ailanthus altissima seed. a) Dimensions of the natural seed. b) Dimensions of the artificial porous seed. c) Image binarization of the surface of the natural seed for the calculation of wing surface and wing loading. d) Image binarization of the surface of the artificial seed for the calculation of wing surface and wing loading. e) Workflow of the development of the artificial Ailanthus altissima seed. f) Micrograph of the wing of the natural seed and g) artificial seed (scalebar 1000 µm).

Full size image
Table 2. Comparison of morphometric and aerodynamic parameters for natural, lignin filled artificial and porous artificial Ailanthus altissima seeds.
Full size table

3.3 Humidity Sensing with CNC Optical Crystals

Figure 3a shows the coupling of the CNC photonic crystal (square 4 × 4 mm, weight 7 mg) with the porous cellulose acetate based artificial Ailanthus altissima seed.

Fig. 3.
figure 3

Colorimetric humidity sensing characterization. a) Picture of the porous artificial Ailanthus altissima seed with the square (4 × 4 mm) CNC photonic crystal humidity sensor over relative humidity variations (30, 60 and 90%). b) Reflectance spectrum of the CNC photonic crystal humidity sensor at 30, 60 and 90% relative humidity. c) Calibration %RH vs wavelength (nm) relative to the maximum reflectance values recorded in (b). d) Spectral data from b) represented in a CIE 1931 diagram of the color perceived by the human eye. e) Colorimetric analysis of the CNC photonic crystal humidity sensor pictures (using ImageJ software) in the blue channel over relative humidity variations (30, 60 and 90%). f) Calibration %RH vs px value relative to maximum intensity values recorded in (e).

Full size image

Figure 3a shows a sequence of pictures of the CNC film exposed to different humidity conditions (30, 60 and 90%) proving structural color changes in the visible spectrum. At %RH of 30% the CNC showed a yellow color. As %RH increased to 60% the CNC changed from yellow to red; with a further increase at 90% the CNC became dark red/transparent.

These data are in good agreement with the reflectance spectrum reported in Fig. 3b. The structural color changes from yellow to dark red/transparent with the increase of humidity is owing to the swelling of the multilayer structure caused by water adsorption.

Figure 3c reports the calibration of the sensor (i.e., %RH vs wavelength relative to maximum reflectance of the photonic bandgap). Data from the spectra reported in Fig. 3b are shown superimposed on a CIE 1931 diagram in Fig. 3d, proving the possibility of being colorimetrically discriminated.

The CNC colors at %RH 30, 60 and 90% extracted from Fig. 3a were also analyzed plotting the histogram in the blue channel using ImageJ software [21] (Fig. 3e). Figure 3f reports the colorimetric calibration of the sensor (i.e., %RH vs pixels value relative to the maximum intensity).

In summary, the CNC photonic crystal sensor coupled with the artificial Ailanthus altissima seed showed promising applicability for colorimetric humidity environmental monitoring using both spectrometer analysis and colorimetric image processing techniques. In perspective, statistical validation and an on-field measurement of the humidity using the colorimetric calibration will be carried out.

4 Conclusions

The distributed monitoring of environmental parameters poses even more challenges in the technologies at the base of the sensors employed. We developed a porous, biodegradable and 3D printable material for the fabrication of biomimetic artificial fliers and we coupled it with a biodegradable optical sensor based on CNC nanocrystals.

The porosity measured with different compositions of cellulose acetate and lignin varied from a minimum of ~30% to a maximum of ~90%, indicating a wide range of porosity achievable. The choice of the right material composition and use of 3D printing was important to tailor the porosity of the artificial seed. The designing of the artificial seed was possible thanks to a thorough analysis of the morphometrics and aerodynamics of the natural Ailanthus altissima seed. This biomimetic approach to design, coupled with the use of 3D printing technologies and leaching technique, constituted an efficient process for the creation of artificial seeds with same morphometries and aerodynamic behavior as the natural model (same mass (M) ~22.4 ± 1.1 mg and same descent speed (vd) ~0.64 ± 0.03 m/s). The characterization of the CNC sensor indicated a reliable readout of relative humidity (30, 60, 90%RH) in controlled conditions.

Future work will include more morphological analysis, i.e., histology, to study and characterize the porosity of the natural Ailanthus altissima seed, and more aerodynamic parameters, such as seed rotations and drift. Outdoor experiments will be done to assess the dispersal abilities of natural and artificial seeds with different wind conditions and to measure humidity in real conditions.