Energy Autonomous Wearable Sensors for Smart Healthcare: A Review

Energy Autonomous Wearable Sensors (EAWS) have attracted a large interest due to their potential to provide reliable measurements and continuous bioelectric signals, which help to reduce health risk factors early on, ongoing assessment for disease prevention, and maintaining optimum, lifelong health quality. This review paper presents recent developments and state-of-the-art research related to three critical elements that enable an EAWS. The first element is wearable sensors, which monitor human body physiological signals and activities. Emphasis is given on explaining different types of transduction mechanisms presented, and emerging materials and fabrication techniques. The second element is the flexible and wearable energy storage device to drive low-power electronics and the software needed for automatic detection of unstable physiological parameters. The third is the flexible and stretchable energy harvesting module to recharge batteries for continuous operation of wearable sensors. We conclude by discussing some of the technical challenges in realizing energy-autonomous wearable sensing technologies and possible solutions for overcoming them.


Introduction
The Internet-of-Things (IoT) is a network of interconnected devices that are playing an important role to improve the quality of individuals' life by providing a "smart" environment. IoTs have found commercial success in application areas such as smart city, transport and home. Healthcare is one of the crucial sectors where IoTs can provide individuals' safety and comfort by continuous health monitoring. In this regard, research efforts are dedicated to the "Internet-of-Medical-Things (IoMT)", linking billions of wearable devices/sensors into a communication network allowing patient-doctor communications periodically or real-time1. Figure 1 describes the concept of IoMT where wearable connected devices collect health data and transfer to healthcare service providers1-6. IoMTs provide the opportunity for medical practitioners to monitor risk patients. On-going research efforts are focused on development of multi modal wearables that can be easily integrated in clothing7-9, wrist watches and bands10-13, contact lenses14-18, tattoo-like sensory skin patches (electronic-skin)19- 25 etc. to enable around-the-clock vital signs monitoring. The battery is a significant part of an IoMT to supply continuous power to the sensors, however it is also bulky, with a limited lifespan and requiring periodic replacement26. For long-time wearable IoMT, energy autonomy is critical. Depending on the number of sensors, size and complexity of electronic circuitry, power requirement for wearables could range from 1 to 100 μW. Such demand could be met with smart nanogenerators (NGs), and that is why several green energy sources are being investigated27, as the possibility to develop "symbiotic" devices28.
An Energy Autonomous Wearable Sensor (EAWS), with low power consumption and a flexible package is indispensable for the successful deployment of IoMTs. In literature, several surveys can be found on EAWS for personal healthcare. For instance, extensive surveys presented in ref2,3,5,29,30 cover aspects of wearable sensors, whereas in ref1 focus is given to IoT for healthcare monitoring covering aspects of short-and long-range communications standards and cloud technologies. This paper provides a thorough review of wearable autonomous sensing system with a focus on sensors, batteries and energy harvesting. In Section 2, we describe the various types of sensors and their sensing mechanisms suitable for health monitoring. In Section 3, we describe the recent developments in flexible energy storage for wearables. In Section 4, we present the recent developments in energy harvesting for enabling long-lasting energy autonomy. We conclude the review in Section 5, where we provide an overview of the main challenges and viable solutions for future IoMTs.

Wearable sensors
Here, we present various sensors and their sensing principle for wearable devices. Different types of wearable sensors are reported to-date such as mechanical, biopotential electrode, optical and biochemical sensors to monitor different physiological vital signals. In the following subsections, we present a survey of mechanical and biochemical sensors.

Wearable mechanical sensors
Mechanical sensors such as strain and pressure sensors are developed and mounted on skin or integrated into clothing. Such sensors work on transduction mechanisms such as piezoresistive, capacitive, piezoelectric, and triboelectric. A representative summary of the performances of mechanical strain/pressure wearable sensors reported is provided in Table 1 (piezoresistive and capacitive) and Table 2 (piezoelectric and triboelectric). Each of them is discussed in details in the following section.

Piezoresistive sensors
Piezoresistive sensors are based on the change of electrical properties of the material when subjected to mechanical deformation. This electromechanical response of the material is called piezoresistive effect or resistive change. Piezoresistive sensor is often named as a resistive strain sensor or simply strain gauge. The piezoresistive effect has been widely exploited using many different deformable materials for the detection of physiological body signals due to many advantages such as simple device design, high sensitivity and simple readout circuits20, 31-41. The important requirements for resistive strain sensors to be employed for wearable applications are: (i) high flexibility and stretchability to absorb more strain with body movements, (ii) compactness and small size, (iii) high sensitivity, and (iv) biocompatibility.
The strain sensitivity is typically characterized with a gauge factor (GF) as = showing signal degradation at about 500 cycles of 2% strain. The microcracking-based approach may result in achieving a high GF, but device stability is a big issue to be resolved. The nanocomposite material based piezoresistive sensors exhibit large stretchability and high sensitivity, but often suffer from drawbacks such as nonlinear response, large hysteresis, and irreversibility34,64. These disadvantages can be explained as nanomaterials cannot completely come back to their initial position when unloaded of the applied strain.

Capacitive sensors
As an alternative approach to piezoresistive sensing technology, capacitive-based sensors offer some advantages, such as higher linearity, less hysteresis, and fast response time, which are important parameters when sensors are intended to be used in real-life scenarios.65-71 The simplest example of mainstream capacitive sensors is the parallel-plate configuration, as it is easy to construct and straightforward to model (Figure 2b). The capacitive change is governed by the classic equation as = ⁄ where ε is the permittivity of the cavity between two plates, and A and d represent the overlap area and distance between two plates, respectively.  Piezoresistive effect is mainly exploited to measure strain, whereas capacitive effect is used to measure pressure. Both piezoresistive and capacitive based strain/pressure sensors are gaining lot of interest as they provide high sensitivity with simple device design and readout circuits.
However, power consumption is a critical issue in the implementation of these sensors for EAWS. Research efforts are on-going to realise ultra-low power sensors to extend the life of rechargeable batteries. Meanwhile, mechanical sensors based on piezoelectric and triboelectric transduction phenomenon are gaining interest due to the possibility to realize self-powered sensors with negligible power consumption over flexible and stretchable substrates. The following sections will discuss on some of the best reported sensors in literature.

Piezoelectric sensors
Piezoelectric (PE) and triboelectric (TE) sensing mechanisms are very similar phenomena of mechanical-to-electrical energy conversion. In this subsection, we present the recent progress on stretchable mechanical strain/pressure sensors using PE effect. The PE effect is shown in Figure 3a and is based on the ability of materials to generate electrical charges under external mechanical force, pressure, or strain.88-95 When a tensile or compressive force is applied over the top of the PE device (Figure 3a), piezopotential difference, between the top and bottom electrodes, is generated. The essence of the potential developed is the relative displacement of the cations and anions centres in the PE material, resulting in a microscale dipole moment.
Polarization from all of the units, inside the material, results in a macroscopic potential drop, called "piezopotential," along the straining direction.

Triboelectric sensors
TriboElectric NanoGenerators (TENGs), introduced in 2012 by Z. L. Wang group at Georgia Tech USA100 are a promising technology to convert mechanical energy to electrical energy because of exciting device features such as self-powered, easy to fabricate and high performance. The device operational mechanism is based on coupled triboelectrification and electrostatic induction phenomenon and is schematically shown in Figure  stretchability and an issue to be resolved108. Both sensor types are more suitable for autonomous detection of dynamic pressure with wide frequency range109.

2 Biochemical sensors
The advancement in the field of nanotechnology has enabled the realization of miniaturized and cost-effective biological and chemical (biochemical) sensors for health monitoring.
Wearable BioChemical Sensors (BCS) are considered as the next generation analytical methods for an emerging third wave of technology to replace bulky and expensive analytical instruments in the health care industry2,119-121. In health monitoring, it is important to detect the presence of analytes into the body´s biofluids such as sweat using wearable biochemical sensors such as: ions (sodium, calcium, potassium, chlorine, pH), glucose, lactate, enzyme (alanine and aspartate aminotransferase, tyrosinase), proteins (troponin, C-Reactive Protein, BNP), hormone (cortisol), alcohol, drugs and antibodies29,120,122-125. However, the correlation between the level of analytes in sweat and the health condition has not been totally understood.
Wearable BCS are devices that can adhere to the body skin and designed in a way to integrate special receptors to detect the presence of biomarkers in human´s biofluids that can be either interstitial fluid (ISF) or sweat. Wearable BCS devices have additional and indispensable requirements compared to classical biochemical sensors such as: (i) sensitivity to analyte in low concentrations between fM to mM, (ii) stretchability and mechanical stability to ensure a good adhesion to the skin (iii) flexibility to certain extent to allow a comfortable movement, (iv) response and recovery times that are safe in case of no drift, (v) low power consumption for continuous monitoring, and (vi) overall sensor lifetime. The first and the last properties are much difficult to fulfil but possibility to replace the sensor might be an adequate solution.

Invasive BCS: The access of wearable biochemical sensors to interstitial fluids (ISF)
from the skin is usually invasive because they necessitate microneedles to penetrate the epidermal/dermal tissues and ensure the transport of the ISF to biomarkers´ receptors on the wearable BCS to skin. Nevertheless, the access to ISF is invasive, which represent a concern and needs the development of minimally invasive microneedles126. In this category of minimal invasive wearable, the glucose sensor is the most well-known commercial device as an easyto-use sensor indicating the sugar level of diabetic patients with high precision.

Non-invasive BCS:
Alternatively, a non-invasive wearable BCS device can use sweat biofluids to detect different analytes. In contrast to ISF, sweat is produced on the skin naturally without invasive methods2. From sweat, the wearable BCS can quantify various biomarkers.
In literature, several human biomarkers are detected into sweat such as sodium, chlorine, potassium, lactate, calcium, glucose, ammonia, ethanol, urea, cortisol, and various neuropeptides and cytokines.5 Table 3 summarizes some of the best reported work on wearable BCS sweat sensors. Various different sensing mechanisms are used to measure analytes in sweat such as colourimetric, optical, electrochemical and impedance-based sensing. The most common and versatile method is electrochemical detection, which measures electrical potential or current of transduced analyte concentrations5. 1.5 mm distance and covered with two layers of glass fiber-based paper, which has a high porosity and water absorption rate. By connecting a purely resistive load of 2 kΩ, the voltage generated by the battery is directly proportional to the sweat conductivity that is absorbed by the electrolyte (paper layer) of the battery. Later, the incorporation of platinum nanoparticles (Pt NPs) and Prussian blue (PB) as an electrocatalytic mediator allows to increase the sensitivity of glucose detection by favoring a selective oxidation of H2O2 and lowering the reaction potential to near 0 V versus Ag/AgCl133- 135. In the second generation, redox mediators as ferrocene are added to interact with enzyme, but such material could not be used in vivo due to its toxicity. The third generation GOx-based sensors use engineered enzymes to facilitate the direct electron exchange between electrodes and enzymes. For example, nanostructured electrodes, such as electrodes treated with CNTs, could be coupled to GOx but reaction still need O2 supply. In addition to glucose oxidase, glucose dehydrogenases have the advantage to induce a redox reaction independently of O2

Flexible energy storage (Batteries and Supercapacitors)
Battery  (Table 4).   Figure 5 summarizes the various structures for flexible and stretchable Li-ion batteries.

Stretchable energy storage devices:
Although this research is nascent, many advances have been made by introducing new materials and optimizing structures for high electrochemical performance suited to wearable devices. Some of the important and interesting design strategies for wearable-compatible soft batteries are discussed in the following section. wearability, which is the major concern in flexible electronic devices. Recent advancement in stretchable energy harvesting technologies is discussed in the following section.

Energy harvesting technologies
Energy harvesting is a key component for EAWS to ensure their energy autonomy over long

Solar cells: Among the popular alternative energy sources, solar energy is inexhaustible
and it is also a form of clean energy. The biggest advantage of using solar cells as the power source is their ease of integration, lack of harmful emission and readily available resources. However, it is important to note that for both outdoor and indoor cases, the peak instantaneous power harvested is not assured at all times because of conditions such as cloud cover, location of sensor, night time or indoor lights turning off. Subsequently, the power demanded by a wearable sensor cannot be met. For example, health monitoring medical patches are generally located under patient's clothes where the amount of light intensity available is very low, making solar cells unsuitable to power such wearable sensors. Thus, to ensure a steady and reliable power supply to wearables, it is necessary to explore additional energy harvesting mechanisms.
In this regard, the ubiquitous kinetic/mechanical energy is a potential source of electrical energy with possibility to fabricate devices over fully flexible and stretchable substrates97,101.
The natural mechanical energy from human body can be exploited for energy generation using  Such output voltage decreases dramatically with the decrease of the external load resistance that means additional circuitry is needed for impedance matching. Such an additional circuit will surely reduce the battery life and impose challenges for the development of EAWS system.

Thermoelectric generators (TEG):
Thermoelectric generators (TEG) have also drawn a lot of interest on using body temperature gradient as a source to generate power. TEG generate electric potential using the principle of Seebeck effect. The voltage developed by thermoelectric material is directly proportional to the temperature gradient that different part of the material is subjected to. The following equation explains this dependence mathematically:

V = α∆T
Where, α is the Seebeck coefficient (V K-1) of the thermoelectric material and ∆T (K) is the temperature gradient induced across the thermoelectric material. The figure of merit of the thermoelectric material (zT) depends on the Seebeck coefficient (α) of the material, the electrical conductivity (σ), the thermal conductivity of the material (k) and the absolute temperature (T) and is shown by the following equation: The human body is a continuous source of thermal energy, which nowadays is widely explored for supplying a substantial temperature gradient to the TEG aiding in generation of uninterrupted energy to replace the bulky energy-deficient batteries. The temperature of the human body changes with the physical activity level and the surrounding atmosphere through its metabolic function216,217. The body heat is transferred through tissues and blood to skin, which is then dissipated to our surrounding atmosphere by conduction, convection and radiation. The difference in temperature of the human body and the outer atmosphere can be of great use to produce useable source of energy, which can power various wireless transceivers, chemical sensors, ECG sensors by which one can successfully develop a health monitoring system218-222. Bismuth telluride based materials dominate with higher thermoelectric properties in wearable applications and fair amount of reviews and works have been extensively reported in the literature218,223-225.
The first commercially available µTEG, to power a wristwatch, was developed by Japanese company Seiko226. The harvested energy of the µTEG was used to power the watch as well as to charge the battery, as the maximum estimated power was 22.5 µW. The next significant commercial development in this field is demonstrated by MicroPelt. They developed their devices using sputtered BiTe or SiGe based thermoelectric materials and using flip-chip bonding approach227. The p-and n-type thermoelectric materials were deposited separately on two different wafers and bonded during the final assembly. This approach of the device fabrication offered them the ease of independent material deposition and optimization.
However, the main challenge was the bonding step of the device, which requires precise control over the thermoelectric material thickness, bonding temperatures, and the bonding material.
For wearable application, in order to exploit the body heat gradient completely, the area of contact between the device and skin has to be sufficient. respectively. An open circuit voltage of about 315 mV was measured for a temperature difference of 10 K using a complete device with 210 TE pairs. These innovative works on stretchable energy harvesters will lead to promising improvements in future autonomous wearable electronics.

Challenges and future outlook
Since the launch of first commercial wearable sensor in 1960s, wearable devices have recently begun to improve healthcare services. Such an improvement can be dedicated to the miniaturization of sensing devices, and the rapid progress in microelectronics and wireless communication technologies. Some of these modern-day wearable sensors are dominated by commercial wrist-watch sensors such as Apple Watch10, and medical patches such as the iRhythm Zio patch234 and Abbott's FreeStyle patch235 for continuous health monitoring.
However, these wearable sensors have limitations such as inaccuracies, large power consumption, short battery lifetime and no multi-sensing functionality. Therefore, several technical challenges still need to be addressed to develop EAWS for personalized healthcare.
Some of these challenges are discussed in this section. Components have to be placed on a flexible substrate in a way so that the system itself remains flexible. Same applies to the stretchability of a system. Furthermore, because of the variety of different electronic components, energy harvesters, sensing units, an energy processing unit, a processing and communication system, the components come in different sizes and packages, thus, different assembly techniques need to be used such as conductive adhesion, soldering or bonding. Therefore, integration of various device components on a single flexible and stretchable package is a major concern for EAWS system.

2) Power consumption and to find suitable source of energy:
Another challenge is to find efficient strategies to supply power for continuous operation of EAWS system. with new challenges regarding the system design such as system integration and rechargeability.

3) Miniaturization:
Wearable systems need to be comfortable for the user and therefore, the goal is to make them as small and thin as possible without sacrificing functionality. The tradeoff between comfort and energy harvesting capabilities depends on the application, thus miniaturization of the device is another issue to be resolved. 3D integration technology offers major advantages in all of these aspects and definitely will be needed in the mid-term horizon frame for miniaturized high-performance medical devices. Nevertheless, to accommodate the limited power capacity of a wearable system, several low energy communication protocols are available such as BLE or ZigBee and are supported by several microcontrollers. A common approach for wearable healthcare systems is to collect data, do some minor data processing and store the data temporarily. Periodically the stored data is sent and processed in a central system.
If the system senses an immediate danger, an emergency signal is sent out. Therefore, adapting and optimizing data sampling frequency and data transmission (according to the stand-by and active mode of the sensor) could increase efficiency of the EAWS system, and thus reduce the power consumption.

4) Sensor robustness:
Repeatability of sensing performance of the developed wearable sensor in lab-to-fab environment is a concern. Most of the best reported wearable sensors are often    Capacitive. RC is contact resistance, RD is device dimensional resistance, RT is tunnelling resistance between two 1D material, and RI is intrinsic resistance of the 1D material.                Table 4 Design Anode/CC Cathode/CC Electrolyte Performance Ref. Table 5 Design Fabrication Active material/s Electrolyte Performance Ref.