*3.4. Banana Ripeness Detection Applications*

The fruit ripeness detection experiments were then designed and conducted in the laboratory to verify the application feasibility of our as-fabricated ethylene sensors by comparison with traditional GC-MS technology. In our experiments, the bananas, a typical climacteric fruit, were adopted, and it had been demonstrated that ethylene was released during the ripeness process of bananas. The level of banana ripeness can be commonly divided into four stages: unripe, slightly ripe, ripe, and overripe [30], accompanying the gradual change of banana color and released ethylene concentration. The unripe bananas exhibit solid light green color with some light greenish-yellow, and show no significant aroma. Mostly yellow with very faint green at tips and along edges is defined as slightly ripe, which usually exhibits some faint banana aroma. After that, bananas step into the ripe

stage, and the color become solid yellow with no green, but sporadic small brown spots. At this stage, the released aroma is the strongest. Afterwards, the bananas enter into the overripe stage with the color of dark brown and even black, during which the released aroma gradually drops into a relatively low level. According to the characteristics of the above four ripeness stages, the color changes of the obtained banana samples could be clearly observed (Figure 5a), and the bananas at different ripeness stages were placed into different contained. The containers were sealed for collection of ethylene, and after about 20 min, the sensors were placed into the containers for ethylene sensing toward banana ripeness detection. The sensing response to bananas at different ripeness stages is shown in Figure 5b, from which it could be clearly seen that the ethylene concentration increases gradually until the bananas reached the ripe stage. The sensing response is calculated to be 16.8%, 25.4%, 33.8%, and 30.2%, corresponding to the unripe, slightly ripe, ripe, and overripe bananas, respectively. The normalized peak area of the released ethylene measured by GC-MS exhibits the same changing trend with the banana ripeness stage i.e., the ethylene concentration increases until the bananas are ripe, and further declines when they become overripe, as shown in Figure 5c.

**Figure 5.** (**a**) Photos of banana samples at different storage stages: yellowish green (1st stage, unripe), all yellow (2nd stage, slightly ripe), all yellow with brown speckles (3rd stage, ripe), and dark brown (4th stage, overripe). (**b**) The real-time resistance versus time curves of the rGO/WSe2/Pd heterojunction-based chemiresistive sensor when placed into the sealed container with bananas at different ripeness stages. (**c**) The normalized peak area of released ethylene from banana samples at different ripeness stages by GC-MS technology.

## **4. Discussion**

According to the calculated adsorption energies, all the negative values manifested the advantage of room temperature operation of the fabricated ternary sensitive materials system with energy-level alignment, which may be caused by the formed chemical bond, further changing the energy band structure. The calculated adsorption energies were consistent with the sensing response of each adsorbed system to ethylene. Previous studies show that rGO is p-type materials. When ethylene molecules interact with p-type rGO sensitive films, electrons will transfer from rGO to ethylene, owing to higher Fermi levels of rGO than ethylene, resulting in a decrease of electron density of the rGO films and thus a negative sensing response to ethylene gas. When composited with Pd NPs, the sensing response to ethylene gas becomes more negative in rGO/Pd composites, which is inconsistent with the positive sensing response arising from the catalytic effects of Pd NPs toward ethylene molecules and thus excludes the role of catalytic properties on sensing response enhancements. This ethylene sensing enhancement could be attributed to the local doping effects of Pd NPs on rGO when the device is exposed to ethylene [31]. The large work function difference between high work function Pd NPs and rGO results in the local hole-doping of the rGO at the Pd NPs/rGO interfaces and thus lowers the Fermi energy levels of the rGO/Pd composites. Combined with the higher adsorption energy, both contributed to more electrons transferred from rGO/Pd to ethylene, and leads to an enhanced negative sensing response to ethylene.

In the case of rGO/WSe2 bilayer heterojunctions, the device resistance is dominated by conductive rGO films due to the bilayer device structure deposited on IDEs and much higher electrical conductivity of rGO than WSe2. When rGO and WSe2 are brought into contact, it is expected that electrons would pass from rGO to WSe2 until the equilibrium of the Fermi level is achieved. As a result, a Schottky-type junction is formed across the rGO/WSe2 interface with a downward band bending and a hole depletion region in WSe2 near the surface [32,33]. When the rGO/WSe2 heterojunction is exposed to ethylene, the electron transfer from WSe2 to the adsorbed ethylene molecules would lead to an increase of hole concentration in WSe2. The increased hole concentration in WSe2 causes a larger Fermi level difference between rGO and WSe2, leading to more holes transferred from WSe2 to rGO, and thus significantly increased sensitivity to ethylene. Furthermore, when Pd NPs are introduced to the rGO/WSe2 bilayer film, the Schottky junction effects are further enhanced by the hole doping effects in WSe2 induced by Pd NPs [34]. Combined with more negative adsorption energy of the rGO/WSe2/Pd composites, more holes will transfer from WSe2 to rGO, resulting in a further increased sensitivity to ethylene. Moreover, the unique structures of 2D materials and the self-assembled sensitive films could increase the adsorption sites for gas molecules, accelerate the charge transfer between rGO and WSe2 nanosheets, and promote the synergistic effects between them.

For climacteric fruit such as the banana, when it changes from ripe to overripe, the released ethylene concentration decreases. This indicates a direct relationship between released ethylene concentration and the banana ripeness stage, which still needs further quantitative research. Moreover, the released aroma of climacteric fruits has a positive correlation with the released inner ethylene during the storage process [3]. Therefore, intelligent ethylene sensor techniques cannot only reflect the ripeness information of climacteric fruits, but also can be potentially adopted to evaluate their flavor quality. The consistent results from our fabricated ethylene sensor and GC-MS confirm the reliability and feasibility of our proposed ethylene sensor methodology toward banana ripeness detection. Furthermore, the proposed ethylene sensor exhibits distinct advantages of low costs, fast and accurate detection, simple operation, and in situ monitoring. The ethylene sensors can also further be developed into electronic noses and integrated into picking robots, grading equipment, packaged boxes, and shelves, which could push forward the development of intelligent agriculture.
