Recent Advances in Touch Sensors for Flexible Displays

A touch screen that combines a display and a touch sensor array is a critical component enabling human-machine interaction. The progress made in flexible touch screen technologies also vigorously drives the development and application of flexible electronics in various fields. Over the past decade, there have been enormous research and development efforts on new structures and materials for touch sensors in flexible displays, especially for flexible organic light-emitting diode (OLED) displays. Herein, this review discusses the mechanics and structures of flexible touch screens, including their benefits and drawbacks. The recent advances in the structures and electrode materials (e.g., ITO, silver nanowires, metal mesh, graphene, carbon nanotubes, and conductive polymers) are reviewed, and the challenges and prospects of these technologies are also explored.

In recent years, the rapidly growing flexible electronics have shown considerable potential in various industries, including information technology, energy, military, and healthcare, due to their flexibility, lightweight, mass production capability, and low cost. Among many electronic devices, the display, as the visual output device, has also been made flexible to enable various new intelligent devices, including foldable smartphones and wearables [1], [2]. Flexible displays have numerous advantages over their rigid counterparts: thin and bendable form factor, durability, and lightweight [3]. For example, a device with a flexible display can be folded or rolled up into a more compact size. Touch technology is critical for fulfilling the needs of the human-machine interaction experience on these devices, transforming the display from one-way output into two-way communication.
Flexible and transparent touch sensors are essential input devices for flexible displays. Typically, a touch panel consisting of touch sensors can be stacked on top of a display. Users can provide inputs by touching the panel with finger(s) and/or a specific stylus [4]. Instead of utilizing a mouse, touchpad, or other similar input devices, the user can interact directly with what has been presented by using the touch screen. Touch screens are widely used in various devices, such as mobile phones, tablets, laptops, game consoles, and information kiosks in malls, which provide users with intuitive, rapid, and accurate interactions.
Touch sensing can be accomplished in various mechanisms, commonly by capacitive, resistive, infrared, and surface acoustic wave technologies [5]. For rigid electronics, the well-established capacitive touch sensors consist of two electrode layers and one dielectric layer sandwiched inside. Currently, flexible touch screens are also predominantly capacitive. However, there are new requirements for the structure and materials of flexible touch sensors.
An Out-Cell structure is used in conventional touch screens, where separate touch and display modules are integrated by lamination. Although this well-established technique is low-cost and reliable, it is not suitable for flexible displays, as it increases the thickness and weight of the display and thus reduces flexibility. Therefore, integrated touch sensors have been proposed, where part or the whole of the touch sensor is integrated into the display module, thereby reducing the thickness of the entire device. Approaches for integrated touch include On-Cell, In-Cell, and Hybrid-Cell structures. Embedding a touch sensor in the display unit not only reduces the thickness of the module but also improves the optical performance of the display and enables higher touch accuracy [6].
Flexible transparent conductive electrodes are essential parts of the flexible touch display, which dominate the touch sensors' optical characteristics, lifetime, stability, and cost [7]. The conventional electrode material is primarily indium tin oxide (ITO) coated on flexible substrates. However, the shortcomings of ITO, such as shortage of indium, hightemperature deposition process, and brittleness, severely limit its widespread application in next-generation flexible electronic devices [8]. Therefore, various alternative materials have been used as flexible transparent electrodes, including carbon nanotubes [9], graphene [10], metal nanowires [11], metal meshes [12], conductive polymers [13], etc. These materials are more suitable due to their excellent optical transmittance, low electrical resistance, and high flexibility.
This paper provides a comprehensive review of these structures and materials. The paper is organized as follows: firstly, the various mechanisms of the touch screen are introduced. Then, Section III discusses the structures for flexible touch screens. Next, advances in transparent electrode materials are reviewed in Section IV. The outline of the paper is shown in Fig. 1.

II. TOUCH SENSING MECHANISMS
The capacitive touch sensing technology is widely used in various applications, such as phones, watches, and PCs, due to its high sensitivity and multi-touch capability [14]. Fig. 2 illustrates the working mechanisms of capacitive touch sensors. A capacitive touch sensor consists of one or more layers of transparent conductors deposited on the glass substrate by sputtering or evaporating and patterning (by lithography and etching). The capacitor becomes conductive when an alternating current (AC) is supplied to the conductive electrode. When a human finger or a stylus is close to the sensor, the electric field is distorted, which can be detected as a change in capacitance. The location of the touch may be determined by the indices of the electrodes, which have a capacitance change. The data are further processed by a controller IC.
Capacitive sensing technologies can, in detail, be categorized into surface capacitive touch sensing technology [15] and projected capacitive touch sensing technology. Furthermore, there are two subcategories for the projected capacitive touch sensing: mutual capacitive touch and self-capacitive touch.

A. SURFACE CAPACITIVE TOUCH SENSING TECHNOLOGY
The surface capacitive sensor [16] consists of an insulating layer with a conductive layer only on one side. Each of the four corners is connected to four synchronized AC voltage signals. An electric field is produced on the touch panel surface through the four corners by the AC, as shown in Fig. 2(a). By detecting the capacitance change on each corner, the location of the touch can be determined. The relationship between the supplementary current and the distances can be expressed as: I A / L 1 = I D / L 2 and I D / L 3 = I C / L 4 , where L 1 and L 2 are the distance between the touch point and corners A and D in the horizontal direction, respectively; while L 3 and L 4 are the distances between the touch point and corners D and C in the vertical direction, respectively. I A , I D , and I C represent the supplementary current from corners A, D, and C. The coordinates of the contact point can be determined by L 1 -L 4 [17]. Surface capacitive touch sensors have several advantages over conventional resistive touch sensors, including hard surface, high sensitivity, extended lifetime, and high optical transmittance. However, they also have several drawbacks, including limited resolution and incompatibility with multi-touch. Currently, it is primarily utilized in large-scale interactive systems such as touch screen kiosks.

B. PROJECTED CAPACITIVE TOUCH SENSING TECHNOLOGY
Projected capacitive technology [18] is usually implemented by one or multiple layers of patterned electrode arrays, in contrast to the single conductor electrode of the surface capacitive technology. As shown in Fig. 2(b), (c), the projected capacitive touch technology [19] comprises patterned electrodes. When the sensor is touched, the electric field is altered, and thus the capacitance of the electrode itself or the capacitance between adjacent electrodes changes.
The location of the touch point can be determined by the indices of the electrodes where capacitances alter. The projected capacitive touch sensing technology supports multi-touch, i.e., a two-point touch or pinch, making human-computer interaction more practical and convenient.
Further, projected capacitive touch technology can be classified into self-capacitive and mutual capacitive touch sensing technologies based on their detecting mechanisms.

1) SELF-CAPACITIVE TOUCH SENSING TECHNOLOGY
Self-capacitance, also known as absolute capacitance, is the capacitance between the electrodes and the ground.
One type of self-capacitance structure is shown in Fig. 3(a). The sensor is an X-Y grid consisting of row and column traces. Each column or row trace forms a self-capacitance. The location of the touch is determined by the indices of the row and column traces with altered self-capacitances. The self-capacitive sensing technique [20] has high scanning frequency and good noise immunity. However, it cannot enable accurate multi-touch sensing, which leads to "ghosting," i.e., ambiguous location sensing.
Another type of self-capacitive sensing technique [21] is illustrated in Fig. 3(b). In this technique, each electrode in the array is connected to the driving IC by an individual wire so that multiple contact points can be detected separately, thus solving the ghost point issue by only a single electrode layer. However, separate electrodes require a large number of wires for connection, which consumes a large area, leaving less area for the electrodes. Therefore, in order to avoid such blind areas, it is often necessary to utilize a separate conductive layer for wires beneath the electrode layer. The other issue is that too many input/out ports are needed for the driving IC, especially when the panel size scales up.

2) MUTUAL CAPACITIVE TOUCH SENSING TECHNOLOGY
Different from self-capacitive touch sensing technology, in mutual capacitive touch, two adjacent electrodes are coupled by the fringing electric field between them. When the finger touches the panel surface, it affects the coupling between the two electrodes around the touch point, thereby reducing the capacitance between them. The touch location is determined by the indices of the two adjacent electrodes [22]. As Fig. 3(c) shows, the longitudinal row electrodes are set as transmitter electrodes (Tx), while the latitudinal electrodes are used as receiver electrodes (Rx). AC signal is applied by the controller IC to each Tx in order, and the IC detects the response from all the Rx's. The coordinates (X, Y) of the touch can be determined by the corresponding Tx and Rx pair that encounter a suppressed response. Compared with self-capacitance touch sensing technology, mutual capacitive touch sensing technology has a longer scanning time and consumes more power, but multi-touch sensing can be achieved.
Typically, a layer of transparent electrodes is employed, which serves as the electrode Tx for transmitting touch signals and the electrode Rx for receiving calls. Between several groupings of Tx/Rx signals, a set of isolated signals must be introduced.
Currently, most touch screens used on flexible displays are capacitive. There are also other uncommon techniques for flexible display touch panels that are under development. In 2015, Lee et al. [23] suggested a highly sensitive nearinfrared a-Si:H phototransistor for touch sensor applications. The device has a simple, easy-to-assemble structure, where the infrared sensor is only located on the narrow border and does not cover the display. Although this technique has less impact on the visual effect, an unavoidable delay arises due to the complicated touch detection. The technique is also sensitive to ambient light and requires a consistent working environment. In addition, the cost is high due to the separate circuit board.

A. OUT-CELL STRUCTURE
As shown in Fig. 4(a), a typical flexible touch screen (i.e., active-matrix organic light-emitting diode, AMOLED) consists of a film cover, an independent touch sensor panel, a polarizer film, and a display module assembled by adhesive lamination. There are several commonly used structures for Out-Cell sensing: Film-Film, Film2, Double-side ITO, and Single-side ITO/bridge touch structure [24]. Up to now, the Out-Cell screen has a simple structure and mature technology [25], but more is needed to meet the demand in bendability and thickness.

B. INTEGRATED TOUCH SENSORS
For further lowering the cost, reducing the weight and thickness of the module, and enhancing flexibility, integrated touch sensor technology has attracted more research interests [26]. As shown in Fig. 4(b)-(d), the integrated touch screen structure can be classified into On-Cell, In-Cell, and Hybrid-cell structures based on their stacking orders.

1) ON-CELL TOUCH
On-Cell touch is a specific technology for integrating touch sensors on an active-matrix organic-light-emittingdiode (AMOLED) display [27]. In the AMOLED, as illustrated by Fig. 4(b), the touch sensor is positioned above the OLED encapsulation layer. Longitude and latitude electrodes can be placed on two different conductor layers sandwiching a dielectric layer or simply on the same conductor layer, while the intersection regions are bridged by another conductor layer [28].
Hsieh et al. [29] integrated an On-Cell touch sensor on an IGZO-driven AMOLED and reported a 45% high transparency and 166-PPI high resolution. Hu et al. [30] proposed a new circuit model of the On-Cell touch sensor for AMOLED to model floating ground and noise interference issues. On-Cell structure features easy-to-use, wide display area, and high display quality. Nevertheless, it has limited visible angles and is easily disturbed by external signals.

2) IN-CELL TOUCH
In-Cell touch is a touch technology that integrates touch sensors into the pixels. The In-Cell design represents the structure in which the touch sensor is built into the liquid crystal panel in the LCD or under the encapsulation layer in the OLED (Fig. 4(c)). The touch sensor patterns are usually integrated with the emitting pixels. Integrating extra electrodes in OLED cells is more challenging for manufacturing. Moreover, more complicated driving circuits are needed for the touch and display to avoid interference. However, for In-Cell, the common electrode VCOM is usually divided into blocks and re-used as touch electrodes, where each block is connected to a touch detection signal line, thus reducing the number of conductor layers. This high degree of integration design enables the thinnest projected capacitive touch screen and high transmittance.
Therefore, further research is still needed to reduce parasitic capacitance, noise interference, and other problems. Su et al. [31] employed an organic double-layer structure in 2021 to lower the parasitic capacitance of the electrodes and demonstrated a 10.95-inch In-Cell touch screen with an improved frame rate of 120 Hz. The same year, Shen [32]. et al. suggested a gate drive circuit with a two-stage precharge design to address the voltage drift issue in the touch detection stage. This circuit significantly suppressed threshold voltage drift and improved stability

3) HYBRID-CELL TOUCH
Hybrid-Cell is a touch-sensing technology that combines On-Cell and In-Cell. In the Hybrid-Cell structure, the Tx electrodes are integrated within the display module, while the Rx is deposited and patterned on the surface of the OLED encapsulation, as shown in Fig. 4(d). Tx electrodes are usually designed on the common electrode layer (VCOM), and Furthermore, the hybrid-cell screen is unsuitable for narrow border designs since the Tx needs to be connected to the IC by additional fanout regions on the left and right borders.
A summary of the capacitive touch sensor structures is shown in Table 1. Conventional Out-Cell touch sensing is mature but is limited in module thickness and image quality. Among the integrated touch sensing techniques, In-Cell has the best performance but is difficult to manufacture, while On-Cell is overall more balanced.

IV. TRANSPARENT ELECTRODE MATERIALS
Transparent conducting electrodes are essential for numerous flexible optoelectronic devices, including touch screens and interactive electronics. Especially for flexible touch sensors, the sensors are required to be highly conductive while as transparent as possible for the display beneath to be visible [33]. For On-Cell, the material used for touch sensors is also required to be feasible for low-temperature processes directly on top of the display panel [34]. Furthermore, considering manufacturability, the materials that can be produced on a large scale at low cost are also favorable.
Currently, among the flexible transparent electrode materials, silver nanowires (AgNWs) [35] and metal mesh [36] have already been used to construct flexible capacitive touch sensors. There are other potential materials for capacitive touches, such as carbon nanotubes [37], graphene [38], and conductive polymer [39].

A. INDIUM TIN OXIDE
Indium tin oxide (ITO) is a deeply doped, highly degenerated n-type semiconductor material with a large energy gap (3.5 -4.3 eV) and favorable optical and electrical characteristics, such as high transmittance, low resistivity, good abrasion resistance, and chemical resistance [40]. Currently, ITO is the predominant transparent conductive electrode material, and it is typically coated on substrates made of stiff glass or flexible plastic by a vapor phase deposition. ITO, however, is currently facing several issues [41] that need to be addressed: (1) Low indium supply leading to high cost; (2) expensive deposition procedure with vacuum deposition equipment; (3) as a metal oxide, brittle and unsuitable for flexible and wearable technologies. ITO films are bent together with the flexible display, and it is easy to break and detach from the substrate, especially in the areas where the stress is more concentrated, such as the edges of the film and the regions with growth defects. Currently, a lot of research efforts have been made to improve the performances of ITO electrodes for flexible touch displays.
Song et al. [42] proposed an epoxy-copper-ITO (ECI) sandwich-structure transparent electrodes, which greatly improved the mechanical properties of ITO electrodes. The film sheet resistance is 50 /sq at 50 nm ITO thickness and is in good balance with transmittance (90% at 600nm). Such work provides a new possibility for the application of ITO in flexible devices. Park et al. [43] found that the crystallinity of ITO significantly affects its mechanical properties and deposited ITO films (sheet resistance of 36 /sq, light transmittance of 88%) at 250°C and transferred them to cyclic olefin polymer films, which showed good stability in repeated bending tests. Kim et al. [44] reported a high-performance flexible ITO film ( Fig. 5(a)) prepared by in-line vertical plasma arc ion plating. The films have a low sheet resistance of 15.75 /sq, a high average optical transmittance of 85.88%, and a small bending radius of 5 mm, because the ionized ITO is driven by the system energy to the substrate led to greater crystallinity and adhesion.

B. SILVER NANOWIRES
Silver nanowires (AgNWs) have attracted significant attention due to their excellent conductivity, high transmittance, and flexibility [45]. In addition, silver is a noble metal, which shows good mechanical strength and high chemical stability in harsh environments. The electrical conductivity and optical transmittance of these AgNWs are comparable with or even better than those of ITO films. The major challenge for the practical application of AgNWs in transparent conductor films is the uniformity of the sheet resistance. Therefore, much effort has been devoted to improving the uniformity of AgNW films.
Herein, Jia et al. [46] presented a dynamic heating technique employing infrared light to produce excellent uniform AgNWs. In the as-prepared films, AgNWs are prevented from adhering to the surfaces, and thus the coffee ring effect is reduced after drying. As shown in Fig. 5(b), the sheet resistance non-uniformity of the film is only 6.7%. The average sheet resistance of 35 /sq and the wavelength transmittance of 95% at 550 nm is comparable to commercialized high-quality ITO films. Fig. 5(c) displays a demo of a writing pad made from this film. The material has passed a mechanical bending test of over 5000 bending cycles with minimal change in resistance.
Choi et al. [47] embedded electrode wires made of silver nanowires and reduced graphene oxide (AgNWs/rGO) into a polyurethane (PU) dielectric layer on a polydimethylsiloxane (PDMS) substrate. An array of 5 × 5 stretchable transparent capacitive touch sensors based on patterned AgNWs/rGO wires embedded in PU dielectrics on PDMS substrates was demonstrated. Fig. 5(d) shows the capacitance of an electrode vs. strain, with or without touch. It is shown that neither capacitance changes significantly, demonstrating that the device

FIGURE 5. (a) Photo of the flexible MAPBI3 perovskite solar cell with ion-plated ITO films [44] (Copyright © 2018, RSC Pub). (b) Histograms of sheet resistance of prepared silver nanowire films reported in [46]. (c) Demonstration for writing input using the uniform AgNW transparent conductive films [46] (Copyright © 2016, American Chemical Society). (d) Capacitance change in a single capacitive touch sensor under stretching strain with (red) and without (black) touching [47] (Copyright © 2017, American Chemical Society). (e) Multitouch and writing ("SYSU") demonstrated on the touch panel made by AgNW [48] (Copyright © 2016, American Chemical Society). (f) Demonstration of a bendable touch screen on a non-flat surface, where letters "ZJU" were written [49]. (Copyright © 2019 IOP Publishing Ltd).
works steadily even when stretched. Direct fabrication of thinfilm electrodes on elastic substrates opens many possibilities for transparent, flexible capacitive touch sensors.
To increase the utilization of silver nanowires, avoiding wasting material during the preparation process is also necessary. Liu et al. [48] provide an efficient and universal transferring method to fabricate extraordinarily stable and high-performance AgNW transparent electrodes on any substrate. A large-area transfer of 70 × 70 mm 2 was achieved, and a multi-touch panel made by AgNWs was demonstrated with "SYSU" written on the panel, as shown in Fig. 5(e).
Using the Mayer rod coating process, Yang et al. [49] of Zhejiang University fabricated a large-scale AgNW-poly(34ethylene dioxythiophene): poly (styrene sulfonate) composite film with a sheet resistance of 12 /sq and a transmittance of 96% at 550 nm. Additionally, they made a 7 × 7 cm 2 transparent touch panel with excellent touch sensitivity and uniformity throughout the entire surface (Fig. 5(f)).
Nevertheless, AgNW transparent conductive films have several drawbacks that limit their commercialization, including rough surfaces, difficulty in mass production, and patentability, including precision printing [50].

C. METAL MESH
Metal mesh, i. e. a micrometer-scale grid structure made of intersecting metal wires, can be utilized in flexible electronics because of its low resistance, high optical transmittance, and superior bending ability. Metal meshes for flexible screens are more promising than silver nanowires since they are more flexible, and their performances are tunable by altering the width and thickness of lines. It can also be fabricated on a large scale at a low cost and has already been applied in commercialized flexible touch screens. However, there are some limitations to the use of metal meshes. For instance, the mesh's uneven surface reduces the adhesion between flexible modules, and the mesh's excessive density of moiré patterns affects the visual effect of the display.
With a process involving electrospinning and metal deposition, Wu et al. [51] demonstrated a new transparent conductive electrode, which exhibits superior performances (sheet resistance of ∼2 /sq, at 90% transmission). The team solved the adhesion issue by employing electrospinning and a gold nano-groove grid to the PET substrate. Subsequently, a transparent touch screen device was implemented based on such films. Such metal nano-groove electrodes could replace the frequently used ITO in solar cells, touch sensors, and flat panel displays and could broaden the application fields, including skin-like sensors.
Zou et al. [52] suggested coating the metal mesh with a passivation layer to reduce the issue of metal surface roughness. For instance, the metal mesh/conductive polymer hybrid transparent electrode is used to fabricate the inverted-structure polymer solar cell. The optical microscope image of an asfabricated metal mesh electrode on a glass substrate is shown in Fig. 6(a). Additionally, the efficiency of the polymer solar cells is increased by 3.2%.
In terms of other works, Zhou et al. [53] made a flexible transparent plastic conductor embedded with a silver network

(b) Electrical resistance testing of the PEAN using an ohmmeter [53] (Copyright © 2016, American Chemical Society). (c) Images of the 8.67-in foldable organic light-emitting diode display with a touch sensor [54] (Copyright 2016 © Society for Information Display). (d) The 5 × 5 arrays of the 3DGF/CNT networked strain sensor [58] (Copyright 2017 © WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim). (e) Photograph of the graphene touch screen installed in a mobile phone (left) in comparison with an ITO-based touch screen phone (right) [59] (Copyright © 2014, American Chemical Society). (f) Unidym touch panel integrated with a full-color LCD [63] (Copyright © 2009 Society for Information Display).
(PEAN). The prepared metal mesh is shown in Fig. 6(b). They used an optimized optocoupler structure to lower the ohmic loss and raise the luminosity of white and green OLEDs. The OLED's power efficiency at 1000 cd/m 2 is higher than 120 mW -1 , and its current efficiency is more than 140 cd/A. White light has a maximum quantum efficiency of 49%. The performance of flexible OLEDs with PEAN is comparable to that of ITO devices, and they are anticipated to be used in large-area non-ITO flexible displays and lighting.
Watanabe et al. [54] fabricated an 8.67-inch foldable OLED display with a touch sensor, in which the touch sensor has an In-Cell structure where metal-mesh sensor electrodes are formed in a counter substrate, shown in Fig. 6(c). To minimize the negative impact on light extraction, the electrode area was solely present outside the pixel aperture. The sensor's parasitic capacitance is only 910 pF, while its parasitic resistance is roughly 1.3 k .
The metal mesh is nonetheless constrained for making flexible touch display devices due to the following issues [55]: (1) When the metal mesh pitch is too dense, it will block the pixel cells and cause moiré patterns and reflection on a black image visible from a particular angle; (2) the expensive process of vacuum metal deposition; (3) small adhesion between the metal mesh and the modules on top, as well as the uneven surface topography, etc.

D. GRAPHENE
Graphene is a two-dimensional carbon nanomaterial with a hexagonal honeycomb lattice made of carbon atoms with sp 2 -hybridized orbitals. Each carbon atom exposed to the environment can function as an active site sensing stimulation since the ideal graphene comprises a single layer of carbon atoms. Graphene is well suited for the fabrication of flexible sensors due to its enormous specific surface area (2360 m 2 /g), high electron mobility (200000 cm 2 /(Vs)), strong conductivity, lightweight, good mechanical flexibility, and compatibility with large-area flexible solid supports [56].
However, the absence of effective methods for synthesizing, transferring, and doping graphene at the scale and quality necessary for applications has stymied the attempts to produce transparent conducting films from graphene.
Savchak et al. [57] encapsulated graphene oxide with polymers, which were subsequently transformed into highly conductive and transparent reduced graphene oxide films by dip-coating in water and thermal reduction, attaining conductivity as high as 10 4 S/cm and optical transmittance of 90%. Cai et al. [58] built a stable and noise-free strain sensor with stretchability by integrating 3D graphene foam and fusing two-step chemical-vapor-deposition-produced carbon nanotubes with a measurement factor of up to 35, a reliable sensing range of up to 85% and good cycling stability (>5000 cycles). Fig. 6(d) depicts a 5 × 5 array of flexible touch sensors generated with a 3DGF/CNT percolation network and then sealed with elastic PDMS. Ryu et al. [59] used hydrogenfree fast chemical vapor deposition, roll-to-roll etching, and batch transfer procedures to create graphene films with an area of more than 400 × 300 mm 2 and a sheet resistance of 249 ± 17 / sq. They also designed a capacitive multi-touch screen. Fig. 6(e) depicts the use of this film on a touch-sensitive mobile phone. Bae et al. [60] used roll-to-roll fabrication and a wet chemical doping approach to create 30-in graphene films on flexible copper substrates. Their optical transmittance was 90%, and their thin film sheet resistance was as low as 30 /sq. It is also built into a fully functional touch panel gadget.
Since the challenges with size, uniformity, and reliability have yet to be resolved to meet industrial standards, there has been no concrete evidence of the usage of graphene in consumer devices.

E. CARBON NANOTUBES
Carbon nanotube (CNT) [61] is a one-dimensional carbon allotropy with a length-to-diameter ratio, or aspect ratio, above 1000. A single CNT has mobility of more than 100000 cm 2 /s and a current carrying capacity of 10 9 A/cm 2 . Although carbon nanotubes have strong electrical and optical properties, preparing large-scale, high-purity, and low-cost CNTs remains challenging. The wet-pulling technique and solution-based spin-coating [62] are frequently used to prepare homogenous CNT at low cost.
Hecht et al. [63] employed a spin coating process using Unidym's CNT solution to build a CNT flexible transparent electrode film with optical transmittance of 88% and sheet resistance of 550 /sq, which was then used to build a four-wire resistive touch panel. The films were integrated into a slightly smaller touch panel with a full-color LCD, as illustrated in Fig. 6(f). Zhang et al. [64] used a solution approach to prepare a self-assembled silver nanoparticle/multi-walled CNT composite thin film with a low layer sheet resistance of 14.5 /sq, an average visible transmittance (AVT) of ≈67%, and a high color rendering index (CRI) of 97. Solar cell device with this solution-processed transparent electrode achieves an AVT of 36% and a high CRI of 90. Choi et al. [65] prepared a flexible transparent electrode film by dispersing single-wall CNTs in deionized water and sodium dodecyl benzene sulfonate and fabricated a five-wire resistive CNT touch sensor.
Although CNT is a high-performance transparent electrode material capable of solution process, the complicated separation process remains an issue for the practical application of CNT. Therefore, more research effort is needed for CNT materials to be implemented in flexible capacitive touch sensors.

F. CONDUCTIVE POLYMERS
Among the transparent conductive materials of various polymers, poly(34-ethylene dioxythiophene): poly(styrene sulfonate) (PEDOT: PSS) appears as an attractive candidate because of its high transmittance, acceptable conductivity, excellent flexibility, and increased work function [66]. Numerous optoelectronic devices based on PEDOT: PSS anodes, such as organic light-emitting diodes (OLEDs), organic solar cells, and supercapacitors, have exhibited very encouraging photoelectric characteristics [67].The excellent optoelectronic capabilities of OLEDs based on stretchable PEDOT: PSS grid anodes suggest they have tremendous potential as ITO-free anodes for flexible and wearable electronic applications.
Although outstanding photoelectric characteristics have been attained, most existing manufacturing procedures are based on spin coating. This typically results in substantial processing losses that significantly hampers overall mass production. Therefore, exploring low-cost solution processing approaches with high scalability, continuous production, large area, and minimal material waste is highly desirable.
Zhou et al. [68] used roll-to-roll compatible screen printing to create PEDOT: PSS mesh-type transparent conductive electrodes on PET with an optical transmittance of 80% and a square resistance of 450 /sq. Fig. 7(a) shows a light-on image of the device. This is the first study to explore screenprinted PEDOT: PSS mesh anodes that allow for the rapid and efficient construction of anode clusters. Fig. 7(b) shows that Kim et al. [69] improved the conductivity of the PEDOT: PSS film to 3000 S/cm by treating the surface with H 2 SO 4 and performing a transfer process, allowing for the practical integration of any flexible substrate or electronic component, which was demonstrated by high-performance OLED displays consisting of this film. These two explorations show that PEDOT: PSS can practically be manufactured on a large scale at a low cost.
However, performance deterioration of PEDOT: PSS films are likely to happen on plastic substrates and thus flexible optoelectronic devices due to the presence of acidic solid residues in the PEDOT: PSS matrix. To facilitate the problem of solid acid residues, Fan et al. [70] used gentler acids such as methane sulfonic acid and phosphoric acid. As a result, they achieved significantly increased conductivity (3500 S/cm, which was applied to OLEDs to achieve 84% optical transmission and considerably improved stability (Fig. 7(c)).
Understanding additives can help improve the stability of flexible electronics used in medical, energy, and other applications. For example, Savagatrup et al. [71] developed wearable strain sensors (Fig. 7(d)) with detectable responses under 20% strain using PEDOT: PSS sheets containing 5% dimethyl sulfoxide and 10% Zonyl fluorosurfactant.
Overall, few touch sensors use conductive polymer, and further development is needed.
At the end of this section, the transmittance and sheet resistance of the transparent electrodes reported by selected works mentioned in this review are summarized in Fig. 8. It can be seen that the properties of AgNWs and metal meshes are already comparable or even better than those of optimal ITO films, while graphene and carbon nanotube materials still need improvements. Additional work on the conductive polymer is not plotted on the graph as their performance is far inferior to ITO.

V. CONCLUSION AND OUTLOOK
The rapid advancement of flexible electronics accelerates the evolution of flexible screens toward customized applications in various scenarios. Research efforts in sensing, manufacturing and display technologies have all been made to achieve that goal. We expect flexible touch screen technologies to dramatically change how human beings interact with everything in the new era of the Internet of Things (IoT) and Industry 4.0.
This study summarized recent research on touch sensors for flexible displays in the mechanisms, structure, and electrode materials. To guide future research on flexible displays, the challenges and prospects of touch sensors in terms of structure and module thickness, cost and manufacture feasibility, and display quality are explored.

A. STRUCTURE AND MODULE THICKNESS
The Out-Cell structure screen has conventionally been integrated as a separate touch module on top of the display module. These two modules are frequently laminated by optically clear adhesive, resulting in a relatively high overall thickness. The integrated touch sensor technologies, including In-Cell, On-Cell, and hybrid-Cell structures, are presented as a solution to this issue. Because the touch module is partially or entirely integrated with the display module, the overall module thickness is significantly decreased. This is also a trend for flexible touch screens, and more smart devices with integrated touch sensors will appear in the future.
In-Cell is the optimal solution considering visual effects and module thickness. However, In-Cell needs to be integrated in the pixel circuit and requires a specified driving IC that coordinates the display and the touch. Furthermore, in some designs, the OLED VCOM cathode is patterned and re-used as In-Cell touch electrodes, leading to a more difficult OLED evaporation process and lower yield. As a result, most of the integrated touch for flexible AMOLEDs on the market is On-Cell, which may continue to dominate the flexible display market.

B. COST AND MANUFACTURE FEASIBILITY
ITO is a rare and brittle material historically utilized as an electrode. New electrode materials have enabled more possibilities for flexible displays. The market share of AgNWs and metal mesh technologies is growing continuously. However, ITO may not be replaced by new materials in a short time. Because these transparent electrodes still have some manufacturing issues, like surface roughness and adhesion issue.
Graphene is currently in research and development, but mass production is still far away. Likewise, industrialized mass manufacture has yet to be achieved for CNTs, too. Moreover, these thin films' electrical conductivity cannot match regular ITO thin films. Therefore, AgNWs and metal meshes are the most promising materials for touch sensors in flexible displays based on their technology readiness and market applicability.

C. DISPLAY QUALITY
Silver, copper, oxides, and other widely accessible and reasonably priced raw materials can be used to build the metal mesh. However, the metal mesh line width of the currently generated touch patterns is frequently larger than 5 μm due to technological limitations. In addition, with high pixel density, the moiré interference ripples will be highly noticeable. Therefore, metal mesh may become inappropriate for extremely high-resolution displays unless the line width and resistance issues can be resolved.
Using technologies like laser-assisted printing [72] can deposit AgNW patterns on flexible substrates with line width down to 50 nm, which can be utilized on displays of different sizes and pixel densities without exhibiting a moiré pattern. AgNWs also have a smaller bending radius than metal mesh films and a low resistance variation when bent. However, due to the random distribution of AgNWs, their films suffer from severe diffuse reflection, which leads to a haze effect and needs to be resolved.