**Design and Fabrication of Flexible Naked-Eye 3D Display Film Element Based on Microstructure**

#### **Axiu Cao , Li Xue, Yingfei Pang, Liwei Liu, Hui Pang, Lifang Shi \* and Qiling Deng**

Institute of Optics and Electronics, Chinese Academy of Sciences, Chengdu 610209, China; longazure@163.com (A.C.); xueli2553@163.com (L.X.); yfpang7647@163.com (Y.P.); liuliweineko@163.com (L.L.); ph@ioe.ac.cn (H.P.); dengqiling@ioe.ac.cn (Q.D.)

**\*** Correspondence: shilifang@ioe.ac.cn; Tel.: +86-028-8510-1178

Received: 19 November 2019; Accepted: 7 December 2019; Published: 9 December 2019

**Abstract:** The naked-eye three-dimensional (3D) display technology without wearing equipment is an inevitable future development trend. In this paper, the design and fabrication of a flexible naked-eye 3D display film element based on a microstructure have been proposed to achieve a high-resolution 3D display effect. The film element consists of two sets of key microstructures, namely, a microimage array (MIA) and microlens array (MLA). By establishing the basic structural model, the matching relationship between the two groups of microstructures has been studied. Based on 3D graphics software, a 3D object information acquisition model has been proposed to achieve a high-resolution MIA from different viewpoints, recording without crosstalk. In addition, lithography technology has been used to realize the fabrications of the MLA and MIA. Based on nanoimprint technology, a complete integration technology on a flexible film substrate has been formed. Finally, a flexible 3D display film element has been fabricated, which has a light weight and can be curled.

**Keywords:** naked-eye 3D; microstructure; flexible; film; fabrication

#### **1. Introduction**

With the development of science and technology, people are hoping to truly restore three-dimensional (3D) information of the object space. As a result, 3D display technology has emerged with the times and has become a research hotspot in the field of image displays [1–6].

As Wheatstone invented the first stereoscopic picture viewer in 1838, the technology of 3D displays has been developed for nearly 200 years [7]. In this development process, head-mounted 3D display technology is very mature in principle and technology [8–11], and there are a large number of commercial products. However, due to the need of wearing equipment, it is always inconvenient. Meanwhile, long-term use depending on the binocular parallax principle will lead to viewing fatigue, and viewers will feel dizzy. Therefore, it is an inevitable trend for the future development of naked-eye 3D display technology without wearing equipment.

Research groups have developed a variety of naked-eye 3D display technologies, including the grating 3D display technology of binocular parallax, holographic technology, and integrated imaging technology. Grating 3D display technology uses the principle of binocular parallax to produce a 3D sensation [12,13]. This technology has the advantages of low cost, simple structure, and easy implementation. However, because the left and right parallax images cannot be completely separated, the viewing area of the viewer is limited, and the 3D image can only be viewed in a relatively fixed position, lacking freedom. Therefore, it is only suitable for a single user and small range of motion.

By using the interference principle, holographic technology interferes the light wave reflected by the object with the reference light wave, and records it in the form of interference fringes to form a hologram [14]. When the hologram is illuminated by a coherent light source, the original light wave will be reproduced based on the diffraction principle, to form a realistic 3D image of the original object. However, the production of high-quality optical holograms requires a high-coherence laser, shockproof platform, and precise optical path setting. In addition, the ambient air flow will also affect the successful recording of holograms. Later, with the development of digital holography technology, using charge coupled devices (CCDs) instead of ordinary holographic recording materials to record holograms and using computer simulations instead of optical diffraction to realize object reproduction, the digitization of the whole process of hologram recording, storage, processing, and reproduction can be realized [15]. However, the resolution of digital holography using CCDs to record coherent light waves cannot be compared to that of holographic dry plates, so the resolution of holograms is relatively low, which seriously affects the image clarity.

The computer-generated hologram, which combines digital computing with modern optics, encoding the complex amplitude of the object light wave from a computer to computer-generated hologram (CGH), has unique advantages and good flexibility [16]. However, there is a large amount of data information and processing time for 3D objects, so it is necessary to select appropriate algorithms and coding methods to overcome [5,17]. In addition, the spatial light modulator plays a very important role in the experiment of CGH photoelectricity reproduction. It is necessary to overcome the influence of spatial light modulation on the quality of the reconstructed image [18,19]. At present, the CGH is still in the research stage of algorithm optimization to realize a 3D display of large scale and large field of view. It is still early to expect a commercial product based on holographic display devices.

Integrated imaging technology is also composed of two basic processes: Recording and reproduction. Unlike holographic technology, the recording and reproduction process do not require the participation of coherent light, which reduces the difficulty of the whole system. This technology was first proposed by Lippmann, a famous French physicist and Nobel Laureate in physics, in 1908 [20]. The microlens units in the microlens array (MLA) are used to record 3D object information from different perspectives to form a microimage array (MIA), and then the recorded MIA is placed on the focal plane of the MLA whose parameters are matched with a microlens in the recording process. The 3D image can be viewed by irradiating with scattered light according to the principle of optical reversibility and the fusion of the human brain. This technology can provide full parallax and a full-color image, without any special equipment, and the viewpoint provided is quasi-continuous. In a certain area, it can be viewed by many people, which has become an important development trend of naked-eye 3D displays.

The recording and reconstruction of 3D objects are formed through the interaction of tens of thousands or even hundreds of thousands of microimages. A camera array can be used to record the MIA. This method requires a large number of expensive and complex camera equipment, and the mechanical error between camera equipment will also affect the final imaging effect [21,22]. The optical recording method [23] uses an MLA to record the MIA, which is easy to be affected by surrounding environmental conditions such as brightness, sensitivity, and uniformity. The experimental operation is difficult, and the adjacent images are prone to crosstalk, resulting in a poor imaging effect of the final MIA. Then, the acquisition of the MIA based on 3D graphics software is proposed [24], which can realize the free crosstalk and high-resolution recording of the MIA. In addition, the display screen is generally used to display the recorded microimage array in 3D image reproduction. At present, the screen resolution of the mainstream high-definition display screen is 1920 × 1080, and the number of pixels per inch is 89, i.e., 89 ppi. Compared to the resolution limit of the human eye of 300 ppi, the resolution is still low. The granular distortion effect will be visible at a certain distance.

In this paper, based on the integrated imaging technology, a flexible naked-eye 3D display film element based on microstructures has been designed and fabricated, which can achieve the 3D imaging effect with a resolution higher than the human eye resolution limit of 300 ppi, and has the advantages of curling and light weight. The main arrangement of this paper is as follows: The second part describes the structure design and imaging principle of the 3D display film element. The third part presents the

structural design of the MLA and acquisition of the MIA. The fourth part describes the preparation and integration of the microstructure, and the final part presents the summary of the whole paper. the preparation and integration of the microstructure, and the final part presents the summary of the whole paper.

*Micromachines* **2019**, *10*, x 3 of 9

#### **2. Structural Design and Imaging Principle 2. Structural Design and Imaging Principle**

The structure of the flexible naked-eye 3D display film element consists of three parts, namely, an MLA, MIA, and flexible substrate material, as shown in Figure 1a. The MIA is imaged by the MLA with different viewing angle information. The sub-images of each imaging channel are fused to form a 3D display effect. The human eye can watch the 3D image of the object in front of them, as shown in Figure 1b. The structure of the flexible naked-eye 3D display film element consists of three parts, namely, an MLA, MIA, and flexible substrate material, as shown in Figure 1a. The MIA is imaged by the MLA with different viewing angle information. The sub-images of each imaging channel are fused to form a 3D display effect. The human eye can watch the 3D image of the object in front of them, as shown in Figure 1b.

**Figure 1.** Flexible naked-eye 3D display film element: (**a**) structure composition; (**b**) imaging principle; (**c**) viewing angle. **Figure 1.** Flexible naked-eye 3D display film element: (**a**) structure composition; (**b**) imaging principle; (**c**) viewing angle.

The aperture (*D*) of the microlens is the same as the size (*T*) of the microimage unit, and the distance (*d*) between the MLA and MIA is the effective focal length (*f*) of the microlens. A single microlens can image the corresponding microimage independently, and several sub-images are fused to form a 3D effect. According to the theory of Gauss optics, when the distance between the MLA and MIA is the focal length of the microlens, the image distance is infinite, which means that light from any angle on the MIA is refracted by the MLA and then emitted as parallel light. Therefore, the number of pixels of the reconstructed 3D image is determined by the number of MLAs (*n* × *n*). Finally, the imaging resolution (Re) of the element can be calculated according to Equation (1), where *L* is the size of the MIA. *L* can be obtained by multiplying the array number (*n* × *n*) of microimages by the size of the microimage unit (*T*). The viewing angle of the film element (Figure 1c) can be obtained from Equation (2). As the focus display mode [25] is used in this paper, the 3D depth range (Δ*Z*) is determined by Equation (3), where *P<sup>I</sup>* is the pixel size of the microimage. The aperture (*D*) of the microlens is the same as the size (*T*) of the microimage unit, and the distance (*d*) between the MLA and MIA is the effective focal length (*f*) of the microlens. A single microlens can image the corresponding microimage independently, and several sub-images are fused to form a 3D effect. According to the theory of Gauss optics, when the distance between the MLA and MIA is the focal length of the microlens, the image distance is infinite, which means that light from any angle on the MIA is refracted by the MLA and then emitted as parallel light. Therefore, the number ofpixels of the reconstructed 3D image is determined by the number of MLAs (*<sup>n</sup>* <sup>×</sup> *<sup>n</sup>*). Finally, the imaging resolution (Re) of the element can be calculated according to Equation (1), where *L* is the size of the MIA. *L* can be obtained by multiplying the array number (*n* × *n*) of microimages by the size of the microimage unit (*T*). The viewing angle of the film element (Figure 1c) can be obtained from Equation (2). As the focus display mode [25] is used in this paper, the 3D depth range (∆*Z*) is determined by Equation (3), where *P<sup>I</sup>* is the pixel size of the microimage.

$$\text{Re} = \frac{n}{L'} \tag{1}$$

$$\theta = 2 \arctan(\frac{T}{2d}),$$
 
$$\dots$$

$$
\Delta Z = 2\frac{dD}{P\_I}.\tag{3}
$$

**3. Structural Design of Microlens Array (MLA) and Acquisition of Microimage Array (MIA)**

*3.1. Structural Design of MLA*

### **3. Structural Design of Microlens Array (MLA) and Acquisition of Microimage Array (MIA)**

#### *3.1. Structural Design of MLA*

Due to the MLA being the key imaging element, the reasonable design of the parameters of the MLA, such as the aperture, focal length, and array number, is related to the integration of the whole element and the quality of the 3D image. For a miniaturized 3D film element, when the microlens is designed with a large aperture or a small number of MIAs, the resolution of images from all perspectives of the 3D image will be very low, which makes the viewing effect worse. In order to meet the requirement of the human eye resolution of 300 ppi, the structural parameters of the microlens are designed as follows: (1) the material of the microlens is a photosensitive adhesive (NOA61), and the refractive index is 1.56 in the visible light band; (2) the aperture (*D*) of the microlens is 80 µm; (3) the curvature radius of the microlens is 47.5 µm; (4) the focal length (*f*) of the microlens is 85 µm; (5) the sag height of the microlens is 22 µm; (6) the array number of the microlens is 250 × 250. Then, the imaging resolution is calculated as 317 ppi, which is higher than the human-eye resolution limit. The viewing angle (θ) is about 50◦ . At this time, the focal length of the microlens is very short, so the distance between the MLA and MIA is 85 µm. Thus, the whole element will reach the thin-film level.

#### *3.2. Acquisition of MIA*

The acquisition of tens of thousands or even hundreds of thousands of microimages has been carried out based on 3D graphics software. This method does not need complicated and expensive optical equipment, and can also avoid human and mechanical errors in the operation of optical equipment. With the use of computer memory, computer generation technology can generate microimages quickly, accurately, and in large quantities.

Based on 3ds MAX software (3ds MAX 2009, San Rafael, CA, USA), a 3D scene was established, as shown in Figure 2. The scene contained the Chinese characters "光电所" and letters "IOE." The central 3D coordinates of "光", "电", and "所" were (15.59 mm, −6.8 mm, 9.5 mm), (30.79 mm, 0, 9.5 mm), and (37.88 mm, 6.8 mm, 9.5 mm), respectively. The central 3D coordinates of "I", "O", and "E" were (37.35 mm, −6 mm, 13 mm), (23.43 mm, 0, 13 mm), and (12.48 mm, 6 mm, 13 mm), respectively. The virtual dynamic camera array was established to simulate the image acquisition process of the whole MLA, and the acquisition of the microimage was carried out for different 3D information from far to near. In the image acquisition, the 3D coordinates of the start point of the camera were A (0, −12.5 mm, 0), and the 3D coordinates of the end point were B (0, 7.42 mm, 19.92 mm). The field of view of the camera was 5◦ and the moving interval was 80 µm, which was matched with the structural parameters of the MLA. Finally, 250 × 250 microimages were acquired. The pixel number of the microimage was 40 × 40, and the pixel size of the microimage (*P<sup>I</sup>* ) was 2 µm. Therefore, the 3D depth range could be calculated as 6.8 mm.

In the process of MIA acquisition, the camera captured images of the 3D scene from different perspectives, as shown in Figure 3, to record the information at different perspectives. In addition, the obtained microimages are shown in the box in the upper right corner of the corresponding perspective. It can be seen that the microimages captured by the virtual camera have a very high image resolution and perfect image quality.

the thin-film level.

*3.2. Acquisition of MIA*

microimages quickly, accurately, and in large quantities.

Therefore, the 3D depth range could be calculated as 6.8 mm.

image resolution and perfect image quality.

Due to the MLA being the key imaging element, the reasonable design of the parameters of the MLA, such as the aperture, focal length, and array number, is related to the integration of the whole element and the quality of the 3D image. For a miniaturized 3D film element, when the microlens is designed with a large aperture or a small number of MIAs, the resolution of images from all perspectives of the 3D image will be very low, which makes the viewing effect worse. In order to meet the requirement of the human eye resolution of 300 ppi, the structural parameters of the microlens are designed as follows: (1) the material of the microlens is a photosensitive adhesive (NOA61), and the refractive index is 1.56 in the visible light band; (2) the aperture (*D*) of the microlens is 80 μm; (3) the curvature radius of the microlens is 47.5 μm; (4) the focal length (*f*) of the microlens is 85 μm; (5) the sag height of the microlens is 22 μm; (6) the array number of the microlens is 250 × 250. Then, the imaging resolution is calculated as 317 ppi, which is higher than the human-eye resolution limit. The viewing angle (*θ*) is about 50°. At this time, the focal length of the microlens is very short, so the distance between the MLA and MIA is 85 μm. Thus, the whole element will reach

The acquisition of tens of thousands or even hundreds of thousands of microimages has been carried out based on 3D graphics software. This method does not need complicated and expensive optical equipment, and can also avoid human and mechanical errors in the operation of optical equipment. With the use of computer memory, computer generation technology can generate

Based on 3ds MAX software (3ds MAX 2009, San Rafael, CA, USA), a 3D scene was established, as shown in Figure 2. The scene contained the Chinese characters "光电所" and letters "IOE." The central 3D coordinates of "光", "电", and "所" were (15.59 mm, −6.8 mm, 9.5 mm), (30.79 mm, 0, 9.5 mm), and (37.88 mm, 6.8 mm, 9.5 mm), respectively. The central 3D coordinates of "I", "O", and "E" were (37.35 mm, −6 mm, 13 mm), (23.43 mm, 0, 13 mm), and (12.48 mm, 6 mm, 13 mm), respectively. The virtual dynamic camera array was established to simulate the image acquisition process of the whole MLA, and the acquisition of the microimage was carried out for different 3D information from far to near. In the image acquisition, the 3D coordinates of the start point of the camera were A (0, −12.5 mm, 0), and the 3D coordinates of the end point were B (0, 7.42 mm, 19.92 mm). The field of view of the camera was 5° and the moving interval was 80 μm, which was

**Figure 2. Figure 2.** 250 × 250 microimages acquired by dynamic camera. 250 × 250 microimages acquired by dynamic camera. perspective. It can be seen that the microimages captured by the virtual camera have a very high image resolution and perfect image quality. perspective. It can be seen that the microimages captured by the virtual camera have a very high

**Figure 3.** Imaging from different perspectives: (a) imaging from one prespective of "所", (b) imaging from one prespective of "O", (c) imaging from one prespective of "电", and (d) imaging from one prespective of "I". **Figure 3.** Imaging from different perspectives: (a) imaging from one prespective of "所", (b) imaging from one prespective of "O", (c) imaging from one prespective of "电", and (d) imaging from one prespective of "I". **Figure 3.** Imaging from different perspectives: (a) imaging from one prespective of "所", (b) imaging from one prespective of "O", (c) imaging from one prespective of "电", and (d) imaging from one prespective of "I".

Furthermore, 250 × 250 microimages captured by the camera were encoded and fused according to the arrangement of the MLA. The MIA was generated by computer processing, as shown in Figure 4. From the enlarged images of different regions, we can see that each microimage is different with information of different perspectives from the 3D scene. Furthermore, 250 × 250 microimages captured by the camera were encoded and fused according to the arrangement of the MLA. The MIA was generated by computer processing, as shown in Figure 4. From the enlarged images of different regions, we can see that each microimage is different with information of different perspectives from the 3D scene. Furthermore, 250 × 250 microimages captured by the camera were encoded and fused according to the arrangement of the MLA. The MIA was generated by computer processing, as shown in Figure 4. From the enlarged images of different regions, we can see that each microimage is different with information of different perspectives from the 3D scene.

**Figure 4.** Microimage array (MIA) with different enlarged images of different regions. **Figure 4.** Microimage array (MIA) with different enlarged images of different regions. characteristics of high toughness, smooth surface, and good light transmittance. The pattern was **Figure 4.** Microimage array (MIA) with different enlarged images of different regions.

There are two key microstructures in the 3D display element, which are the MLA and MIA. The MLA was prepared by photolithography and the hot melting method, and the preparation results are shown in Figure 5, which shows the prototype of the MLA (Figure 5a), micromagnifier of the microlens (Figure 5b), and surface profile of the microlens (Figure 5c). The sag height of the microlens was 21.97 μm, which is consistent with the design result. The MIA was also prepared by photolithography. The pattern was prepared on the substrate material by a series of processes such as exposure, development, and etching. The MIA and enlarged images of different regions of the MIA are shown in Figure 6. The substrate material was a polyethylene terephthalate (PET) film, which has characteristics of high toughness, smooth surface, and good light transmittance. The pattern was

There are two key microstructures in the 3D display element, which are the MLA and MIA. The MLA was prepared by photolithography and the hot melting method, and the preparation results are shown in Figure 5, which shows the prototype of the MLA (Figure 5a), micromagnifier of the microlens (Figure 5b), and surface profile of the microlens (Figure 5c). The sag height of the microlens was 21.97 μm, which is consistent with the design result. The MIA was also prepared by photolithography. The pattern was prepared on the substrate material by a series of processes such as exposure, development, and etching. The MIA and enlarged images of different regions of the MIA are shown in Figure 6. The substrate material was a polyethylene terephthalate (PET) film, which has

**4. Preparation and Integration**

**4. Preparation and Integration**

#### **4. Preparation and Integration**

There are two key microstructures in the 3D display element, which are the MLA and MIA. The MLA was prepared by photolithography and the hot melting method, and the preparation results are shown in Figure 5, which shows the prototype of the MLA (Figure 5a), micromagnifier of the microlens (Figure 5b), and surface profile of the microlens (Figure 5c). The sag height of the microlens was 21.97 µm, which is consistent with the design result. The MIA was also prepared by photolithography. The pattern was prepared on the substrate material by a series of processes such as exposure, development, and etching. The MIA and enlarged images of different regions of the MIA are shown in Figure 6. The substrate material was a polyethylene terephthalate (PET) film, which has characteristics of high toughness, smooth surface, and good light transmittance. The pattern was made by lithography with a high resolution of minimum linewidth of 2 µm, which is equal to the pixel size of the microimage. *Micromachines* **2019**, *10*, x 6 of 9 made by lithography with a high resolution of minimum linewidth of 2 μm, which is equal to the pixel size of the microimage. *Micromachines* **2019**, *10*, x 6of 9 made by lithography with a high resolution of minimum linewidth of 2 μm, which is equal to the pixel size of the microimage.

**Figure 5.** Preparation results of microlens: (**a**) prototype of microlens array (MLA); (**b**) micromagnifier of microlens; (**c**) surface profile of microlens. **Figure 5.** Preparation results of microlens: (**a**) prototype of microlens array (MLA); (**b**) micromagnifier of microlens; (**c**) surface profile of microlens. **Figure 5.** Preparation results of microlens: (**a**) prototype of microlens array (MLA); (**b**) micromagnifier of microlens; (**c**) surface profile of microlens.

**Figure 6.** Preparation results of (**a**) MIA in different areas: (**b**) few microimages of "I", (**c**) few microimages of "O", and (**d**) few microimages of "E". **Figure 6.** Preparation results of (**a**) MIA in different areas: (**b**) few microimages of "I", (**c**) few microimages of "O", and (**d**) few microimages of "E". **Figure 6.** Preparation results of (**a**) MIA in different areas: (**b**) few microimages of "I", (**c**) few microimages of "O", and (**d**) few microimages of "E". **Figure 6.** Preparation results of (**a**) MIA in different areas: (**b**) few microimages of "I", (**c**) few

microimages of "O", and (**d**) few microimages of "E".

Subsequently, integration of the two key microstructures needs to be carried out. During the integration, the MLA and MIA need to be aligned one by one to realize 3D image reconstruction. In the experiment, nano-imprinting alignment technology was used to achieve alignment integration of the two microstructures, as shown in Figure 7. Subsequently, integration of the two key microstructures needs to be carried out. During the integration, the MLA and MIA need to be aligned one by one to realize 3D image reconstruction. In the experiment, nano-imprinting alignment technology was used to achieve alignment integration of the two microstructures, as shown in Figure 7. Subsequently, integration of the two key microstructures needs to be carried out. During the integration, the MLA and MIA need to be aligned one by one to realize 3D image reconstruction. In the experiment, nano-imprinting alignment technology was used to achieve alignment integration of the two microstructures, as shown in Figure 7. Subsequently, integration of the two key microstructures needs to be carried out. During the integration, the MLA and MIA need to be aligned one by one to realize 3D image reconstruction. In the experiment, nano-imprinting alignment technology was used to achieve alignment integration of the two microstructures, as shown in Figure 7.

**Figure 7.** Integration of thin film element: (**a**) pouring of polydimethylsiloxane (PDMS) materials; (**b**) **Figure 7.** Integration of thin film element: (**a**) pouring of polydimethylsiloxane (PDMS) materials; (**b**) **Figure 7.** Integration of thin film element: (**a**) pouring of polydimethylsiloxane (PDMS) materials; (**b**) baking and curing; (**c**) generation of imprinting mold; (**d**) photosensitive adhesive (NOA61) dropped on the prepared MIA; (**e**) leveling; (**f**) imprinting and UV curing; (**g**) mold peeled off. **Figure 7.** Integration of thin film element: (**a**) pouring of polydimethylsiloxane (PDMS) materials; (**b**) baking and curing; (**c**) generation of imprinting mold; (**d**) photosensitive adhesive (NOA61) dropped on the prepared MIA; (**e**) leveling; (**f**) imprinting and UV curing; (**g**) mold peeled off.

the mold is used to carry out the integration. During the mold preparation, PDMS stroma and curing agent were poured into a clean beaker at a volume ratio of 10:1, continuously stirring with a glass rod. A large number of bubbles were generated in the PDMS prepolymer until the bubbles disappeared gradually. In addition, the PDMS prepolymer was poured on the prepared MLA, as shown in Figure 7a. Then, the substrate was placed on the coater at a speed of 250 rpm with a time of 20 s, shaking off the excess PDMS. Subsequently, it was placed in a vacuum oven until all the bubbles disappeared for curing, as shown in Figure 7b. The baking temperature and time were set as 65 °C

First, the imprinting mold with the structural information of the MLA should be prepared. The imprinting mold is composed of polydimethylsiloxane (PDMS). The free energy of the interface of the PDMS mold is low and has chemical inertness. Therefore, the mold is easy to be separated when the mold is used to carry out the integration. During the mold preparation, PDMS stroma and curing agent were poured into a clean beaker at a volume ratio of 10:1, continuously stirring with a glass rod. A large number of bubbles were generated in the PDMS prepolymer until the bubbles disappeared gradually. In addition, the PDMS prepolymer was poured on the prepared MLA, as shown in Figure 7a. Then, the substrate was placed on the coater at a speed of 250 rpm with a time of 20 s, shaking off the excess PDMS. Subsequently, it was placed in a vacuum oven until all the bubbles disappeared for curing, as shown in Figure 7b. The baking temperature and time were set as 65 °C

First, the imprinting mold with the structural information of the MLA should be prepared. The imprinting mold is composed of polydimethylsiloxane (PDMS). The free energy of the interface of the PDMS mold is low and has chemical inertness. Therefore, the mold is easy to be separated when the mold is used to carry out the integration. During the mold preparation, PDMS stroma and curing agent were poured into a clean beaker at a volume ratio of 10:1, continuously stirring with a glass rod. A large number of bubbles were generated in the PDMS prepolymer until the bubbles disappeared gradually. In addition, the PDMS prepolymer was poured on the prepared MLA, as shown in Figure 7a. Then, the substrate was placed on the coater at a speed of 250 rpm with a time of 20 s, shaking off the excess PDMS. Subsequently, it was placed in a vacuum oven until all the bubbles disappeared for curing, as shown in Figure 7b. The baking temperature and time were set as 65 °C

First, the imprinting mold with the structural information of the MLA should be prepared. The

as shown in Figure 8.

viewing angles.

First, the imprinting mold with the structural information of the MLA should be prepared. The imprinting mold is composed of polydimethylsiloxane (PDMS). The free energy of the interface of the PDMS mold is low and has chemical inertness. Therefore, the mold is easy to be separated when the mold is used to carry out the integration. During the mold preparation, PDMS stroma and curing agent were poured into a clean beaker at a volume ratio of 10:1, continuously stirring with a glass rod. A large number of bubbles were generated in the PDMS prepolymer until the bubbles disappeared gradually. In addition, the PDMS prepolymer was poured on the prepared MLA, as shown in Figure 7a. Then, the substrate was placed on the coater at a speed of 250 rpm with a time of 20 s, shaking off the excess PDMS. Subsequently, it was placed in a vacuum oven until all the bubbles disappeared for curing, as shown in Figure 7b. The baking temperature and time were set as 65 ◦C and 4 h, respectively. Finally, the PDMS imprinting mold with negative structural information of the MLA on the surface could be peeled off from the MLA, as shown in Figure 7c. *Micromachines* **2019**, *10*, x 7 of 9 and 4 h, respectively. Finally, the PDMS imprinting mold with negative structural information of the MLA on the surface could be peeled off from the MLA, as shown in Figure 7c. Secondly, the structural information of the MLA should be imprinted on the surface of the MIA by using the imprinting mold to realize the integration of the MLA and MIA. The process flow was as follows: First, the photosensitive adhesive (NOA61) was dropped on the prepared MIA, as shown in Figure 7d. After it was leveled (Figure 7e), the imprinting mold was used to imprint photosensitive *Micromachines* **2019**, *10*, x 7 of 9 and 4 h, respectively. Finally, the PDMS imprinting mold with negative structural information of the

Secondly, the structural information of the MLA should be imprinted on the surface of the MIA by using the imprinting mold to realize the integration of the MLA and MIA. The process flow was as follows: adhesive. During imprinting, the high-precision alignment device was used to align the microlenses with the microimages one by one. On the basis of alignment, the photosensitive adhesive was irradiated by ultraviolet light with a wavelength of 365 nm until it was cured, as shown in Figure 7f. MLA on the surface could be peeled off from the MLA, as shown in Figure 7c. Secondly, the structural information of the MLA should be imprinted on the surface of the MIA by using the imprinting mold to realize the integration of the MLA and MIA. The process flow was

First, the photosensitive adhesive (NOA61) was dropped on the prepared MIA, as shown in Figure 7d. After it was leveled (Figure 7e), the imprinting mold was used to imprint photosensitive adhesive. During imprinting, the high-precision alignment device was used to align the microlenses with the microimages one by one. On the basis of alignment, the photosensitive adhesive was irradiated by ultraviolet light with a wavelength of 365 nm until it was cured, as shown in Figure 7f. Finally, the PDMS mold was peeled off (Figure 7g) to obtain an integrated 3D display film element, as shown in Figure 8. Finally, the PDMS mold was peeled off (Figure 7g) to obtain an integrated 3D display film element, as shown in Figure 8. Figure 8a shows the 3D effect of the planar display, and the 3D display effect can be seen from various angles. Figure 8b shows the 3D display effect after bending of the element. Meanwhile, the weight of the 3D display element on the flexible substrate has been characterized, which is less than 1 g, as shown in Figure 8c, reaching the lightweight level. Figure 9 shows the 3D effects from different viewing angles. as follows: First, the photosensitive adhesive (NOA61) was dropped on the prepared MIA, as shown in Figure 7d. After it was leveled (Figure 7e), the imprinting mold was used to imprint photosensitive adhesive. During imprinting, the high-precision alignment device was used to align the microlenses with the microimages one by one. On the basis of alignment, the photosensitive adhesive was irradiated by ultraviolet light with a wavelength of 365 nm until it was cured, as shown in Figure 7f. Finally, the PDMS mold was peeled off (Figure 7g) to obtain an integrated 3D display film element,

**Figure 8.** Integration of thin film element: (**a**) planar display; (**b**) curved display; (**c**) weight. **Figure 8.**Integration of thin film element: (**a**) planar display; (**b**) curved display; (**c**) weight.

Figure 8a shows the 3D effect of the planar display, and the 3D display effect can be seen from various angles. Figure 8b shows the 3D display effect after bending of the element. Meanwhile, the weight of the 3D display element on the flexible substrate has been characterized, which is less than 1 g, as shown in Figure 8c, reaching the lightweight level. Figure 9 shows the 3D effects from different viewing angles. **Figure 8.** Integration of thin film element: (**a**) planar display; (**b**) curved display; (**c**) weight.

time, the element has the characteristics of miniaturization and light weight, which can be applied to product packaging, handicrafts, anti-counterfeiting, and other industries. Using the 3D display effect **Figure 9.** 3D effects from different viewing angles: (**a**) −15°; (**b**) −10°; (**c**) 0°; (**d**) 5°; (**e**) 20°. **Figure 9.** 3D effects from different viewing angles: (**a**) −15◦ ; (**b**) −10◦ ; (**c**) 0◦ ; (**d**) 5◦ ; (**e**) 20◦ .

#### to replace the original two-dimensional image display has a certain market application prospect, and **5. Conclusion 5. Conclusions**

review and editing, L.L. and L.S.

also lays the technical foundation for wearable display equipment. **Author Contributions:** Conceptualization, A.C.; Formal analysis, A.C. and L.X.; Funding acquisition, Q.D.; Investigation, L.X.; Software, Y.P.; Validation, H.P.; Visualization, L.L.; Writing – original draft, A.C.; Writing – review and editing, L.L. and L.S. In this paper, we propose to design and fabricate a 3D display film element based on microfabrication. The imaging resolution is higher than that of the human eye at 300 ppi. At the same time, the element has the characteristics of miniaturization and light weight, which can be applied to In this paper, we propose to design and fabricate a 3D display film element based on microfabrication. The imaging resolution is higher than that of the human eye at 300 ppi. At the same time, the element has the characteristics of miniaturization and light weight, which can be applied

product packaging, handicrafts, anti-counterfeiting, and other industries. Using the 3D display effect

Youth Innovation Promotion Association, CAS and CAS "Light of West China" Program. The authors thank

**Author Contributions:** Conceptualization, A.C.; Formal analysis, A.C. and L.X.; Funding acquisition, Q.D.; Investigation, L.X.; Software, Y.P.; Validation, H.P.; Visualization, L.L.; Writing – original draft, A.C.; Writing –

**Funding:** This research was supported by the National Natural Science Foundation of China (Nos. 61605211, 61905251); The Instrument Development of Chinese Academy of Sciences (No. YJKYYQ20180008); The National R&D Program of China (No. 2017YFC0804900); Sichuan Science and Technology Program (No. 2019YJ0014); Youth Innovation Promotion Association, CAS and CAS "Light of West China" Program. The authors thank

their colleagues for their discussions and suggestions to this research.

their colleagues for their discussions and suggestions to this research.

**Conflicts of Interest:** The authors declare no conflict of interest.

**Conflicts of Interest:** The authors declare no conflict of interest.

to product packaging, handicrafts, anti-counterfeiting, and other industries. Using the 3D display effect to replace the original two-dimensional image display has a certain market application prospect, and also lays the technical foundation for wearable display equipment.

**Author Contributions:** Conceptualization, A.C.; Formal analysis, A.C. and L.X.; Funding acquisition, Q.D.; Investigation, L.X.; Software, Y.P.; Validation, H.P.; Visualization, L.L.; Writing—original draft, A.C.; Writing—review and editing, L.L. and L.S.

**Funding:** This research was supported by the National Natural Science Foundation of China (Nos. 61605211, 61905251); The Instrument Development of Chinese Academy of Sciences (No. YJKYYQ20180008); The National R&D Program of China (No. 2017YFC0804900); Sichuan Science and Technology Program (No. 2019YJ0014); Youth Innovation Promotion Association, CAS and CAS "Light of West China" Program. The authors thank their colleagues for their discussions and suggestions to this research.

**Conflicts of Interest:** The authors declare no conflict of interest.

## **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Review* **Emerging Designs of Electronic Devices in Biomedicine**

**Maria Laura Coluccio <sup>1</sup> , Salvatore A. Pullano <sup>2</sup> , Marco Flavio Michele Vismara <sup>2</sup> , Nicola Coppedè 3 , Gerardo Perozziello <sup>1</sup> , Patrizio Candeloro <sup>1</sup> , Francesco Gentile 4,\* and Natalia Malara 1,\***


Received: 30 November 2019; Accepted: 20 January 2020; Published: 22 January 2020

**Abstract:** A long-standing goal of nanoelectronics is the development of integrated systems to be used in medicine as sensor, therapeutic, or theranostic devices. In this review, we examine the phenomena of transport and the interaction between electro-active charges and the material at the nanoscale. We then demonstrate how these mechanisms can be exploited to design and fabricate devices for applications in biomedicine and bioengineering. Specifically, we present and discuss electrochemical devices based on the interaction between ions and conductive polymers, such as organic electrochemical transistors (OFETs), electrolyte gated field-effect transistors (FETs), fin field-effect transistor (FinFETs), tunnelling field-effect transistors (TFETs), electrochemical lab-on-chips (LOCs). For these systems, we comment on their use in medicine.

**Keywords:** biodevices; integration; miniaturized devices

#### **1. Introduction**

Theranostics is universally understood to be the use of a combination of nanoscale agents and techniques that have both diagnostic and therapeutic effects on a disease. Theranostics provides a transition from conventional medicine to a personalized and precision medicine approach.

It includes a large variety of themes including, for example, molecular imaging, personalized medicine, or pharmacogenomics that expand the field of knowledge on targeted therapies and enhance our understanding of the molecular mechanisms of drugs. In the last years, recent advances in microfluidics [1–4] and nanotechnology [5–8] gave significant support to theranostics for developing new procedures and treatments of diseases.

This review focuses on the development of strategies and potential applications of emerging theranostic nanosystems based on the transport of electroactive species (i.e., ions or electrons) at the nanoscale. Faster diagnosis and screening of diseases have become increasingly important in predictive personalized medicine as they improve patient treatment strategies and reduce cost as well as the burden of healthcare. However, the correct extraction, acquisition, and sampling of physiological signals is still unsatisfied by many of the currently available approaches; for these, the analysis of body fluids such as tears, sweat, saliva, or interstitial fluid, is heavily conditioned by the type of biomarker integrated with the electrochemical/bioelectronics sensor that performs the analysis.

The choice of the correct biomarker is a critical step that can compromise the entire process of measurement and overrule even the more advanced technology of a sensing device. The blind biomarker should exhibit the properties of high specificity and sensitivity in monitoring a disease. This adherence must be investigated in a preliminary stage, before the use of the biomarker in the device in its final configuration. Then, the aim of the (electrochemical) device is that of enhancing the sensing abilities of the biomarker in terms of limit of detection (i.e., the smallest quantity or concentration of the analyte to be detected), resolution (i.e., the smallest incremental unit of the analyte that the biosensor can detect), and sensitivity (i.e., how much the response of the system changes as the input changes). A nanoscale architecture of the device can improve the limit of detection, resolution, and sensitivity of the biosensor. This review is an account of how nanotechnology can enter the field of electrochemical transistors to impact clinical medicine. The circuit functionalities and their applications will also be addressed, with attention to current trends in the field.

For their characteristics of high sensitivity and fast response times, electrochemical biosensors may provide early diagnosis of diseases and increase the possibility of a patient's recovery. Perhaps more importantly, electrochemical biosensors are able to translate a chemical signal into an electrical signal and this enables us to detect and quantify several different molecular or cellular species in the body. Moreover, these systems can be integrated with lab-on-chips to obtain point of care (POC) analytical platforms.

In the fields of theranostics and prognosis, it is necessary to develop devices with a high impact on the detection sensitivity and specificity of the biomarker that in turn must be specific and adherent to the disease under examination. The integration between electrochemical biosensors and lab-on-chips (LOCs) to obtain POC analytical platforms is discussed with plenty of examples. We report, describe, and comment on latest generation transistors, electrochemical biosensors (fin field-effect transistors (FinFETs), tunnelling field-effect transistors (TFETs), and organic electrochemical transistors (OECTs)), and the combination of electrochemical biosensors with lab-on-chips for medical applications.

#### **2. Fin Field-E**ff**ect Transistor (FinFET), Tunnel FET**

The field-effect transistor is a type of transistor that uses an electric field to control the flow of current. A typical FET device has three terminals: source, gate, and drain. The application of a voltage to the FET's gate modifies the conductivity between the drain and the source, allowing the control of the flow of current.

One of the first examples of field-effect devices for the evaluation of ionic species was introduced by Bergveld in the 1970s and named ion sensitive field-effect transistor (ISFET) [9].

In this class of devices, the gate consists of an SiO<sup>2</sup> layer placed in solution, and consequently, the drain current is influenced by analyte activity by varying the potential at the gate/electrolyte interface [9,10]. Subsequently, different technologies of field-effect devices were developed to overcome the main limits imposed by FET–based devices, such as the threshold voltage drift. In chronological order the extended gate field effect transistor (EGFET) was developed by Van der Spiegel in the early 1980s [11]. The continuous scaling of planar metal-oxide-semiconductor field-effect transistor (MOSFET) evidenced come technological difficulties whereby the device can no longer be classified as a long channel MOSFET. The main short-channel effect is due to the two-dimensional distribution of potential and high electric fields in the channel region, which mainly lead to variable threshold voltage, saturation region that does not depend on drain potential, and drain current that does not depend on the inverse of channel length [12]. Subsequently a variety of different planar and non-planar topologies were investigated, such as the FinFET and the TFET [13,14]. The FinFET resulted in an attractive option for the fabrication of a non-planar device with self-aligned double-gate using a standard complementary metal-oxide semiconductor (CMOS) process [15]. Conversely, the FET is considered a promising design which guarantees immunity to subthreshold swing degradation at

short channel length [16]. More recently, the development of more sophisticated organic materials led to the development of devices that are trying to replace the role of silicon, such as the organic thin-film transistors (OTFTs). Even though the OTFTs' performances are actually not comparable with inorganic ones in terms of carrier mobility, operating frequency, and subthreshold swing, the low cost and easier fabrication, as well as the possibility to chemically modify the material properties, represent a key role in the development of organic biosensors [17]. *Micromachines* **2020**, *11*, 123 3 of 21 of silicon, such as the organic thin‐film transistors (OTFTs). Even though the OTFTs' performances are actually not comparable with inorganic ones in terms of carrier mobility, operating frequency, and subthreshold swing, the low cost and easier fabrication, as well as the possibility to chemically modify the material properties, represent a key role in the development of organic biosensors [17].

The electronic interface of a field-effect biosensors' electronic interface is a standard and/or custom designed MOSFET, which ensures long-term stability and insulation from the chemical environment to the device (where the sensing layer is generally placed) [16]. In linear region, the sensor output (i.e., drain current) is related to the analyte as follows: The electronic interface of a field‐effect biosensors' electronic interface is a standard and/or custom designed MOSFET, which ensures long‐term stability and insulation from the chemical environment to the device (where the sensing layer is generally placed) [16]. In linear region, the sensor output (i.e., drain current) is related to the analyte as follows:

$$I\_{DS} = \mu \mathcal{C}\_{\alpha \chi} \frac{W}{L} \left[ (V\_{Ref} - V\_{th}^\*) V\_{DS} - \frac{1}{2} V\_{DS}^2 \right] \tag{1}$$

here *W* is the width and *L* is the length of the channel, µ is the carrier mobility, *COx* is the gate oxide capacitance per unit area, and *VRe f* and *VDS* are the applied reference electrode and the drain-to-source voltages. *V* ∗ *th* is the threshold voltage, which is related to the device and the chemical environment as follows: oxide capacitance per unit area, and ோ and ௌ are the applied reference electrode and the drain‐ to‐source voltages. ௧ <sup>∗</sup> is the threshold voltage, which is related to the device and the chemical environment as follows:

$$V\_{th}^{\*} = V\_{th} + E\_{ref} + \chi\_{sol} - \frac{W\_M}{q} - \phi \tag{2}$$

In Equation (2), *Vth* is the threshold voltage of the field-effect device. *ERe f* is the reference electrode potential, χ*sol* is the dipole potential of the electrolyte, *W<sup>M</sup>* is the work function of the reference electrode, *q* is the charge, and ϕ is the potential at the sensing interface [10]. In Figure 1 is reported a comparison between an EGFET and a Fin-FET device for biosensing. In Equation (2), ௧ is the threshold voltage of the field‐effect device. ோ is the reference electrode potential, ௦ is the dipole potential of the electrolyte, ெ is the work function of the reference electrode, is the charge, and is the potential at the sensing interface [10]. In Figure 1 is reported a comparison between an EGFET and a Fin‐FET device for biosensing.

**Figure 1.** Representative view of a device based on extended gate field effect transistor (EGFET) (**a**) and (**b**) fin field‐effect transistor (Fin‐FET) technology (not in scale). **Figure 1.** Representative view of a device based on extended gate field effect transistor (EGFET) (**a**) and (**b**) fin field-effect transistor (Fin-FET) technology (not in scale).

In the EGFET aspect ratio / influences the characteristics of the devices in terms of transconductance . A higher transconductance is desirable because it results in lower flicker (1/) noise device. There are several advantages in FinFET such as a lower off current, a lower ௧ due to a reduced bulk (depletion) capacitance and a very low output conductance (higher voltage gain). Conversely, FinFETs suffer from a high series resistance and thus a lower peak transconductance [18]. In both cases the literature reports that most of the developed biosensors are developed using commercial devices, evidencing how the development of biosensors over the years has been mostly oriented toward the sensing part, using off‐the‐shelf components because of the easier fabrication process and lower cost [19,20]. Even though they are a more recent technology, FinFETs have more recently been commercialized and thus are mature for biosensor applications [21]. One of the key metrics in the development of biosensors is their sensitivity, which through the years has been the object of intensive investigation, especially for the detection of analytes at even lower concentration. Being inherently characterized by theoretical limits other approaches were investigated instead of classical planar and non‐planar geometries. The TFET is one of the most recent devices, with base conduction mechanism on the band‐to‐band tunneling. In this class of device, the analyte influences the tunneling barrier, and hence the tunneling current. Literature has evidenced that the use of TFET technology results in devices with improved sensitivity and reduced response time, while retaining In the EGFET aspect ratio *W*/*L* influences the characteristics of the devices in terms of transconductance *gm*. A higher transconductance is desirable because it results in lower flicker (1/ *f*) noise device. There are several advantages in FinFET such as a lower off current, a lower *Vth* due to a reduced bulk (depletion) capacitance and a very low output conductance (higher voltage gain). Conversely, FinFETs suffer from a high series resistance and thus a lower peak transconductance [18]. In both cases the literature reports that most of the developed biosensors are developed using commercial devices, evidencing how the development of biosensors over the years has been mostly oriented toward the sensing part, using off-the-shelf components because of the easier fabrication process and lower cost [19,20]. Even though they are a more recent technology, FinFETs have more recently been commercialized and thus are mature for biosensor applications [21]. One of the key metrics in the development of biosensors is their sensitivity, which through the years has been the object of intensive investigation, especially for the detection of analytes at even lower concentration. Being inherently characterized by theoretical limits other approaches were investigated instead of classical planar and non-planar geometries. The TFET is one of the most recent devices, with base conduction mechanism on the band-to-band tunneling. In this class of device, the analyte influences the tunneling barrier, and hence the tunneling current. Literature has evidenced that the use of TFET

all other advantages of FET biosensors [22–24]. FinFET technology was originally proposed as an

technology results in devices with improved sensitivity and reduced response time, while retaining all other advantages of FET biosensors [22–24]. FinFET technology was originally proposed as an improved technology characterized by higher sensitivity, stability, and reliability [25]. Literature reports some attempts to fabricate FinFET-based sensors for biomolecule detection such as cellular ion activities [26], pH [25,27], and the detection of avian influenza (AI) antibody [28]. Change in current was recently linked to change in gate capacitance, allowing the detection of proteins linked to early detection of diseases (e.g., streptavidin, biotin) [29]. Moreover, being a relatively recent technology, modelling tools are still under development to optimize the design phase [30].

The continuous efforts to improve the sensing performances attracted significant attention through the recent technological advancement in synthesis and deposition on high performance materials, such as graphene (i.e., nanopores, nanoribbon, reduced graphene oxide and graphene oxide), carbon nanotube, nanowires, and nanoporous materials [31–38]. Graphene is a high-performance material, recently investigated in different fields due to the availability of synthesis and mature deposition technologies, characterized by high carrier mobility and low inherent 1/ *f* noise [39]. As result, different attempts at using a graphene-FET (GFET) as biosensor were reported in literature. Most of the applications are focused on low-concentration nucleic acid detection, exploiting the site-specific immobilization of probes [32]. The reported resolution of GFETs, often conjugated with metal nanoparticles (e.g., Au) can be lowered down to the pM range [40,41]. Despite the interesting sensing applications, continuous investigations are still required to improve the reduced DNA translocation dynamics and the low-frequency noise levels [32,42,43]. Other applications are concerned with protein detection, living cell and bacteria monitoring [33,44]. A commercial graphene-based biosensor is the agile biosensor chip—NTA, used overall for research purposes, allowing the immobilization of recombinant proteins.

The efforts to develop the FinFET device often lead to biosensors with performances comparable with that of other multigate or planar MOSFETs [25]. Subsequently, in order to reduce the development time, commercial FET devices are sometimes preferred.

#### **3. Organic Electrochemical Transistor Devices**

The role of organic electrochemical transistors in biomedicine is becoming increasingly relevant. OECTs are devices based on a semiconductor, conventionally named channel, that is physically placed in contact with a solution. Upon the action of an externally controlled voltage, ions of the solution are displaced and can be either injected or removed from the channel, doping or dedoping it, changing the bulk conductivity of the entire device. Alterations in the electrical characteristics of the device can be in turn correlated to the physical and chemical characteristics of the electrolyte. State of the art OECTs are mostly based on the conducting polymer poly(3,4-ethylenedioxythiophene) doped with polystyrene sulfonate (PEDOT:PSS) [45], thus OECTs share many of the characteristics of polymers such as low weight, low density, low cost, high resilience, elevated specific strength, biocompatibility, and facile deposition. Moreover, because they have a transistor architecture, OECTs feature high sensitivity, high signal-to-noise ratio, high gain [46–48], enabling amplification of weak electric signals such as those generated by biological systems.

OECTs, in fact, can operate at low voltages, in aqueous environments such as efficient ion-to-electron converters, providing an interface between the worlds of biology and electronics.

As key constituents of biosensors and analytical devices, OECTs have been used for electrophysiological recording, for bio-sensing applications and applications at the bio-interface [45,49, 50], bio-computing [51], neuromorphic engineering [52–54], as constituents in electronic bio-devices [55], and as a sensor for cells [56].

As flexible, high-sensitivity low-cost devices, OECTs have found different applications in biomedicine and biology. They have been applied as sensors for simple analytes such as hydrogen peroxide [57] and ions [47,58]. In reference [59] Seong-Min Kim and colleagues examined the long term stability of PEDOT:PSS, examined the correlations among the microstructure, composition, and device performance of PEDOT:PSS films for possible applications in the development of long-term stable implantable bioelectronics for neural recording/stimulation. In reference [60], devices based on the conducting polymer PEDOT:PSS have been demonstrated for the real-time processing and manipulation of signals from living organisms; devices with the characteristics of miniaturization and bio-compatibility with human skin have been used to analyze neurophysiological activity. In references [61–63] it is discussed how similar OECTs can be used as sensing interfaces of cells and systems of cells. OECTs can very practically determine the physiological conditions of living cells, follow the processes at the basis of their life cycle, including reproduction processes, and track their transition to apoptosis [63].

Moreover, OECTs can be used as electronic switches or components of logic gates, also in interaction with biological interfaces, creating a multiple interactive logic that is perfectly suited to talk with living systems [64]. The in vivo monitoring of biological-driven phenomena is likewise heavily investigated, including, for example, enzymatic interactions [65] useful to detect metabolites relevant in in the biological processes and function of cells and organs, such as lactate or glucose, which are indicative of the physiological conditions of a patient [66]. Finally, a wide range of applications is still open in the detection of biomolecules based on specific antigens, which are crucial in the in vivo early diagnosis of bacteria and specific illnesses [67].

The typical configuration of an OECT is reported in Figure 2. An electrolyte, i.e., a solution containing electroactive species, is contacted to the device with three electrodes: the gate, the drain, and the source. The gate is the reference electrode that is directly connected to the electrolyte. Instead, the drain and the source are bridged together by the conductive polymer channel (Figure 2a). Upon application of a voltage between the gate and the source (*Vgs*) and the drain and the source (*Vds*), ions in the electrolyte are propelled towards the polymer channel, penetrate into the channel, and generate a current *Ids* that flows from the drain to the source (Figure 2b). Thus, the current *Ids* is the typical output of an OECT device. The reaction between the ions and the polymer in the channel is described by the following equation:

$$\text{PEDOT}^+ : \text{PSS}^- + \text{M}^+ + e^- \rightarrow \text{PEDOT} + \text{M}^+ : \text{PSS}^- \tag{3}$$

where M<sup>+</sup> represents the cations. The presence of cations in the PEDOT : PSS film depletes the number of available carriers, reducing the source–drain current *Ids*. Thus, *Ids* is modulated by the inflow or outflow of ions—from the electrolyte to the channel—and values of current measured by the device can be indicative of the concentration, size, and charge of the specie initially dispersed in solution, and of the geometrical characteristics of the system. The form of the *Ids* current is similar to response of a first order system: values of current are a function of time and increase from a reference value (minimum background current, no flux) to a steady state value (maximum value of current, constant regime) (Figure 2c). The signal can be modelled by an exponential function of the type *Ids* ∼ *m* 1 − *e* −*t*/τ , where *t* is time, and *m* and τ are the modulation and time constants of the system. The modulation is the signal increment normalized to its initial value, *m* = *I f in ds* − *I* 0 *ds* /*I* 0 *ds*; it is proportional to the strength of the signal. The time constant is the time after which the signal attains 67% of its steady state value, *Ids*(τ) ∼ 0.67 *I f in ds* ; it is indicative of the inertia of the system. Remarkably, using mathematical models described elsewhere [68] *m* and τ can be associated with the diffusivity, charge, concentration, and other physical, chemical and geometrical characteristics of a system, so that the information encoded in the *Ids* can be broken to extract the characteristics of the system in analysis. Typical *Ids* values and increments fall in the 0–5 mA range. The values of *Vds*, instead, are controlled by the operator and are typically varied in discrete increments in the 0–1 V range.

0–1 V range.

transition to apoptosis [63].

by the following equation:

*Micromachines* **2020**, *11*, 123 5 of 21

and bio‐compatibility with human skin have been used to analyze neurophysiological activity. In references [61–63] it is discussed how similar OECTs can be used as sensing interfaces of cells and systems of cells. OECTs can very practically determine the physiological conditions of living cells, follow the processes at the basis of their life cycle, including reproduction processes, and track their

Moreover, OECTs can be used as electronic switches or components of logic gates, also in interaction with biological interfaces, creating a multiple interactive logic that is perfectly suited to talk with living systems [64]. The in vivo monitoring of biological‐driven phenomena is likewise heavily investigated, including, for example, enzymatic interactions [65] useful to detect metabolites relevant in in the biological processes and function of cells and organs, such as lactate or glucose, which are indicative of the physiological conditions of a patient [66]. Finally, a wide range of applications is still open in the detection of biomolecules based on specific antigens, which are crucial

The typical configuration of an OECT is reported in Figure 2. An electrolyte, i.e., a solution containing electroactive species, is contacted to the device with three electrodes: the gate, the drain, and the source. The gate is the reference electrode that is directly connected to the electrolyte. Instead, the drain and the source are bridged together by the conductive polymer channel (Figure 2a). Upon application of a voltage between the gate and the source (௦) and the drain and the source (ௗ௦), ions in the electrolyte are propelled towards the polymer channel, penetrate into the channel, and generate a current ௗ௦ that flows from the drain to the source (Figure 2b). Thus, the current ௗ௦ is the typical output of an OECT device. The reaction between the ions and the polymer in the channel is described

where Mା represents the cations. The presence of cations in the PEDOT: PSS film depletes the number of available carriers, reducing the source–drain current ௗ௦. Thus, ௗ௦ is modulated by the inflow or outflow of ions—from the electrolyte to the channel—and values of current measured by the device can be indicative of the concentration, size, and charge of the specie initially dispersed in solution, and of the geometrical characteristics of the system. The form of the ௗ௦ current is similar to response of a first order system: values of current are a function of time and increase from a reference value (minimum background current, no flux) to a steady state value (maximum value of current, constant regime) (Figure 2c). The signal can be modelled by an exponential function of the type ௗ௦~൫1 െ ି௧/ఛ൯, where is time, and and are the modulation and time constants of the

it is proportional to the strength of the signal. The time constant is the time after which the signal

Remarkably, using mathematical models described elsewhere [68] and can be associated with the diffusivity, charge, concentration, and other physical, chemical and geometrical characteristics of a system, so that the information encoded in the ௗ௦ can be broken to extract the characteristics of

ௗ௦, instead, are controlled by the operator and are typically varied in discrete increments in the

system. The modulation is the signal increment normalized to its initial value, ൌ ൫ௗ௦

attains 67% of its steady state value, ௗ௦ሺሻ~0.67 ௗ௦

PEDOTା: PSSି Mା ି → PEDOT Mା: PSSି (3)

െ ௗ௦

; it is indicative of the inertia of the system.

 ൯ ௗ௦ ൗ ;

in the in vivo early diagnosis of bacteria and specific illnesses [67].

**Figure 2.** Scheme of a conventional organic electrochemical transistor (OECT) device, in which a solution is connected to the device through the gate, source and gate electrodes, and a conductive **Figure 2.** Scheme of a conventional organic electrochemical transistor (OECT) device, in which a solution is connected to the device through the gate, source and gate electrodes, and a conductive polymer channel (**a**). Upon application of an external voltage at the gate and the drain (**b**), a current of ions flows to the source generating a continuous function of time (**c**). *Micromachines* **2020**, *11*, 123 6 of 21 polymer channel (**a**). Upon application of an external voltage at the gate and the drain (**b**), a current

of ions flows to the source generating a continuous function of time (**c**).

In this scheme, the electrolyte is contained in a channel, a chamber, or a reservoir; the interface of the solution with the external regions of the device is a flat surface with zero curvature. This automatically implies that the motion of ions in the system to the active sites of the device is driven by diffusion: the process is inefficiently controllable by the outside. By nanostructuring of the surface of the polymer channel, one can make the surface super-hydrophobic. Super-hydrophobicity prevents wetting of the surface, allowing a drop of solute positioned on that surface to maintain its originating spherical shape. The curvature of the drop can be modulated by tailoring the geometry of the super-hydrophobic surface and by regulating the amount of the solute partitioned in each drop. Thanks to a non-zero curvature, convective Marangoni flows arise in the drop (Figure 3a). In this scheme, the electrolyte is contained in a channel, a chamber, or a reservoir; the interface of the solution with the external regions of the device is a flat surface with zero curvature. This automatically implies that the motion of ions in the system to the active sites of the device is driven by diffusion: the process is inefficiently controllable by the outside. By nanostructuring of the surface of the polymer channel, one can make the surface super‐hydrophobic. Super‐hydrophobicity prevents wetting of the surface, allowing a drop of solute positioned on that surface to maintain its originating spherical shape. The curvature of the drop can be modulated by tailoring the geometry of the super‐hydrophobic surface and by regulating the amount of the solute partitioned in each drop. Thanks to a non‐zero curvature, convective Marangoni flows arise in the drop (Figure 3a).

**Figure 3.** Nanoscale modification of a surface can make that surface super‐hydrophobic (**a**). In a drop on a super‐hydrophobic surface, the motion of particles is determined by the combination of diffusion and Marangoni convective flows (**b**). The inset reports a graphical representation of the potential functions ψ that describe the velocity field within a spherical drop (**c**). The displacement of particles in a drop can be determined from the velocity field using the Langevin equation and a numerical scheme (**d**). **Figure 3.** Nanoscale modification of a surface can make that surface super-hydrophobic (**a**). In a drop on a super-hydrophobic surface, the motion of particles is determined by the combination of diffusion and Marangoni convective flows (**b**). The inset reports a graphical representation of the potential functions ψ that describe the velocity field within a spherical drop (**c**). The displacement of particles in a drop can be determined from the velocity field using the Langevin equation and a numerical scheme (**d**).

Depending on the value of curvature, the size of the drop, and the gradient of temperature between the substrate and the drop, the intensity of the convective fields can be equal, or greater or less, than the intensity of diffusion. Thus, convection represents the additional degree of freedom Depending on the value of curvature, the size of the drop, and the gradient of temperature between the substrate and the drop, the intensity of the convective fields can be equal, or greater or

and . The stream functions are potential functions that describe the characteristics of a fluid flow,

they have been originally derived by Tam and collaborators and [70] read:

less, than the intensity of diffusion. Thus, convection represents the additional degree of freedom introduced in the system [69], and the competition between convection and diffusion drives the solute species on predefined spots. The velocity field developed within the drop owing to Marangoni flows is derived as the derivative of the stream functions ψ(*r*, θ) with respect to the coordinates *r* and θ. The stream functions are potential functions that describe the characteristics of a fluid flow, they have been originally derived by Tam and collaborators and [70] read:

$$\Psi(r,\theta) = -\frac{1}{8} \Big( 1 - r^2 \Big) \Big[ 1 + r \cos \theta - \frac{1 - r^2}{(r^2 + 1 - 2r \cos \theta)^{\frac{1}{2}}} + \sum\_{n=2}^{\infty} \frac{(n-1) - 2(n-1)\aleph i}{(2n-1)((n-1) + \aleph i)} r^n (P\_{n-2} \cos \theta - P\_n \cos \theta) \Big], \tag{4}$$

where (*r*, θ) is the position of a point in the drop in polar coordinates, Bi is the Biot number, and *P<sup>n</sup>* is the Legendre polynomial of order *n*. The corresponding velocity field *v* is then derived as:

$$v\_r = -\frac{1}{r^2 \sin \theta} \frac{\partial \psi}{\partial \theta}, \ v\_\theta = \frac{1}{r \sin \theta} \frac{\partial \psi}{\partial r}, \tag{5}$$

that can be used, in turn, in the Langevin equation [71–73] to find the distribution of a trace in the drop:

$$m\frac{\partial \mathbf{u}}{\partial t} = 6\pi\mu a \left(\mathbf{K}\_p \mathbf{u} - \mathbf{K}\_f \mathbf{w}\right) + F\_e + F\_b. \tag{6}$$

In Equation (6) *u* is the unknown velocity vector for the particle, *v* is the unperturbed fluid velocity, *m* and *a* are the mass and radius of the solute particulates. Moreover, *F<sup>e</sup>* is the electrostatic force, *Fb* is the Brownian force that depends on the temperature as *F<sup>b</sup>* ∝ √ *T*, *K<sup>p</sup>* and *K<sup>f</sup>* account for the hydrodynamic hindrance of the system, and *t* is time. Equation (6), solved using a numerical scheme, allows us to determine how a solute propagates in a super-hydrophobic drop because of convection and diffusion.

Inspired by Lotus leaves and by nature, super-hydrophobic surfaces have been reproduced using combinations of nano-fabrication techniques [74,75]. Typically, the artificial analogue of a super-hydrophobic surface is an array of microsized pillars, where the size (*d*), spacing (δ), and height (*h*) of the pillars in the array can vary over large intervals. The upper surface of the pillars contains, in turn, details at the nanoscale, and the combined effects across length-scales cause the surface to be super-hydrophobic [76–78]. The pillars of those surfaces are often made out of silicon, nano-machined using reactive ion etching techniques, or of polymers, created using optical or electron beam lithography techniques [79]. Fluorinated polymers with low friction coefficients, nano-porous silicon, or nano-rough materials with low surface energy densities, can be deposited on the pillars representing the second-level roughness of the hierarchical nanomaterial device [80]. The contact angle of a drop on similar super-hydrophobic surfaces can be predicted using the celebrated model of Cassie and Baxter [74]:

$$\cos(\mathfrak{d}^{\mathbb{C}}) = -1 + \phi \cos(\mathfrak{d}),\tag{7}$$

where ϑ is the contact angle of the drop on the surface without texture, ϑ *c* is the contact angle on the surface with the texture, and φ is the solid fraction of the surface. Typical design values of *d* and δ are *d* = 10 µm and δ = 20 µm, so that φ = π/4(*d*/δ) <sup>2</sup> <sup>∼</sup> 0.087. With these values of *<sup>d</sup>*, <sup>δ</sup> and <sup>φ</sup>, any originating contact angle ϑ > 60◦ will lead to final contact angles ϑ *<sup>c</sup>* > 150◦ , i.e., a super-hydrophobic surface. For this combination of *d* and δ, the height of the pillars should be chosen such that *h* ≥ 20 µm, in that ratio of *h* to δ is greater than one, *h*/δ > 1, to assure stable adhesion of the drop on the surface [81].

Motivated by the need of new sensor devices with higher sensitivity, higher accuracy, and increased selectivity with respect to available approaches, beginning in 2014 some of the authors of this paper started to nanostructure OECTs with the aim to harness their functionalities [82]. Starting from conventional OECT devices, we modified the geometry of those devices at the nanoscale, and created a new class of bio-devices that we called surface enhanced organic electrochemical transistors (SeOECTs) [69]. SeOECTs are a third generation of organic thin film transistors, in which the electrolyte medium is an active part of the device gating and the surface micro and nanostructure enhances the properties of the electrochemically active conductive polymer. In SeOECTs, a 3-dimensional design and topographical modification of the surface enables selectivity, enhances sensitivity, and enables the detection of multiple analytes in very low abundance ranges.

SeOECTs are based on the fine tailoring of surface microstructure and nano structure. The device comprises arrays of super-hydrophobic micro-pillars, functionalized with a conductive PEDOT:PSS polymer sensitive to the ionic strength of the electrolyte. Each pillar has a diameter of 10 µm and a height of 20 µm. The pillars are positioned on the substrate to form a non-periodic lattice (Figure 4a). A similar non-uniform tiling of pillars generates a system of radial forces that recalls the drop to the center of the lattice for automatic sample positioning. Some of the pillars are individually contacted to an external electrical probe station for site selective measurement on the sample surface (Figure 4b,c); they incorporate nano-gold contacts with sub-micron reciprocal distance that generate enhanced and localized electric fields (Figure 4d–h). Due to the microstructure of the device, the device is super-hydrophobic with contact angles up to 165◦ (Figure 4i). The device takes advantage of a combination of scales to resolve, identify, and measure complex biological mixtures. At the micro-scale, arrays of super-hydrophobic micro pillars enable manipulation and control of biological fluids. At the nano-scale, some pillars are modified to incorporate nano-electrodes for time and space resolved analysis of solutions. *Micromachines* **2020**, *11*, 123 8 of 21 polymer sensitive to the ionic strength of the electrolyte. Each pillar has a diameter of 10 μm and a height of 20 μm. The pillars are positioned on the substrate to form a non‐periodic lattice (Figure 4a). A similar non‐uniform tiling of pillars generates a system of radial forces that recalls the drop to the center of the lattice for automatic sample positioning. Some of the pillars are individually contacted to an external electrical probe station for site selective measurement on the sample surface (Figure 4b,c); they incorporate nano‐gold contacts with sub‐micron reciprocal distance that generate enhanced and localized electric fields (Figure 4d–h). Due to the microstructure of the device, the device is super‐hydrophobic with contact angles up to 165° (Figure 4i). The device takes advantage of a combination of scales to resolve, identify, and measure complex biological mixtures. At the micro‐scale, arrays of super‐hydrophobic micro pillars enable manipulation and control of biological fluids. At the nano‐scale, some pillars are modified to incorporate nano‐electrodes for time and space resolved analysis of solutions.

**Figure 4.** Surface enhanced organic electrochemical transistors (SeOECT) devices are based on a non‐ periodic array of micro‐pillars (**a**). Some of those pillars are contacted through circuits to an external electric probing station (**b**) and are equipped with nanoelectrodes for site selective measurement of the ionic current: each of those pillars is named sensor (**c**–**f**). The device imaged with a camera lens, the distance between the parallel gold circuits is 1 cm (**g**). The sensors are placed in line on the device, symmetrically with respect to the center of the device (**h**). Due to the characteristics of the substrate, during operation the liquid sample maintains a spherical shape (**i**). **Figure 4.** Surface enhanced organic electrochemical transistors (SeOECT) devices are based on a non-periodic array of micro-pillars (**a**). Some of those pillars are contacted through circuits to an external electric probing station (**b**) and are equipped with nanoelectrodes for site selective measurement of the ionic current: each of those pillars is named sensor (**c**–**f**). The device imaged with a camera lens, the distance between the parallel gold circuits is 1 cm (**g**). The sensors are placed in line on the device, symmetrically with respect to the center of the device (**h**). Due to the characteristics of the substrate, during operation the liquid sample maintains a spherical shape (**i**).

SeOECTs are obtained by the superposition of different layers as explained in detail in reference [69]. To perform a measurement, a liquid sample is positioned on the device. Due to the super‐ SeOECTs are obtained by the superposition of different layers as explained in detail in reference [69]. To perform a measurement, a liquid sample is positioned on the device. Due to the super-hydrophobic characteristics of the surface, the sample takes a quasi-spherical shape (drop) and

the sample through the circuit and is measured by the points of measurements (sensors) on the device. Ions in the sample drop are transported by buoyancy and Marangoni flows that originate in the drop because of its curvature. Since the motion of ions—under the combined effect of convective flows and electric field—is directly proportional to the charge, directly proportional to the diffusion is automatically centered on the device. Then, the device is driven by an externally applied voltage, ranging between 0 and 1 volt. Upon the application of the voltage, a current of ions *I*ds flows from the sample through the circuit and is measured by the points of measurements (sensors) on the device. Ions in the sample drop are transported by buoyancy and Marangoni flows that originate in the drop because of its curvature. Since the motion of ions—under the combined effect of convective flows and electric field—is directly proportional to the charge, directly proportional to the diffusion coefficient, and inversely proportional to the size, the process achieves the migration and spatial separation of species in solution. Arrays of sensors, spatially positioned on the device, can resolve this separation in space and time. Thus, the device measures ionic current transients at different positions on the substrate and for different values of voltage. While the information content of the solution is mapped into a whole set of variables, statistical techniques of analyses can be used to decode such information and determine the characteristics of target molecules. *Micromachines* **2020**, *11*, 123 9 of 21 coefficient, and inversely proportional to the size, the process achieves the migration and spatial separation of species in solution. Arrays of sensors, spatially positioned on the device, can resolve this separation in space and time. Thus, the device measures ionic current transients at different positions on the substrate and for different values of voltage. While the information content of the solution is mapped into a whole set of variables, statistical techniques of analyses can be used to decode such information and determine the characteristics of target molecules.

The state of the system in a specific configuration is a point in the *m* − τ plane. Samples with different characteristics are placed in different regions of the diagram (Figure 5a). Thanks to this graphical representation, it is possible to operate sample separation, clustering, and classification (Figure 5b). The separation can be optimized if one considers sensors positioned at opposite extremes of the device or high values of voltage, for which the convective transport effects are amplified and the differences between species with different charge and size are maximized (Figure 5c). Points in the *m* − τ diagram can be parametrized by voltage (Figure 5d) or by time (Figure 5e). In both cases, data processing and analysis generate trajectories, the shapes of which are indicative of the time evolution and of the characteristics of the system (Figure 5f). The state of the system in a specific configuration is a point in the െ plane. Samples with different characteristics are placed in different regions of the diagram (Figure 5a). Thanks to this graphical representation, it is possible to operate sample separation, clustering, and classification (Figure 5b). The separation can be optimized if one considers sensors positioned at opposite extremes of the device or high values of voltage, for which the convective transport effects are amplified and the differences between species with different charge and size are maximized (Figure 5c). Points in the െ diagram can be parametrized by voltage (Figure 5d) or by time (Figure 5e). In both cases, data processing and analysis generate trajectories, the shapes of which are indicative of the time evolution and of the characteristics of the system (Figure 5f).

**Figure 5.** The response of the SeOECT devices is described by the sole variables modulation and time constant (**a**). The scatter plot of modulation against time constant may be indicative of differences between samples and can be used to operate sample separation, clustering, and classification (**b**). Separation between species can be improved by setting high voltage values and using sensors positioned at the border of the drop, where Marangoni flows are maximized (**c**). The modulation and time constant variables can be parametrized by voltage (**d**) and by time (**e**). The form of these trajectories can be indicative of the time evolution of the system (**f**). **Figure 5.** The response of the SeOECT devices is described by the sole variables modulation and time constant (**a**). The scatter plot of modulation against time constant may be indicative of differences between samples and can be used to operate sample separation, clustering, and classification (**b**). Separation between species can be improved by setting high voltage values and using sensors positioned at the border of the drop, where Marangoni flows are maximized (**c**). The modulation and time constant variables can be parametrized by voltage (**d**) and by time (**e**). The form of these trajectories can be indicative of the time evolution of the system (**f**).

SeOECT devices have been used to evaluate tumors [83]. Since cancerous states are associated with an altered protonation state of the intra/extracellular microenvironment, one can estimate the onset and progression of a cancerous disease from a measurement of the ionic content of a blood‐ derived cell culture. We used SeOECTS to evaluate potential perturbation of protein protonation state (i.e., charge) of cell secretome in the extracellular compartment in vitro. We applied the analysis to the conditioned medium of blood culture after a short‐time expansion, derived from patients with, without, and suspected of cancer. Using data from the bio‐chip and statistical techniques of analysis, we developed algorithms that segregated tumor patients from non‐tumor patients. For the ~30 patients across two independent cohorts, the method identified tumor patients with high sensitivity SeOECT devices have been used to evaluate tumors [83]. Since cancerous states are associated with an altered protonation state of the intra/extracellular microenvironment, one can estimate the onset and progression of a cancerous disease from a measurement of the ionic content of a blood-derived cell culture. We used SeOECTS to evaluate potential perturbation of protein protonation state (i.e., charge) of cell secretome in the extracellular compartment in vitro. We applied the analysis to the conditioned medium of blood culture after a short-time expansion, derived from patients with, without, and suspected of cancer. Using data from the bio-chip and statistical techniques of analysis, we developed algorithms that segregated tumor patients from non-tumor patients. For the ~30 patients

The sensing platform is then a crucial point for a specific application of LOCs, consisting of a sort of recognition element required for the capture of the target. The high affinity between antigen–

and 93% specificity [83].

**4. Combined Electrochemical Biosensor and Lab‐on‐Chip**

across two independent cohorts, the method identified tumor patients with high sensitivity and 93% specificity [83].

#### **4. Combined Electrochemical Biosensor and Lab-on-Chip**

The sensing platform is then a crucial point for a specific application of LOCs, consisting of a sort of recognition element required for the capture of the target. The high affinity between antigen–antibody (Ag–Ab) or protein–aptamer makes them largely applied for biosensing design. Nanotechnologies offer support as electrochemical devices because the nanomaterials employed, from silicon to graphene or graphene oxide to carbon nanotubes or to metal nanoparticles, are characterized by excellent electrical, mechanical, and, generically, physical/chemical properties, which, combined with the novel nanotechnology techniques (lithography, metal deposition, plasma treating), allow the realization of appropriate architectures with the appropriate functionalities too. In particular the optical and electrical properties of the sensing nanodevices are related to their materials as well as to their geometry, and an optimal combination of them can amplify the signals coming from the analytes. Consequently, high-sensitivity analyses of biomarkers (usually cellular biomarkers or biomolecules) become possible adopting nanodevices with different techniques (e.g., cyclic voltammetry (CV), electrochemical impedance spectroscopy (EIS), IR, or Raman spectroscopy) [84,85]. Reference [7] reports one of the first examples of a sensor able to capture circulating tumor cells (CTCs), which could be used to study tumor staging, to guide the study of the risk of recurrence. Folic acid (FA) was selected as transducer molecule, because CTCs characterized by high expression of folate receptor (FR) and, consequently, evidencing also 5-methylcytosine-positive nuclei, are potentially dangerous for their dissemination power in healthy tissue. The FA surface can trap cancer-cell-expressing FR, encouraging the attack and the growth of CTC subsets with a higher content of methylated genomic DNA, with notable consequences on tumor prevention strategies and on prognosis definition.

Microfluidic paper-based analytical devices (µPADs) represent a technology of hydrophilic/hydrophobic micro-channel networks and associated analytical devices for development of portable and low-cost diagnostic tools that improve point of care testing (POCT) and disease screening [86] involving the specific detection of biomolecules. They have the ability to perform laboratory operations on micro-scale, using miniaturized equipment, and can be fabricated by using 2-D [86–88] or 3-D [89,90] methods to transport fluids in both horizontal and vertical dimensions, depending on complexity of the diagnostic application. The principal techniques in the literature for fabrication of paper-based microfluidic devices include: wax printing [90], inkjet printing [91], photolithography [92], flexographic printing [93], plasma treatment [94], laser treatment [95], wet etching [96], screen printing [97], and wax screen printing [98]. The µPADs can be used with the naked eye for qualitative testing but can also be used as quantitative assays based on specific detection methods. The choice of the method to detect the binding events that occur on a transducer surface depends on the type of biomarker. The detection methods currently used for sensitive measurements with low detection limits are represented by colorimetry, electrochemistry, fluorescence, chemiluminescence (CL), electrochemiluninescence (ECL), and photoelectrochemistry (PEC). Moreover, the surfaces can be modified to impart high selectivity to the binding of the target analyte, which is desirable in complex biological samples. In fact, sensitivity and specificity of the µPADs can be enhanced through a combination use of the reaction mechanisms categorized into biochemical, immunological, and molecular detections, transforming the device in a multiplexed testing [99–101]. However, chemical amplification or multiplex procedures are often disadvantaged in the µPADs as they require expensive reagents, multiple steps that must be performed by the end user, and complex protocols for the interpretation of the final data.

#### **5. Future Trends of Combined Electrochemical Biosensor and Lab-on-Chip**

The µPADs promise to meet the critical needs of rapid analytical tests in the diagnostic area. These devices represent the diagnostic field, a platform for a wide variety of chemical and biochemical

reactions and detection patterns that can be used to assess the health status of the general population. They are useful for people who must reach very distant health facilities (Figure 6) [102]. These devices can also be used to monitor intoxications in occupational medicine. Finally, their ever-increasing application in the qualitative assessment of foods is beginning. Further research is needed to address several common challenges, such as the poor reproducibility, the need of high detection limits, the inadequate specificity, and the risk of a subjective interpretation of data. Most µPADs successfully address most of these challenges through the association of a machine learning procedure based on "kernel machines." Kernel machines have considerable appeal in the machine learning research community due to a combination of conceptual elegance, mathematical tractability, and state-of-the-art performance [102], and, applied on Android smartphones for image processing and paper-based devices, they are able to solve many of the several common challenges mentioned. *Micromachines* **2020**, *11*, 123 11 of 21 procedure based on "kernel machines." Kernel machines have considerable appeal in the machine learning research community due to a combination of conceptual elegance, mathematical tractability, and state‐of‐the‐art performance [102], and, applied on Android smartphones for image processing and paper‐based devices, they are able to solve many of the several common challenges mentioned.

**Figure 6.** Flowchart of the use of an Android smartphone for image processing and paper‐based devices. In this example, the test is performed at home, the data collected from the paper‐based device through the Android application, and adequate software is sent to qualified medical personnel to support the management of the results obtained**. Figure 6.** Flowchart of the use of an Android smartphone for image processing and paper-based devices. In this example, the test is performed at home, the data collected from the paper-based device through the Android application, and adequate software is sent to qualified medical personnel to support the management of the results obtained.

The calculations of the limit of detection (LOD) and limit of concentration (LOC) for the combination of Android smartphones with image processing and paper‐based devices are considered the frontier for their use. Despite its limitations, the interesting combination of the μPAD and Android applications provides a coherent and objective analysis of colorimetric data without the need of complicated interpretation data methods, and consequently it represents a significant milestone in the current and future development of ePADs for clinical diagnostics Finally, attracting increased attention from the research community is the development of The calculations of the limit of detection (LOD) and limit of concentration (LOC) for the combination of Android smartphones with image processing and paper-based devices are considered the frontier for their use. Despite its limitations, the interesting combination of the µPAD and Android applications provides a coherent and objective analysis of colorimetric data without the need of complicated interpretation data methods, and consequently it represents a significant milestone in the current and future development of ePADs for clinical diagnostics

mobile device‐based healthcare as a revolutionary approach for monitoring medical conditions. More systems have computerized traditional clinical tests to design functions and interactions of smartphone‐based rehabilitation systems [103] regarding diseases like as stroke and cardiac failure [104]. Moreover, sophisticated platforms were developed for a better‐targeted cancer therapy and improved follow‐up care, to make the care process more effective in terms of clinical outcome. On the other hand, there is also the need to develop the μPAD for personalized toxicity studies. Therefore, the future trends on theranostic applications of the μPAD will be developed from the bench‐to‐bedside and updated to produce patient‐friendly analytical assays [105]. **6. Medical Clinical Applications** Finally, attracting increased attention from the research community is the development of mobile device-based healthcare as a revolutionary approach for monitoring medical conditions. More systems have computerized traditional clinical tests to design functions and interactions of smartphone-based rehabilitation systems [103] regarding diseases like as stroke and cardiac failure [104]. Moreover, sophisticated platforms were developed for a better-targeted cancer therapy and improved follow-up care, to make the care process more effective in terms of clinical outcome. On the other hand, there is also the need to develop the µPAD for personalized toxicity studies. Therefore, the future trends on theranostic applications of the µPAD will be developed from the bench-to-bedside and updated to produce patient-friendly analytical assays [105].

#### In cancer, the next generation of POC will probably be represented by 2‐D material‐based **6. Medical Clinical Applications**

[111].

electrochemical biosensors/sensors [106] such as electrochemical apparatus [107], lateral flow assays (LFAs) [108], or paper‐based colorimetric solutions [109]. The main biomarkers relevant in this field are nucleic acids such as mRNA and DNA; proteins such as antigens, enzymes, and peptides; some small molecules such as the reactive species of oxygen and nitrogen; and, notably, the protonation state [83]. Due to the chaotic nature of this medical condition, multiple marker solutions can fit the clinical needs better. Aptamers [110] are artificial oligonucleotides selected through a Systematic In cancer, the next generation of POC will probably be represented by 2-D material-based electrochemical biosensors/sensors [106] such as electrochemical apparatus [107], lateral flow assays (LFAs) [108], or paper-based colorimetric solutions [109]. The main biomarkers relevant in this field are nucleic acids such as mRNA and DNA; proteins such as antigens, enzymes, and peptides; some small molecules such as the reactive species of oxygen and nitrogen; and, notably, the protonation state [83]. Due to the chaotic nature of this medical condition, multiple marker solutions can fit the clinical needs better. Aptamers [110] are artificial oligonucleotides selected through a Systematic evolution of ligands by exponential enrichment (SELEX) procedure. They are gaining appreciation as ultra-specific, stable probes. Their use spans therapeutics to diagnostics, where they can be used as biomarker surrogates or in so-called aptahistochemistry, an evolution of immunohistochemistry [111].

In this section of the review some examples related to the more diffuse cancer types and infectious diseases are reported. Cancer-related applications are summarized in Table 1, infectious disease applications are summarized in Table 2.

In breast cancer aptamers are gaining momentum, and they find usage in therapy, as well as in the detection of diagnostic and prognostic markers such as aberrant HER-2 forms [112,113], α-estrogen receptor status [114], vascular endothelial growth factor (VEGF) [115], osteopontin [116], Michigan Cancer Foundation-7 (MCF-7) cells [117], anterior gradient homolog 2 (AGR-2) protein [118].

In lung cancer, one of the main issues is early diagnostics and effective screening. The problem is to find a proper balance between sensitivity and specificity, and complexity, time, and expenses. According to a recent review by Roointan and colleagues [119], the main biosensor approaches in lung cancer early diagnosis are electrochemical, optical, and piezoelectric (mass-based). The best clinical, analytical, and technological performances are achieved by electrochemical sensors. The main biomarkers sensed by those instruments, are: VEGF165 [120,121], EGFR [122], Annexin II and MUC5AC [123], HIF-1α [124], NADH levels [125].

In colorectal cancer (CRC) the main early diagnostic biomarker is the fecal occult blood test (FOBT) [126,127]. The most advanced FOBT tests commonly available on the market are immunochemical [128,129]. These can be further divided into qualitative and quantitative [130]. One of the main advantages of the immunological-based approach is that patients are allowed to stay on a regular diet without the need to stop drugs known to interfere with the guaiac-based FOBT [128].

Instead of looking for occult blood in stools, it is possible to approach early diagnosis looking for common genetic aberrations commonly found in CRC, such as K-Ras, adenoma polyposis coli (APC), p53, and microsatellite instability. Moreover, the novel DNA tests comprise epigenetic analysis of methylated genes for vimentin, secreted frizzled-related protein 2 (*SFRP2*), bone morphogenetic protein 3 (*BMP3*), N-Myc downstream-regulated gene 4 protein (*NDRG4*), and tissue factor pathway inhibitor 2 (*TFPI2*), using as analytical matrix feces or venous blood. Another promising target for early diagnosis and screening are fecal miRNAs [129–138].

A totally disruptive approach to FOBT is an ingestible micro-bio-electronic device (IMBED) based on environmentally resilient biosensor bacteria for in situ biomolecular detection, coupled with miniaturized luminescence readout electronics that wirelessly communicate with an external device. According to the authors, gut biomolecular monitoring could be more precise and faster than any other laboratory methods [139].

Another biomarker relevant in GI tract neoplasms is sarcosine. It is not only associated with CRC and stomach cancer, but also with prostate cancer and neurodegenerative disorders. Analytically it is relevant to human pathology in the food and fermentation industry. Many biosensor-based approaches have been tried to measure this analite: amperometric biosensors, potentiometric sarcosine biosensors, impedimetric sarcosine biosensors, photoelectrochemical (PEC) biosensors, and immunobiosensors. All these methods have been recently reviewed by Pundir at al. [140].

In the infectious disease market, many different tests received the Clinical Laboratory Improvement Amendments (CLIA) waivers that enable POC use [141]. Most POC rapid tests use lateral flow immunoassay (LFIA) technology, a limited number of POC diagnostics utilize molecular approaches. One of the most interesting molecular methods is nicking enzyme amplification reaction (NEAR) [142], an isothermal nucleic acid amplification. Back in 2015, the FDA approved the Alere i influenza A & B test [143] based on NEAR technology, which is an isothermal DNA amplification technique. Later also a test for group A Streptococcus (GAS) that uses throat swabs as samples was CLIA waived [144], and one for respiratory syncytial virus (RSV) [145].

Recently some reviews summarized monographically the state of the art in POC testing of common conditions. Kozel and collaborators [141] provided data for cryptococcal antigen meningitis and malaria. Grebely and colleagues in 2017 reviewed the offer for HCV (Hepatitis C Virus) POC testing [146]; Gaydos et al. *Trichomonas vaginalis* [147]; Hurt et al. HIV (Human Immunodeficiency Virus) [148]; Kelly and his group *Chlamydia trachomatis* [149]; Basile and collaborators in 2018 reviewed POC testing for respiratory viruses [150]; and Nzulu and colleagues in 2019 the one for *Leishmania* [151].

Concerning other experimental techniques in infection diagnostics and specific biosensor approaches, refer to the work by Datta et al. [152].

There are two other major chapters in infectious disease relevant for this review, apart from pure diagnosis: antibiotic susceptibility testing [153,154] and sepsis early diagnosis [155–158].


**Table 1.** Clinical application in oncology.

**Table 2.** Clinical application in infectious diseases.


#### **7. Conclusions**

The size of the global market for sensors is expected to increase in the near future due to the (i) growing demand for devices that meet the needs of early and reliable analysis of diseases and (ii) the fast pace at which technology is developing, providing products that satisfy those needs, with the additional characteristics of low energy consumption, eye-catching design, and ease of use. Here, we have reported on the most up-to-date nanotechnology prototypes in the field of sensor devices. Despite the frontier technology many of these devices possess, their functionality is often compromised by a lack of care and strategy in the pre-analytical and device-maintenance phases. In this context it is important, for example, to identify the optimal operation-interval of the sensor and tune this interval such that it matches with the range of values of the biological sample signal that are clinically relevant. Such a range, in turn, depends on the characteristics of the biological sample, or matrix (tissue position and blood/urine collection), and on the degree of advancement of the disease. This is to say that the optimization and use of a device in biomedicine is not simple and requires cooperation across disciplines and a multidisciplinary approach to ensure the right technology is adopted in the right conditions and at the right time. The many interesting contributions to biomedical nanoelectronics that we have reviewed in this paper are examples of how technology has developed to meet the needs of medicine. The nature of the problems that a new technology has to face is bifold: on one side, it has to solve specific scientific problems, and on the other side it has to adapt such a solution at the interface with medicine. Thus, it is likely that the evolution of the field of nanoelectronics in the near future will be guided by the definition of new problems in medicine.

**Author Contributions:** Conceptualization, N.M. and F.G.; writing—original draft preparation, N.M., F.G., M.L.C., S.A.P., M.F.M.V., N.C.; writing—review and editing, S.A.P., N.C., M.L.C.; visualization, P.C., G.P.; supervision, F.G. and N.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Micromachines* Editorial Office E-mail: micromachines@mdpi.com www.mdpi.com/journal/micromachines