Flexible Cellulose-Based Assembled with PEDOT: PSS Electrodes for ECG Monitoring

Electrocardiography is one of the most signicant technologies for detecting cardiovascular diseases. Nowadays, the problems of various electrodes still meet a great challenge. Herein, we design a low cost, environmentally friendly and exible conductive electrode using cellulose and polyvinyl alcohol as a substrate assembled with conductive polymer polythiophene by in-situ oxidative polymerization, and the green solvent 1-butyl-3-methylimidazolium chloride as a crosslinking agent. The polyvinyl alcohol/cellulose/PEDOT:PSS(PCPP) composite electrode has excellent features of exibility, low skin contact impedance and comfortable contact with skin. When the load of EDOT reaches 15 wt%, the electrode is stable and can clearly monitor the characteristic wave of ECG signals. Therefore, based on cellulosic biopolymer and conductive polymer PEDOT:PSS, an environmentally friendly, exible and stable PCPP composite electrode is obtained and can be a promising candidate applied in the elds of energy storage and ECG sensing.


Introduction
With the aging of the population and the prevalence of unhealthy lifestyles, cardiovascular and related diseases show an obvious growing trend nowadays(Akter Shathi et al. 2020a). Electrocardiogram (ECG) is one of the most important technologies for monitoring physiological function (Acar et al. 2019), based on the monitor many bioelectric signals containing important information of the human body, have been widely studied in medicine and scienti c elds (Ankhili et al. 2018). By collecting the bioelectrical signals with electrodes on the surface of the body (Yamamoto et al. 2017), it is not only a clinical diagnosis of various heart diseases but also a simple and effective tool to explain arrhythmia and conduction disorders (Yapici and Alkhidir 2017). The mechanism of these electrodes is the depolarization and repolarization of myocardial cells will generate electric current. After this current is extracted, analyzed and processed by the instrument placed on the body surface, the dynamic curve of ECG signals can be obtained (Kaplan Berkaya et al. 2018). Obviously, the electrode is a very critical part of collecting bioelectric signals (Acar et al. 2019).
Electrodes mainly include wet electrodes, dry electrodes and exible electrodes (Amale et al. 2018). The wet electrode mainly refers to the disposable electrode Ag/AgCl, which is a commercially available electrode at present. It has certain limitations that demand to be matched with a conductive gel to work properly. However, the conductive gel will cause allergic and other symptoms of the skin. Moreover, the gel will become dry in a short time, leading to the increase of noise and artifacts of the obtained electrical signals (Castrillón et al. 2018). Dry electrodes don't need conductive gel (Puurtinen et al. 2006), but need to be rmly xed on the skin with medical tape, which will also incur the uncomfortable feeling of the patients (Obukhov et al. 2012). Therefore, in order to meet the needs of long-term ECG acquisition, the design of a lightweight, exible and highly sensitive electrode is extremely important (Shathi et al. 2020).
Recent ndings have shown that cellulose is a biocompatible and low-cost electrode substrate candidate(Akter Shathi et al. 2020b). Because of its excellent exibility and mechanical stability, it has the potential to be regarded as structural support for wearable devices (Zhang et al. 2019;Zhang et al. 2020). When polythiophene (Zhao et al. 2017), polypyrrole (Abinaya and Muthuraj 2020;Chen et al. 2015;Jyothibasu and Lee 2020;Yang et al. 2020a), and polyaniline (Wang et al. 2019) are assembled with cellulose through strong interaction, to highly conductive exible electrodes can be fabricated. For example, the introduction of conducting polymer poly (3,4-ethylene dioxythiophene): Polystyrene trans acid (PEDOT:PSS) can endow the electrodes with higher sensitivity to biological and chemical molecules, shorter response time, and realizes on-line real-time monitoring (Pani et al. 2016). A transparent and biocompatible strain sensor based on silver nanowires, regenerated cellulose lm and PEDOT:PSS were prepared by coordination and hydrogen bonding (Xu et al. 2019). Using green solvents like ion liquid is a promising way to recycle cellulose from biomass (Rieland and Love 2020). Studies have found that 1butyl-3-methylimidazolium hydrochloride is able to dissolve cellulose, realizing the recycling of cotton fabrics (Swatloski et al. 2002). After dissolution, a single cellulose macromolecular chain which could be further modi ed and self-assembled was obtained (Wang et al. 2012). As known to all, polyvinyl alcohol (PVA) has a large number of hydroxyl groups on the molecular chain with good biocompatibility, stability, and the built-in moisture content is comparable to human cell tissue (Huafeng and Shaopeng 2017). It can be combined with cellulose bers due to the formation of hydrogen bonds between macromolecular chains (Sun et al. 2014). In particular, PVA can modify the stability of PEDOT:PSS electrode Liu et al. 2011). Therefore, in this paper, cotton fabric which is dissolved by the green solvent of 1-butyl-3methylimidazolium chloride ([Bmim]Cl), is mixed with PVA turning into the substrate of a exible electrode (Sellam and Hashmi 2013). Then PEDOT formed in the process of in-situ oxidation polymerizes with PSS to form polyvinyl alcohol/cellulose/PEDOT:PSS(PCPP) lm electrode. This strategy may provide new ideas for the production of low cost, environmentally friendly conductive electrodes for ECG monitoring.

Materials
In this experiment, cotton fabric was obtained from Shanghai San yuan Co. Ltd., sodium hydroxide (NaOH) was obtained from Ping Hu Chemical Reagent Factory, [Bmim]Cl was received from Shanghai Bidder Medical Technology Co. Ltd, EDOT and PSS were acquired from Shanghai Yi Shi Chemical Co.
All chemicals are analytical grade reagents and used without puri cation.

Pretreatment of cotton fabric
Cotton fabric was dipped into sodium hydroxide solution with a mass fraction of 2% for 2h at 80℃ to remove impurities and grease on the fabric surface. Then it was rinsed by water at room temperature for washing and soaking for a certain time. After ltering distilled water, the fabric was dried and sheared to pieces.

Preparation of the PVA/Cellulose/PEDOT:PSS (PCPP) lms
[Bmim]Cl (3 g) was added to an eggplant shaped ask and heated at 100°C for 10 minutes. Then cotton fabric (3 wt% 0.09g) was added to the ask and mechanically stirred until a transparent and uniformly stable IL/cellulose system was received. Then PVA (1% 0.03g) was added into the mixture system. EDOT (0.3, 0.36, 0.45g) was dissolved in solution respectively to obtain IL/PVA/CEL/EDOT solution with different content of (10 wt%, 12 wt%, 15 wt%)EDOT, reacting 30min. The stirring rate of the reaction system needed to be reduced in order to allow IL/PVA/CEL/EDOT/ to perform an adequate cross-linking reaction. When the reaction was nished, keeping 30min under 100℃ to remove bubbles. The mixture solution was immediately cast into a Te on mold while it was hot, and then the gel was cooled to room temperature, getting an IL/PVA/CEL/EDOT gel. Then the gel was soaked in distilled water, obtaining a white lm. Polystyrene sulfonic acid (PSS) was used as a dopant; the lm was then soaked in PSS solution at 35°C for 30 minutes. Ammonium persulfate (APS) and ferric sulfate (Fe 2 (SO 4 ) 3 ) were added into the solution to initiate oxidative polymerization. The oxidation time was 12 h, and the color of the solution changes from clear to black. Then unreacted PSS was removed by washing with distilled water, obtaining a PVA/CEL/PEDOT:PSS hydrogel. The hydrogel was dried in a vacuum oven at 55°C for 8 hours to obtain a black PVA/CEL/PEDOT:PSS(PCPP) lm.

Raman spectra
Raman imaging microscope (DXR2xi, Thermo Scienti c, USA) was used to analyze the chemical structure of the samples. The laser beam length was 532 nm.

Scanning electron microscope (SEM)
The morphology of pure cellulose lm, PVA/CEL lm, PVA/CEL/EDOT lm and PVA/CEL/PEDOT:PSS lm were observed by scanning electron microscopy (SEM, TM-1000, Hitachi) Before testing, the samples were dried and then gold plated on the samples using a vacuum sputtering coater.

Conductivity testing
A four-point probe was used to measure the sheet resistance of the lm. The water content in the conductive lm would affect the test results. The lm was pretreated in a constant temperature and humidity chamber for 24 h. The average resistance was calculated by taking 10 measured values, and the conductivity was calculated using the formula.
Rs is the sheet resistance of the lm and d is the thickness of the lm.

XPS spectra
X-ray photoelectron spectroscopy (XPS) was used to detect the surface element composition and atomic valence state of 15 wt% PCPP lm in the range of 0-1200ev.

Electrochemical Measurements
CV and EIS were studied by CHI 660E electrochemical analyzer (CHI 660E, Chen Hua, China). The CV scan data were obtained in a 1.0 V voltage window, using simulated sweat as the electrolyte, and the scan rate was 0.01-0.09V/s. The EIS measurement was performed in the frequency range of 0.01 Hz to 100 kHz.

ECG measurements
A standard -lead ECG measurement was performed using 15 wt% PCPP lm electrodes to evaluate the bio-signal sensing performance. The electrodes are connected to the left and right wrists of the volunteer's body and then connected to a commercial ECG device (BDM101) through lead wires to record ECG signals. The digital signal is transmitted to the PC or mobile phone through the ampli er and ADC converter.

Design and synthesis of the PVA/CEL/PEDOT:PSS(PCPP) lms
Figure1a shows the process of obtaining PVA/CEL/PEDOT:PSS conductive lm. Cellulose was crosslinked with PVA, forming a exible template to guide the in-situ polymerization of EDOT along the polymer chain. Firstly, the pretreated waste cotton fabric was dissolved into a single cellulose macromolecular chain in the [Bmim]Cl system. At the beginning of dissolution, the bers with linear and micron-level can be clearly seen under the optical microscope, which was orderly arranged and had uniform thickness. After a period of dissolution treatment, the existence of cellulose can not be observed under the magni cation of 100 times, indicating that cellulose was completely dissolved (Fig. 1b). After that, PVA was added into the homogeneous and transparent solution of IL/CEL, and cross-linked with cellulose through hydrogen bonding to obtain a network structure, which provided a favorable exible template for the oxidative polymerization of EDOT (Zhao et al. 2017). The solution was cast into a mold while it was hot, and a white PVA/CEL/EDOT(PCE) lm was obtained after cooling and solidi cation. This lm was obtained after water exchanged the IL solvent. EDOT was oxidized and polymerized along the exible polymer chain in the environment of PSS as dopant, ammonium persulfate and ferric sulfate as oxidant. In the process of doping, PEDOT may be attached to one PSS chain or more PSS chains, and gradually cross-linked to form particles with strong force (Kirchmeyer and Reuter 2005). During the oxidation process, the white gel lm gradually turns black over time (Fig. 1c, d). After drying, a black exible PCPP lm was obtained which can be curled and folded. In the reaction process, the amount of EDOT played a decisive role, which determined the conductivity of PCPP lm. If the quantity of EDOT is insu cient, the conductivity of PCPP lm is uneven. As the addition of EDOT ( Figure S1), the accumulation of it was taken place and the optimum amount of EDOT should be selected as 15 wt%. From the XPS survey spectrum of 15 wt% PCPP lm (Fig. 3), C1s, O1s and S2p numbers can be clearly seen, which indicates that PEDOT:PSS successfully bonded to the cellulose/PVA matrix lm. The existence of C, O and S elements is further con rmed by energy spectrum analysis ( Figure S2). The highresolution C1s spectra of 15wt % PCPP lm showed ve different peaks (Fig. 3b). The peak generated at 284.71 eV is attributed to C = C in the aromatic structure of PEDOT, and the aliphatic C-C peak of the PSS chain is generated at 284.81 eV. The peak at 286.10 eV corresponds to the chemical environment of C element in the C-O/C-S structure. The peak at 286.40 eV generally refers to the C-O in the polyvinyl alcohol. Two main peaks are observed in the S 2p spectrum (Fig. 3c) (Guo et al. 2013). The two low binding energy peaks at 163.71 eV and 164.84 eV are attributed to the S atom in PEDOT (Zahed et al. 2020), while the two high binding energy peaks at 168.81 eV and 168.27 eV are due to the interaction of neutral S atoms and S ions in PSS (Merche et al. 2010). The two main peaks of the PEDOT and PSS chains indicate that PSS was successfully doped into the PVA/CEL/EDOT lm structure and had a strong π-π interaction with PEDOT during the oxidation process. The 532.34 eV and 532.11 eV shown in the O 1s spectrum (Fig. 3d) are homogeneously dispersed to the PEDOT and PSS.

Chemical Structure and Morphology of the Composite Films
The SEM images show the surface morphology of the samples (Fig. 4). The comparison between Fig. 4a and Fig. 4b shows that the surface of cellulose lm is rough, and the addition of PVA makes the lm surface smooth. This may be due to the large number of hydroxyl groups in the molecular chain of PVA and cellulose that cross-link with each other to form hydrogen bonds, optimizing the morphology of the lm. As shown in the image of the PCE lm (Fig. 4c), EDOT particles are deposited on the structure of the lm. After in-situ polymerization and oxidation, pores of different degrees appear on the surface of 10 wt%, 12 wt% and 15 wt% PCPP lms (Fig. 4d-f). On one side, it may be caused by a solvent exchange of ionic liquids in the coagulation bath, the other side of the shield, it is speculated that it is caused by PSS doping. The number of pores distributed on 10 wt% PCPP lm is much less than that of 15 wt% lm. In addition, the pore distribution on the 15 wt% PCPP lm was uniform. From the statistical distribution of the average diameter of 15 wt% PCPP lm (Fig. 4f´), it can be clearly observed that the average pore diameter is mainly 8 µm. However, large pores are easier to collapse. When water is used as a solvent to exchange ionic liquids, it will have a signi cant impact on the morphology of the cellulose lm. According to the existing research (Khare et al. 2007;Liu and Budtova 2012;Swatloski et al. 2002), when ethanol is used as a coagulation bath, the lm with smaller pore size and denser pore distribution can be obtained. To avoid the problem of macroporous defects, it is necessary to carry out experimental exploration. When the cellulose, PVA and (10 wt%, 12 wt% and 15 wt%) EDOT were dissolved in the ionic liquid to form a uniform and transparent solution, and it was cooled and solidi ed in the mold, the next step was to use ethanol as coagulation bath instead of deionized water to precipitate the ionic liquid. The other steps were unchanged. After drying, the surface morphology of PCPP lms was observed by SEM. As shown in Fig. 4 (g-i), the pore size on the surface of the PCPP lms reaches a few microns with a uniform pore distribution. At the same time, the sheet resistances of PCPP lms were measured by a four-point probe, and it is found that the resistance decreases slightly, but the overall trend does not change ( Figure S5). Correspondingly, ECG monitoring was performed, and there was no difference between the ECG signal spectra obtained by the original method.
It can be seen from the test results of mechanical properties (Fig. 5a), in the cellulose/PVA composite lm, the addition of PVA can improve the tensile strength of the membrane. The increase of mechanical properties is attributed to the interaction between cellulose and PVA, especially the hydrogen bond. When EDOT is added and oxidized, the tensile strength of PCPP lms decreases greatly. It can be speculated that oxidants (APS and Fe 3+ ) will reduce the mechanical properties of PCPP lms. The tensile strength of the three groups of PCPP lms has no obvious difference. The outstanding one is that the elongation at break of 15% PCPP lm is the highest, and the stronger the tear resistance, the better the exibility and elasticity. The electrode is the interface between the human body and the bioelectric collection device.
The current in the biological body in the form of ions is converted into the electronic current in the collection device through the electrode . Therefore, the conductivity of the electrode will affect the accuracy of ECG signal transmission. The sheet resistance value of 10 wt% PCPP lm is 5.543 kΩ·sq − 1 , the sheet resistance value of 12 wt% PCPP lm is 3.201 kΩ·sq − 1 , and the sheet resistance value of 15 wt% PCPP lm is 0.196 kΩ·sq − 1 (Fig. 5b). The curled state has no obvious in uence on the resistance value. The sheet resistance of the lm electrodes increased with the increase of the EDOT loading, and the 15 wt% PCPP lm showed the optimum conductivity because EDOT was evenly dispersed into the network structure of PVA/CEL, when its addition amount reached 15 wt%. Because of the evaporation of water, the resistance of the PCPP lms increased to some extent with time. Compared with the rst day, the resistance increased signi cantly on the seventh day (Fig. 5c).
The electrochemical tests were carried out to exhibit the reaction process between the skin and the electrode lm. Sweat is inevitable on the surface of human skin, so when measuring ECG signals, lm electrode and sweat constitute an electrochemical system, in which lm electrode reacts as working electrode and sweat acts as an electrolyte. In order to simulate the reaction process between ECG electrode and skin as much as possible, the electrolyte is simulated sweat. The components of simulated sweat are 20g/L sodium chloride and 1g/L urea. Cyclic voltammetry curve and EIS impedance curve were obtained. Figure 5d shows the test results of cyclic voltammetry curves of different samples at 0.01 Vs − 1 .
The three curves all conform to the CV shape of the polymer as the electrode material. Compared with the CV curve area of 10 wt% PCPP, 12 wt% PCPP, 15 wt% PCPP lm at a scanning rate of 0.01V/s, the larger the area enclosed by the curve, the more conducive to the charge and discharge reaction, which means the better the redox performance. The area of the 15 wt% PCPP lm is obviously the largest, and the curve is close to a symmetrical rectangle. Redox peaks generate belonging to the energy storage mechanism of pseudocapacitance, which further proves the existence of PEDOT in the composite lm. The curve area of 12 wt% PCPP is roughly half of that of 15 wt% PCPP electrode, and the area of 10 wt% sample is the smallest, which shows that with the increase of EDOT concentration, the performance of the PCPP electrode is better. Then three samples were tested by cyclic voltammetry at different scanning speeds (Fig. 6a-c). The redox peaks are observed from the CV curves of 10 wt% PCPP lm. This is because the conductive polymer PEDOT generates capacitance through redox reactions, and the energy storage it produces throughout the material, not just on the surface(Anton et al. 2017). EIS impedance diagram was used to test the electron transfer performance of PCPP lm (Priniotakis et al. 2007) and the EIS of 10 wt% PCPP lm with electron transfer resistance tended to a straight line in both high and low-frequency regions (Fig. 6d), indicating that the electrode reaction was not affected by electron transfer restriction, but controlled by the diffusion process. Small semicircles appeared in the high-frequency region of EIS ( Fig. 6e, f), which indicated that there was charge transfer resistance between the electrode and the skin surface. The charge transfer resistance at the interface of 15 wt% PCPP was signi cantly lower than that of 12 wt% PCPP lm, which could be attributed to the dense structure formed by the uniform dispersion of PEDOT on the 15 wt% PCPP lm. In addition, the porous structure of the PCPP lm surface provides a large surface area and a large number of active sites on the electrode surface. The cross-linked structure of PVA/CEL through hydrogen bonding makes the conductive path more succession and uninterrupted, resulting in faster electron transfer and higher conductivity(Nur A fah et al. 2016).

Performance of PCPP lm Electrodes in ECG Monitoring
The equipment for ECG signal acquisition is a commercial BMD101 ECG (Ai et al. 2018) development module (Fig. 7a). The main components of the analog front-end include a low-noise ampli er and analogto-digital(A/D)converter ADC (Rashkovska et al. 2020), which can digitally process the analog signal (Fig. 7b), and the data value can be transmitted to PC or mobile phone through the standard Bluetooth serial port Pro Le (SPP). The acquired signal is processed by MATLAB (R 2018a). The BMD101 ECG acquisition device comes with two lead wires lead I so the PCPP lm electrodes are connected to the wrists of the left and right hands (Lo et al. 2018), using the standard lead I, and the ECG signals of two volunteers at rest are collected. ECG signals were collected by PCPP lm electrode, and the acquired data is ltered by MATLAB to obtain the ECG signal spectrogram. A normal ECG signal is composed of a series of wave groups, including P wave, QRS wave, and T wave. The p wave represents the depolarization of the atrium, the QRS wave corresponds to the depolarization wave of the ventricle, and the T wave represents the repolarization wave of the ventricle. The ECG spectrum obtained by 10 wt% PCPP lm electrode showed a regular signal value, but the PQRST characteristic wave was not detected, which is closely related to the insu cient load of EDOT in the 10 wt% lm electrode and the high sheet resistance.
On the ECG spectrum obtained by 12 wt% lm electrode, there are still many obvious noises after processing, but the existence of Q and R waves can be clearly observed (Fig. 7d). Interestingly, these waves could be clearly detected from the ECG spectra of two volunteers using the 15 wt% lm electrode (Fig. 7e, f). After 7 days, the ECG signal spectrum detected by 15 wt% PCPP lm electrode showed more noise, compared with 10 wt% and 12 wt% ECG, QRS wave could still be detected, which indicated that the conductivity of ECG lm electrode played an important role in the detection of ECG signals. The ECG signals obtained by 10 wt% and 12 wt% PCPP lm electrodes are unstable, with a lot of noise and too much chance, causing poor characteristic waves. Compared to others, the performance of 15 wt% PCPP lm electrode is more stable, which is due to the excellent conductivity of PEDOT and the stable structure assembled by EDOT and PVA /CEL in 15 wt% PCPP lm.

Conclusions
In this work, a exible cellulose-based assembly with PEDOT:PSS composite electrode was prepared to monitor the ECG signals. The conductivity of PCPP lm can be optimized by controlling the content of EDOT. When the addition of EDOT reaches 15 wt%, the sheet resistance value of PCPP lm decreases sharply, compared with that of 10%, 12% PCPP lm. Stable electrode performance can be maintained even in a crimped state. The characteristic wave of the ECG signal obtained through the 15 wt% PCPP lm is clear. During the signal measurement, no skin irritation was observed, and because of the excellent exibility of PCPP lm, it is more comfortable to wear. We believe that conductive composite materials based on cellulose biopolymer and conductive polymer PEDOT:PSS show a bright future in the elds of energy storage and electrical signal sensing.

Figure 7
Page 20/20 (a) Photograph of PC software for real-time monitoring of ECG curve. (b) Photograph of 15 wt% PCPP electrode and analog front end circuitry in the top right. (c) Photograph of contacting with the skin without any irritation. (d) ECG spectrum obtained from 10 wt% and 12 wt% PCPP electrodes. (e-f) ECG spectra of two volunteers obtained from 15 wt% PCPP electrode and ECG spectra obtained after 7 days.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Graphicalabstract.docx SupplementaryInformation.docx