Outdoor‐Useable, Wireless/Battery‐Free Patch‐Type Tissue Oximeter with Radiative Cooling

Abstract For wearable electronics/optoelectronics, thermal management should be provided for accurate signal acquisition as well as thermal comfort. However, outdoor solar energy gain has restricted the efficiency of some wearable devices like oximeters. Herein, wireless/battery‐free and thermally regulated patch‐type tissue oximeter (PTO) with radiative cooling structures are presented, which can measure tissue oxygenation under sunlight in reliable manner and will benefit athlete training. To maximize the radiative cooling performance, a nano/microvoids polymer (NMVP) is introduced by combining two perforated polymers to both reduce sunlight absorption and maximize thermal radiation. The optimized NMVP exhibits sub‐ambient cooling of 6 °C in daytime under various conditions such as scattered/overcast clouds, high humidity, and clear weather. The NMVP‐integrated PTO enables maintaining temperature within ≈1 °C on the skin under sunlight relative to indoor measurement, whereas the normally used, black encapsulated PTO shows over 40 °C owing to solar absorption. The heated PTO exhibits an inaccurate tissue oxygen saturation (StO2) value of ≈67% compared with StO2 in a normal state (i.e., ≈80%). However, the thermally protected PTO presents reliable StO2 of ≈80%. This successful demonstration provides a feasible strategy of thermal management in wearable devices for outdoor applications.

RF Analysis of Device: The resonance frequency (ƒ res ) of the coil on the device was measured using a vector network analyzer (VNA) (E5071C ENA, Keysight, USA) with a RF near-field probe (PBS1, AARONIA, Germany). The S11 magnitude mode was set to measure the ƒ res and Q-factor of the coil. The ƒ res of the coil is the frequency resulting in the minimal reflected power, as shown in Figure S4. The ƒ res equation and Q-factor equation were used to design the coil, which can be expressed as where L c is the inductance of the coil and C c is the total capacitance of the device. The Qfactor is determined by L c , C c , and the coil resistance (R c ). The Q-factor represents the sharpness of the resonance peak, and thus can be defined as an approximate calculation, where ƒ 2 and ƒ 1 is −3 dB frequency of ƒ res and ƒ 2 −ƒ 1 is the bandwidth of ƒ res .

Fabrication of NMVP:
The optimized NMVP is prepared by a two-step phaseinversion process: 1) Drop-casting porous styrene-ethylene-butylene-styrene (p-SEBS) layer on slide glass and 2) spray-coating porous polymethylmetacrylate (p-PMMA) layer on the p-SEBS layer ( Figure S17). For p-SEBS, 55.2 g of chloroform (C2432, Sigma-Aldrich) is mixed with 10.6 g of IPA and 2.4 g of SEBS beads (Tuftec TM H1062, Asahi Kasei, Japan) into a test tube at room temperature. For p-PMMA, 50 g of acetone is mixed with 5 g of DI water and 4 g of PMMA beads (182230-500G, Sigma-Aldrich). The solutions are then sonicated for one day to completely dissolve the SEBS and PMMA beads.
After the complete dissolution, the SEBS solution is dropped on a glass slide, and then the solvent (i.e., chloroform) and nonsolvent (i.e., IPA) are allowed to evaporate; thus, this procedure generates the p-SEBS layer by forming a polymer chain containing air voids.
An air-brushing tool is used to spray the p-PMMA solution on the formed p-SEBS layer. The spraying is performed from a distance of ~10 cm for 10 s. During and after the spraying step, similar to the p-SEBS layer, the solvent (i.e., acetone) and nonsolvent (i.e., DI water) are evaporated, and the PMMA produces porous polymer chains.
Bending test of PTO and NMVP: A cyclic bending test was performed on PTO and NMVP with cylindrical objects of ~1.3 cm RoC for 200 bending cycles in 20 increments, respectively. The frequency characteristics of PTO were measured every 20 bending repetitions with VNA. The solar reflectance and LWIR emissivity of NMVP were also measured every 20 bending cycles. Additionally, in order to evaluate the performance of the PTO in the bent state, the frequency characteristic was measured in the bent state, and the StO 2 and temperature measurements were compared before and after the bending of PTO on the finger (Figure S12). After measuring StO 2 and temperature for a certain period of time by placing a flat PTO on a finger, the PTO was bent on the finger for measuring StO 2 and temperature in a bent state.

Raw Data Processing for Tissue Oxygenation:
The tissue oximeter data analysis was performed using commercial software (MATLAB R2013b, MathWorks Inc, USA). The smartphone Android application was designed using commercial software (Android Studio, Google, USA) based on JAVA program language. The raw data detected by the thermistor and PD are collected at a sampling rate of ~12.5 Hz. The two LEDs blinked with a frequency of 12 Hz. A local minima-maxima finding algorithm was used to separate the two levels of the LED signals. The extracted signals were reassembled by spline interpolation of each LED signal. In order to remove high-frequency noise, a low-pass filter (0.1 Hz, 10 th -order Butterworth digital filter). The filtered data were calculated as optical densities as follows: where I t (λ) is the time-dependent signal and I 0 (λ) is the initial value of I t (λ), which were calculated for both wavelengths (λ 1 (850 nm) and λ 2 (750 nm)). In order to obtain the ΔHbO 2 and ΔHHb, the modified Beer-Lambert law for diffusive media was applied. [1] (S4) where ρ is the interoptode distance (9 mm) and DPF(λ) is the differential path length factor depending on the wavelength. [2] The ε(λ) are extinction coefficients of HbO 2 and Hb. [3] StO 2 was calculated as follows: The initial values of the total hemoglobin concentration at the forearm are (Hb t ) (0.08 mM) and StO 2 (at t = 0, 80%), respectively. [4,5] Because the calculation method is driven by the change in oxygen saturation of tissues based on the change in optical density compared to the initial value, it cannot be used as a gold standard for measuring the exact tissue oxygen saturation of a subject. However, it can be used for healthcare purposes by observing the change in tissue oxygenation in a local area as changes occur in the body (e.g., exercise or temperature). For reliable data processing and plotting of StO 2 , raw data must be accumulated for at least 3 s. In the Android application, by processing accumulated real-time data, StO 2 can be plotted at least every 3 s, and the data accumulation time can be adjusted.

Energy Balance Equation:
Using the measured emissivity spectra, the cooling temperature and cooling power are estimated by MATLAB R2018A (MathWorks, Inc., USA) based on the thermal equilibrium equation given as follows:

P rad T Sample -P Sun -P atm T ambient +h c T sample -T ambient (S6)
where P rad (T sample ) is the power radiated by the structure per unit area, P atm (T ambient ) is the absorbed power per unit area from the atmosphere, P Sun is the incoming solar power absorbed by the structure per unit area, and h c (T sample -T ambient ) is the conductive and convective heat exchange powers. These four terms are evaluated using P rad T Sample = ∫ ∫ ∫ I BB T sample ,λ ϵ λ,θ cos θ sin θ dλdθdϕ ∞ 0 π/2 0 2π 0 (S7) Here, I BB =(2hc 2 /λ 5 )/[e hc λ B T -1] is the spectral radiance of a blackbody at temperature T, where h, c, k B , λ, and h c are Planck's constant, the velocity of light, the Boltzmann constant, the wavelength, and the nonradiative heat-exchange coefficient, respectively. The atmospheric emissivity is given by amb λ,θ =1-t λ 1/ cos θ , where t is the sky transmission calculated by MODTRAN 6 using the conditions of an urban site at mid-latitude in summer (i.e., humid LWIR window). To calculate P Sun , the solar irradiation is expressed by the AM1.5G Global Tilt spectrum with an intensity of 1000 W/m 2 .
In Vivo Experiment: Before proceeding with the experiment, the subject was allowed to acclimate to the environment for approximately 120 s, sitting comfortably. The device was attached to the forearm and thigh, and the bio-signals were captured by an NFC reader at a distance of ~ 0.7 cm. The maximum communication distance between the NFC reader and PTO is ~ 1 cm (Video S6). All of the in vivo experiments in this study were performed in compliance with the protocol approved by the institutional review board at Gwangju Institute of Science and Technology (GIST). Four healthy subjects, aged 25-30 years, participated in the study. Informed consent was obtained from all subjects involved in the study.
Vein Occlusion Test: A standard method for assessing peripheral vascular diseases is venous occlusion test. An inflating cuff on the subject's bicep temporarily occluded venous blood flow for ~80 s, the variation in the measured signal at the forearm was monitored by our tissue oximeter integrated with NMVP.
Exercise experiment with commercial device: As shown in Figure

Supplementary Note 1. Ray-tracing simulation for PTO design
The distance between the LEDs and the PD, namely the interoptode distance, sets the penetrat ion depth of backscattered light into the skin. Generally, increasing the interoptode distance e xtends the light penetration depth [6] and enlarges the probing volume, thus enhancing the sens itivity. [1,7] However, an increase in interoptode distance decreases the light intensity arriving at the PD owing to absorption and scattering of tissue, water, melanin, and fat, which increase s the signal-to-noise ratio of the detected signals. [7] Therefore, considering the light intensity and noise received from PD, we set the interoptode distance as 9 mm based on a commercial non-sequential ray-tracing simulation (OpticStudio 16.5, ZEMAX LLC., USA). Figure S2A and S2B display the simulation domain of the designed structure in the top view and side view. The background material was set to 'skin' by using Henyey-Greenstein phase function with optical constants such as absorption coefficient, scattering coefficient, anisotropic factor, and refractive index ( Table S3). [8,9] To investigate the dependence of light power on the distance of the photodiode, we set seven rectangular detectors with distances of 3, 5, 7, 9, 11, 13, and 15 mm from the two LEDs. In addition, to mimic the light shielding of NMVP, we placed a diffuse reflector on the black encapsulation layer as the light shield layer. The black encapsulation layer was simplified by a perfect absorber.
Noise ratio= P e P t (S10) Noise ratio was defined as the ratio of light penetrated through only the epidermis without blood vessels (P e ) to light penetrated through the epidermis into the dermis with blood vessels (P t ). In order to calculate P e , we considered the optical properties of a 0.1-mm-thick epidermis and reflection at the contact interfaces between the epidermis and dermis. In the case of P t , the optical properties of both 0.1-mm-thick epidermis and 2-mm-thick dermis were considered. Based on the simulation, when the interoptode distance was more than 9 mm, the scattered light was quenched, and thus, the noise ratio was 0. Moreover, with increasing interoptode distance, the amount of light reaching the PD decreases ( Figure S2C and S2D). As can be seen in the contour map of Figure S2E and S2F, the light scattered only into the epidermis is almost extinguished at a distance greater than the interoptode distance of 9 mm. Therefore, we designed an interoptode distance of 9 mm so that we not only achieved a lower noise ratio, but also allowed large amounts of light to reach the PD.

Supplementary Note 2. Comparison between white elastomers and NMVP
A commercial white dye based on TiO 2 nanoparticles is the easiest and most affordable material that can be used to fabricate white-colored radiators. In our study, we also compared the optical and cooling features between the white elastomer (WE) and NMVP. For this, we first fabricated WE with polydimethylsiloxane (PDMS) by doping different white dye densities (Silc Pig TM ; White, Smooth-on, USA). The doping densities were 0.2, 0.4, 0.6, and 0.8 mL with a fixed PDMS weight (11g; 10:1 ratio of base elastomer and curing agent). Based on these ratios, we made sufficiently thick white PDMSs (~ 3.6 mm) to confirm the optical and cooling features, excluding the thickness effect ( Figure S18A). Figure S19A displays the thickness difference between the white PDMSs and NMVP. The NMVP has a thickness of ~ 0.2 mm.
The configuration of the individual temperature measurement unit is described in Figure S18B. A commercial blackbody absorbs sunlight through the samples. A temperature sensor measures the temperature of the blackbody, and the bottom Styrofoam thermally isolates the sample from the ground. Figure S18C shows the measured temperatures of the four white PDMSs and NMVP. The NMVP maintains the lowest temperature compared to other white PDMSs. Interestingly, all white PDMSs show a similar temperature regardless of the white dye density. Because the fabricated white PDMSs has a sufficiently thick layer, the density of the white dye does not significantly affect the optical feature. Figure S18D demonstrates the above-mentioned discussion using the reflectance spectra of white PDMSs and NMVP. The white PDMSs exhibit intense absorption at ultraviolet (UV) wavelengths under 400 nm. In the near-infrared region (NIR), the absorption of white PDMSs is strong. However, the NMVP shows more powerful reflection than white PDMSs in both spectral regions. Owing to its exceptional optical feature, NMVP sustains lower temperatures than all white PDMSs.
To investigate the influence of thickness on white PDMS, we also fabricated three white PDMSs with thicknesses of ~ 190, 440, and 630 μm. The PDMS weights and white dye amounts were fixed at 11 g (10:1 ratio of base elastomer and curing agent) and 0.8 mL, respectively. Figure S19A exhibits photographs of three white PDMSs and NMVP. Notably, thin white PDMSs such as ~ 190 and 440 μm cannot sufficiently block the visible light. The bottom blackbody is seen through the white PDMSs. However, the blackbody is not seen in thick white PDMS and NMVP. Figure S19B displays the measured reflectance spectra of three white PDMSs and NMVP in the solar spectrum range. Depending on the thickness, the reflectance of the white PDMS varies remarkably. In particular, the white PDMS with a thickness of ~ 630 μm, which has a three-fold thickness compared to that of NMVP, has poor reflectance compared to that of NMVP in all ranges. The measured temperature results show that white PDMSs with thicknesses of ~ 190, 440, and 630 μm cannot achieve daytime radiative cooling, whereas the NMVP shows daytime cooling under strong solar intensity, ~ 1000 W/m 2 ( Figure S19C).
Based on the above experiments, the commercial white dye-doped elastomer cannot achieve daytime radiative cooling at a thickness of ~ < 630 μm. In addition, the sufficiently th ick white elastomers cannot obtain a notable radiative cooling performance than NMVP beca use of the strong absorptions in the UV and NIR regions. However, the blackbody NMVP sh ows daytime cooling under strong sunlight (~ 1000 W/m 2 ). Therefore, the NMVP is the best material that can be used to realize thermally protected wearable optoelectronics.    2 kΩ) are LED resistor. R 3 (100 kΩ) is reference resistor for thermistor. R 4 (5 MΩ) is amplification resistor. C 1 (9 pF) is resonance capacitor for resonance frequency tuning of NFC system. C 2 (0.1 μF) and C 3 (1 μF) are decoupling capacitor to remove noise. (C) Block diagram of device system to explain principle of device operation. The acquired data by the customized smartphone application is inserted in the left of Figure S3C. The device operating system is functionally classified into three parts: the first part is the NFC reader (i.e., smartphone) that transfers power to the device and collects data from the device. The second part is the NFC interface composed of an NFC module (i.e., NFC chip, RF coil). The third part is the detection part with a μcontroller, LED, PD, and thermistor. Once the NFC reader supplies power to the device, the NFC chip rectifies the power and transfers to the μ-controller to drive the LEDs to emit light. The PD detects the backscattered light from the tissue and the thermistor obtains raw data corresponding to temperature. The collected data are transferred wirelessly to the NFC reader and processed to obtain the bio-signal (i.e., tissue oxygenation and temperature).   The overall layer of the optimum p-SEBS is composed of a porous structure, whereas the p-SEBS with the SEBS weight of 3.0 g has saturated layers on the top and bottom sides. The p-SEBS with the SEBS weight of 3.0 g is the thickest sample, but this is due to the saturated layers, which are not helpful to enhance reflectance. This means that excessive doping of polymer hinders highly porous structures. Figure S7. Optimization for porous p-PMMA in terms of weight ratio of solvent (acetone), nonsolvent (DI water), and PMMA (polymer). (A) Reflectance spectra of p-PMMA depending on different PMMA weights from 2.0 to 5.0 g. The weights of acetone and DI water are 50 and 5 g, respectively. (B) Summary of p-PMMAs with different PMMA weights with respect to the average reflectance from the wavelengths of 280 to 2000 nm and the layer thickness. The optimum p-PMMA was selected as that with the PMMA weight of 4 g. Unlike the p-SEBS case, in the p-PMMAs, the thicker sample exhibited the higher reflectance. However, the high doping of polymer cannot lead to thicker porous layers. The p-PMMA with the PMMA weight of 4 g is the optimum. (C) Optical images of best and worst cases of p-PMMA (i.e., 2 and 4 g). The p-PMMA with the PMMA weight of 2 g was not densely formed, because the polymer chains were insufficient to fully connect with each other. In contrast, the p-PMMA with the PMMA weight of 4 g shows a white and dense layer. (D, E) SEM images of the p-PMMAs with the PMMA weights of 2 and 4 g, respectively. In the p-PMMA with the PMMA weight of 2 g, the formed porous layer is too thin (~4 μm), which is mechanically very weak. These results confirm that the p-PMMA with the PMMA weight of 4 g is the optimum.     Sun and P atm (E) and T sample −T bac ground (F) as a function of θ s . At oblique angles, P sun is remarkably reduced in black elastomer, but the sample is always hotter than the background. In contrast, NMVP always maintains cooling status, except for θ s = 80°, owing to the near cancellation of reduced P Sun and increased P atm . The frequency characteristics of the device barely degrade with geometric changes and remain in the recommended range. The recommended range for Q factor is from 20 to 35. As the Q factor is high, the coil may be selective, which can result in narrow bandwidth of the resonance and also affect the NFC signal. [10,11] (C) Measured frequency characteristics of PTO for every 20 bending cycles. (D) Raw data, (E) StO 2 (blue) and temperature (red) measured by PTO at fingertip in flat and bent states. (F) Image of NMVP-integrated PTO with index finger. (G) Thermal image of fingers with temperature of ~36 °C. Since more tight contact with the skin increases the total amount of penetrated light to the skin, the raw data are slightly increased in a bent state. StO 2 and temperature results measured in the bent state barely changed compared to the flat state (~80 % and ~36 °C). The temperature results measured by NMVP-integrated PTO are similar to the result of the thermal image.       We tilted the samples and solar sensor at 30° to measure the temperature under st rong sunlight (the measurement was performed in late autumn on November 4, 2020). Before tilting, the solar intensity is remarkably weak, ~ 600 W/m 2 . 'Amb. T' denotes ambient air tem perature. Table S1. Specifications of previous reported wearable sensors in terms of operating system, sensing type, thermal management method, outdoor usability, cooling performance, and remarks. Our proposed layout only deals with wireless/battery-free optoelectronics with outdoor usability. Orange, green, and blue font are used to describe wireless/battery-free operating system, optoelectronics sensor, and the thermal management method, respectively. Our device satisfies all these aspects. Table S2. Recently-developed non-metallic radiative coolers for various applications. Our radiative cooler demonstrates the exceptional radiative cooling performance and the applicability for wearable optoelectronics. Table S3. The optical characteristics and thickness of epidermis and dermis considered for simulation. [26,27]