# Fileset

[Tashiro-npg24_metasurface.pdf](https://mdr.nims.go.jp/filesets/d9fb9c1c-4c33-4600-809d-3a6e53cfe858/download)

## Creator

Masaya Tashiro, Kosuke Ide, Kosei Asano, [Satoshi Ishii](https://orcid.org/0000-0003-0731-8428), [Yuta Sugiura](https://orcid.org/0000-0003-3735-4809), [Akira Uchiyama](https://orcid.org/0000-0001-7563-6191), [Hiroki Wakatsuchi](https://orcid.org/0000-0002-5774-1614)

## Rights

[Creative Commons BY Attribution 4.0 International](https://creativecommons.org/licenses/by/4.0/)

## Other metadata

[Metasurface-enabled multifunctional single-frequency sensors without external power](https://mdr.nims.go.jp/datasets/c2aff32a-2142-4c4f-beea-40204456c15e)

## Fulltext

Metasurface-enabled multifunctional single-frequency sensors without external powerTashiro et al. NPG Asia Materials           (2024) 16:55 https://doi.org/10.1038/s41427-024-00574-4 NPG Asia MaterialsART ICLE Open Ac ce s sMetasurface-enabled multifunctional single-frequency sensors without external powerMasaya Tashiro1, Kosuke Ide1, Kosei Asano1, Satoshi Ishii 2,3,4, Yuta Sugiura 2,5, Akira Uchiyama 2,6 andHiroki Wakatsuchi 1,2AbstractIoT sensors are crucial for visualizing multidimensional and multimodal information and enabling future ITapplications/services such as cyber-physical spaces, digital twins, autonomous driving, smart cities and virtual/augmented reality (VR or AR). However, IoT sensors need to be battery-free to realistically manage and maintain thegrowing number of available sensing devices. Here, we provide a novel sensor design approach that employsmetasurfaces to enable multifunctional sensing without requiring an external power source. Importantly, unlikeexisting metasurface-based sensors, our metasurfaces can sense multiple physical parameters even at a fixedfrequency by breaking classic harmonic oscillations in the time domain, making the proposed sensors viable for usagewith limited frequency resources. Moreover, we provide a method for predicting physical parameters via the machinelearning-based approach of random forest regression. The sensing performance was confirmed by estimating thetemperature and light intensity, and excellent determination coefficients larger than 0.96 were achieved. Our studyaffords new opportunities for sensing multiple physical properties without relying on an external power source orrequiring multiple frequencies, which markedly simplifies and facilitates the design of next-generation wirelesscommunication systems.IntroductionThe ability to perceive and interpret information fromseveral dimensions and modes is essential for advancingfuture information technologies1–3. Currently, varioussensors are used to detect diverse physical parameters,including temperature, light intensity, humidity, pressure,sound, angle, posture, pollution and radiation, as shown inFig. 1a4. These sensors serve numerous purposes, such asoptimizing power consumption, enhancing health care,preserving the environment, supporting agriculture andensuring security. Recently, they have been incorporatedinto wireless networks such as Internet of Things (IoT)systems, which enable future cyber-physical spaces, digitaltwins, autonomous driving, smart cities and virtual/aug-mented reality (VR or AR)1,2,5. This trend is reflected inthe global prevalence of IoT devices, specifically thesubstantial rapid annual growth rate of two billion ormore devices per year. Nevertheless, the growing demandfor IoT sensors has raised notable concerns aboutmanaging numerous devices with limited human resour-ces. More precisely, although these devices rely on bat-teries to establish communication with remote systems, itis not feasible to manually provide a new battery for everyindividual device. Therefore, battery-free or maintenance-free sensors are ideal for next-generation IoT systems.Metasurfaces can serve as a viable option in this sce-nario for detecting physical quantities without needingbattery replacement6–10. Metasurfaces are artificiallyengineered structures that exhibit distinct behavior on thebasis of the properties of their subwavelength unit cellsand the spectra of the incoming wave11–13. Metasurfacesexhibit a robust response to an incoming wave at adesigned resonant frequency and can efficiently sensephysical properties such as light intensity and temperatureby incorporating vanadium dioxide, MEMS, thermistorsand/or photocells (i.e., photoresistors) into the© The Author(s) 2024OpenAccessThis article is licensedunder aCreativeCommonsAttribution 4.0 International License,whichpermits use, sharing, adaptation, distribution and reproductionin any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate ifchangesweremade. The images or other third partymaterial in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to thematerial. Ifmaterial is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtainpermission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.Correspondence: Hiroki Wakatsuchi (wakatsuchi.hiroki@nitech.ac.jp)1Department of Engineering, Nagoya Institute of Technology, Nagoya, Aichi466-8555, Japan2Precursory Research for Embryonic Science and Technology (PRESTO), JapanScience and Technology Agency (JST), Kawaguchi, Saitama 332-0012, JapanFull list of author information is available at the end of the article1234567890():,;1234567890():,;1234567890():,;1234567890():,;http://orcid.org/0000-0003-0731-8428http://orcid.org/0000-0003-0731-8428http://orcid.org/0000-0003-0731-8428http://orcid.org/0000-0003-0731-8428http://orcid.org/0000-0003-0731-8428http://orcid.org/0000-0003-3735-4809http://orcid.org/0000-0003-3735-4809http://orcid.org/0000-0003-3735-4809http://orcid.org/0000-0003-3735-4809http://orcid.org/0000-0003-3735-4809http://orcid.org/0000-0001-7563-6191http://orcid.org/0000-0001-7563-6191http://orcid.org/0000-0001-7563-6191http://orcid.org/0000-0001-7563-6191http://orcid.org/0000-0001-7563-6191http://orcid.org/0000-0002-5774-1614http://orcid.org/0000-0002-5774-1614http://orcid.org/0000-0002-5774-1614http://orcid.org/0000-0002-5774-1614http://orcid.org/0000-0002-5774-1614http://creativecommons.org/licenses/by/4.0/mailto:wakatsuchi.hiroki@nitech.ac.jpmetasurface unit cells (Fig. 1b)14–18. In this approach,every identified physical property is associated with asingle resonant frequency. Thus, two physical propertiescan be identified via two independent resonant fre-quencies, which implies that two physical parameterscannot be independently detected at a single frequency(Fig. 1c). Moreover, in practice, the allocation of fre-quency resources is rigorously regulated19–22. Thus, theoptimal scenario would include overcoming thefrequency-domain restrictions imposed by classic reso-nant mechanisms and detecting multiple physical quan-tities at only a single frequency. For this reason, this studyproposes metasurface-based sensors that change scatter-ing profiles depending on the physical properties of thesurrounding environment even at the same frequency(Fig. 1d, e). Our sensors are specifically designed to detectRCC L0 0C RS11Structural elementsLumped elements(Sensing elements)(a)(b) (c)(d)C L0 0CRS11Light sensorTemperature sensorHumidity sensorPressure sensorSound sensorGyroscope sensor(e)C L0CC 0CRFrequency-domaincharacterizationTime-domaincharacterizationRCFrequency+ CRS11Time+C RS11+Fig. 1 Concept underlying the metasurface-based sensors. a Various sensors that are used in daily life. b Conventional metasurface-based sensordesign using frequency-domain scattering profiles that address the issue of regular battery replacement and (c) a corresponding equivalent circuitmodel with scattering characteristics. Two circuit values (two physical properties) cannot be independently detected at a single frequency.d Proposed metasurface-based sensor design using time-domain scattering profiles that additionally address the restriction of limited frequencyresources and (e) a corresponding equivalent circuit model with scattering characteristics. Even at the same frequency, two physical properties canbe independently detected in the time domain by using diode bridges with sensing circuit elements such as photocells and temperature-dependentcapacitors. This sensor design breaks harmonized oscillations and attains a large degree of freedom to detect multiple physical parameters.Tashiro et al. NPG Asia Materials           (2024) 16:55 Page 2 of 10    55 light intensity and temperature by integrating photocellsand temperature-dependent capacitors. Importantly,however, these physical quantities are identified usingonly a single frequency because of time-domain scatteringchanges with a machine learning methodology23,24. Byaltering the integrated circuit layout, the design conceptof our metasurface-based sensors may be extended todetect additional physical properties. Thus, this studyhelps achieve maintenance-free and sustainable next-generation wireless communication systems.Results and discussionFundamental design theoryA key solution for detecting multiple physical quantitiesat the same frequency is to break the harmonized time-domain response. To this end, time-varying metasurfaceshave been intensively studied and exploited for wavefrontcontrol, which aids in designing reconfigurable intelligentsurfaces (RISs)25–27, nonreciprocal wave propagation28,29,and radiofrequency (RF) and optical devices30,31.Although most time-varying metasurfaces require exter-nal power sources such as active metasurfaces26,32–34,passive and time-varying metasurfaces have been recentlyproposed to change the electromagnetic response even atthe same frequency in accordance with the duration of theincoming pulse31,35–37; this approach is exploited in thisstudy. In fact, these pulse-width-dependent metasurfaces,or so-called waveform-selective metasurfaces, rely on thewell-known transients of classic direct current (DC) cir-cuits. More specifically, waveform-selective metasurfacescomprise a periodic conducting pattern and resonate atresonant frequencies, as seen in ordinary meta-surfaces38,39. However, by introducing a set of four diodesas a diode bridge into the gap between conductor edges,the waveform of the incoming wave (the sine function inthis study) is fully rectified (as in the waveform based on |sin|), generating an infinite set of frequency componentsin which most of the energy is concentrated around thezero-frequency band. Therefore, transient phenomenacan occur if the reactive circuit components are includedinside the diode bridge. Specifically, this study uses acapacitor connected to a resistor in parallel inside thediode bridge. Under this circumstance, the reflection fromthe metasurface is reduced during an initial periodbecause the incoming energy is stored in the capacitorand dissipated with the parallel resistor. However, byincreasing the incident pulse width, the capacitor is fullycharged so that the incident wave is poorly absorbed andstrongly reflected even at the same frequency.In particular, transient responses are characterized bytime constants and steady-state resistance, which areassociated with sensing circuit components whose circuitvalues vary in accordance with physical quantities. Forexample, some capacitors are well known to change theircapacitance due to temperature changes. Additionally,photocells have variable resistance values in accordancewith the surrounding light intensity. Therefore, byincorporating these circuit elements as parallel capacitorsand resistors inside diode bridges, transients (or time-varying responses) change depending on the temperatureand light intensity, which can be detected from scatteringwaves. Specifically, as explained in the literature40 and theSupplementary Information (Supplementary Note 1), thetime constant of our metasurface τ is determined byτ ¼ CRCRdRC þ Rd; ð1Þwhere C and RC represent the capacitance and resistance,respectively, of the discrete components inside the diodebridge (i.e., the parallel capacitor and resistor). Addition-ally, Rd denotes the resistance of the two diodes at theturn-on voltage. In particular, if RC >> Rd, τ is simplifiedtoτ � CRd; ð2Þwhich indicates that the transition time is mostlychanged by C since the Rd is not adjustable. Additionally,in the steady state, the capacitor approaches an opencircuit so that the reflecting state is related to the valuesof RC and Rd. Thus, RCs can be exploited to control thesteady-state response. Moreover, because C and RC varybecause of the temperature dependence of capacitorsand the light-intensity dependence of photocells, ourmetasurface can detect temperature and light intensityin accordance with the reflected waveform. Note thatalthough this study limits the multifunctional sensingcapability to only two physical quantities, the proposedconcept can be further extended to detect additionalphysical quantities by introducing extra circuit compo-nents. For example, the above capacitor-based circuitconfiguration can be integrated with an inductor-basedcircuit configuration to produce a reflectance peak, dip,or more advanced waveform, which can be associatedwith more than two circuit parameters in the timedomain41,42.Numerical demonstrationBefore experimental validation, we numerically showhow metasurface-based sensors vary their time-domainresponse in accordance with circuit parameters. As shownin Fig. 2a, the unit cells of our metasurface comprise aground plane, a dielectric substrate (Rogers 3003) andconducting patches with minor trimming to deploy smallconducting pads and form a diode bridge including aparallel RCC circuit (RC= 10 kΩ and C= 1 nF). Forsimplicity, these simulations use ordinary capacitors andresistors for C and RC instead of a temperature-dependentTashiro et al. NPG Asia Materials           (2024) 16:55 Page 3 of 10    55 capacitor and a photocell, respectively. Detailed infor-mation for these simulations is provided in the “Simula-tions” subsection of the Methods Section and theSupplementary Information (Supplementary Note 2),including the design parameters for the conducting geo-metry, substrate and lumped circuit elements. Underthese circumstances, the reflecting profiles of the meta-surface for 50-ns short pulses and continuous waves(CWs) varied, as shown in Fig. 2b. According to thesesimulation results, the metasurface considerably reducesthe reflectance magnitude only for short pulses near3.9 GHz, which is consistent with the pulse width-dependent absorbing mechanism explained earlier. Tofurther clarify this reflectance trend, C and RC were var-ied, as shown in Fig. 2c, which indicates that the transitiontime was shifted by increasing C from 10 nF to 100 nF.Additionally, the steady-state reflectance decreased withdecreasing RC from 10 kΩ to 1 kΩ. Importantly, these twoconclusions also support the aforementioned theoreticaldesign and ensure that two independent physicalquantities can be detected if they are associated withchanges in C and RC. Related results and information areprovided in the Supplementary Information (Supple-mentary Note 2).Experimental demonstrationOn the basis of the simulation results shown above, wefabricated and experimentally tested the metasurface-based sensor, as shown in Fig. 3a. For these measure-ments, our metasurface was composed of 15 × 15 unitcells and placed on a programmable hot plate that arbi-trarily and directly controlled the temperature of themetasurface sample. Additionally, a light source waspositioned in front of the metasurface and controlled bypulse width modulation (PWM) signals. In addition tothese thermal and light sources, incident signals wereradiated by a standard horn antenna, and another hornantenna was used to receive the reflected waveform. Theincident and reflected angles were set to 30°, and theincident wave was a transverse electric (TE) polarizedwave. Detailed information about the measurements isprovided in the “Measurement Samples” and “Measure-ment Methods” subsections of the Methods Section andthe Supplementary Information (Supplementary Note 2).Under these circumstances, the temperature of themetasurface and the surrounding light intensity were setto 22.1 °C and 328 lux, respectively, which provided nearlyidentical values of C and RC as those shown in Fig. 2b(specifically, 10 nF and 10 kΩ). Thus, the frequency-domain profiles shown in Fig. 3b demonstrated a rela-tively low transmittance for short pulses near 3.76 GHzdespite the use of the same frequencies, which was con-sistent with the simulation results in Fig. 2b. Note that thetransmittance in Fig. 3b was entirely lower than that inFig. 2b because the measurement was performed in openspace to consider a realistic sensing environment, whereasthe simulation was conducted with periodic boundaries asa simplified situation. Additionally, a minor frequencyshift appeared in the measurements because of differencesbetween the simulations and measurements, e.g., theincident angle and parasitic circuit parameters that onlyappeared in the measurement sample. However, despitethese differences, the reflectance profile evidently variedin accordance with the incident waveform even during themeasurements.Next, we clarified how the time-domain responsevaried with changes in temperature and light intensity.In these measurements, we simplified the loaded circuitsinside the diode bridges and used pairs of eithertemperature-dependent capacitors and fixed resistors orfixed capacitors and photocells, which facilitated theanalysis of the temperature and light intensity depen-dences. First, pairs of temperature-dependent capacitorsand fixed resistors (10 kΩ) were used with variable100 101Time (µs)-25-20-15-10-50Reflectance (dB)(b) (c)(a)3 3.5 4 4.5 5Frequency (GHz)-20-15-10-50Reflectance (dB)CWPulseC RCEHkFig. 2 Numerically simulated reflectance profiles. a Realistic unit cell model with periodic boundaries applied for the incident E and H directions.b Frequency-domain reflectance. C, RC and the input power were set to 1 nF, 10 kΩ and 0 dBm, respectively. c Time-domain reflectance at 3.9 GHzwith various C and RC values.Tashiro et al. NPG Asia Materials           (2024) 16:55 Page 4 of 10    55 temperature values at 3.82 GHz, as shown in Fig. 3c.According to these measurements, the transition timedecreased with increasing metasurface temperaturefrom 23.5 °C to 65.0 °C because the capacitance of thetemperature-dependent capacitors decreased, whichdecreased the time constant. This result was consistentwith the numerical simulation in Fig. 2c and Eq. (2).Additionally, when pairs of fixed capacitors (1 nF) andphotocells were alternatively used within the diodebridges, the metasurface varied the steady-state reflec-tance, as shown in Fig. 3d. Specifically, the reflectancedecreased from −30.9 dB to −37.5 dB as the lightintensity increased from 3 lux to 1970 lux, whichresulted in a decrease in the effective resistivecomponent of the photocells and strong absorption ofthe incident wave in the steady state.Moreover, the metasurface was experimentally eval-uated using the temperature-dependent capacitors andthe photocells, as shown in Fig. 3e. This measurementresult also ensures that the changes in temperature andlight intensity affect the reflectance profiles. For example,by increasing the temperature from 40 to 70 °C, the timeconstant of the metasurface was reduced so that thereflectance curves were shifted to a smaller time scale (i.e.,comparing the solid curves with the dashed curves).Additionally, increasing the light intensity from approxi-mately 300 to 2000 lux decreased the steady-state reflec-tance from approximately −23 dB to −33 dB (see the100 101Time (µs) -60-50-40-30-20Reflectance (dB)309 Lux, 40.0 deg.2000 Lux, 40.0 deg.302 Lux, 70.0 deg.1928 Lux, 70.0 deg.3 3.2 3.4 3.6 3.8 4Frequency (GHz)-60-50-40-30-20CWPulseReflectance (dB)100 101Time (µs) -60-50-40-30Reflectance (dB)23.5 deg.38.0 deg.52.5 deg.65.0 deg.100 101Time (µs) -60-50-40-30Reflectance (dB)3 Lux370 Lux970 Lux1970 Lux(b) (c)(d) (e)(a) MetasurfacesampleLight sourceHeat sourceFig. 3 Experimental validation. a Measurement sample and measurement system using light and heat sources in free space. b Frequency-domainreflectance. The temperature, light intensity and input power were set to 22.1 °C, 328 lux and 30 dBm, respectively. c–e Time-domain reflectance ofthe metasurface-based sensor using (c) temperature-dependent capacitors and fixed resistors (10 kΩ), (d) fixed capacitors (1 nF) and photocells and(e) temperature-dependent capacitors and photocells. The frequency was set to optimal values to increase the time-domain variation in (c–e)(specifically, 3.82 GHz in (c) and (d) and 3.76 GHz in (e)).Tashiro et al. NPG Asia Materials           (2024) 16:55 Page 5 of 10    55 difference between the black curves and the red curves).Therefore, by associating C and the RC with physicalquantities, our metasurface design independently con-trolled the time constant and the steady-state response. Inthe Supplementary Information (Supplementary Note 3),this metasurface-based approach is also demonstrated viasimpler versions of structures such as microstrips andone-dimensional metasurface lines, which further vali-dates the time-varying scattering effect even at the samefrequency in accordance with the temperature and lightintensity.Estimation of physical quantitiesWe further present an approach to estimate tempera-ture and light intensity on the basis of the physicalquantities associated with C and RC. Although otherapproaches are potentially applicable for predicting thesetwo physical quantities (e.g., use of theoretical equivalentcircuit models40,43–45), we adopted a machine learningapproach based on random forest regression24. Our esti-mation approach was composed of four steps. First, themeasured time-domain reflectance profiles were dividedinto 40 segments of time on a log scale. Second, in eachsegment, an average reflectance value was obtained.Third, these average reflectance values were used asexplanatory variables for the training data. Finally, on thebasis of the training data, the temperature and lightintensity were estimated and compared to their actualvalues. We varied the number of training datasets and testdatasets, while the total number of these datasets wasfixed at 2290 (i.e., the ratio between the training datasetsand test datasets was varied within the entire dataset of2290). Further details are provided in the “EstimationMethod” subsection of the Methods Section.The corresponding estimation results are shown inFig. 4. When only 11 datasets were used as trainingdatasets in Fig. 4a, b, the correlation between the esti-mated values and the original values was poor, withdetermination coefficients of 0.6456 and 0.5841 for tem-perature and light intensity, respectively. However, byincreasing the number of training datasets to 458, thesedetermination coefficients improved to 0.9861 and 0.9610for temperature and light intensity, respectively. Theseresults indicate that a proper number of datasets need tobe used for the training process, which is consistent withother reports on AI-based metasurface studies46–48. Moreimportantly, these results validate that our metasurface-based sensors can be used to estimate physical quantitiesat the same frequency. The Supplementary Information(Supplementary Note 4) provides additional results,(a) (b)(c) (d)20 40 60 80 100Original value (deg.)20406080100Estimated value (deg.)0 500 1000 1500 2000 2500Original value (Lux)05001000150020002500Estimated value (Lux)0 500 1000 1500 2000 2500Original value (Lux)05001000150020002500Estimated value (Lux)20 40 60 80 100Original value (deg.)20406080100Estimated value (deg.)Fig. 4 Simultaneous estimation of temperature and light intensity from reflected waveforms. a, b Use of 11 training datasets for (a)temperature and (b) light intensity estimation. c, d Use of 458 training datasets for (c) temperature and (d) light intensity estimation. Thedetermination coefficients of (a)–(d) were 0.6456, 0.5841, 0.9861 and 0.9610, respectively. Additional results are provided in the SupplementaryInformation (Supplementary Note 4).Tashiro et al. NPG Asia Materials           (2024) 16:55 Page 6 of 10    55 including the use of an estimation method other thanrandom forest regression.DiscussionUnlike conventional structures11–13,38,49, our meta-surfaces are passive yet able to break classic harmonicoscillations even at a fixed frequency owing to the lumpedtransient circuit elements, more specifically, their circuitvalues, such as capacitance and resistance. In particular,our metasurfaces enable us to control time-varyingresponses as the capacitance and resistance changedepending on the surrounding environment, which hasnot been demonstrated in the literature22,35–37,41. Addi-tionally, although our metasurface-based sensors behavedifferently in accordance with the pulse duration of theincident wave, as shown in Fig. 2b, the sensing processitself requires CWs (i.e., not different pulses), as shown inFig. 2c. Therefore, our metasurface-based sensors do notneed additional complicated modulation techniques if thepulse width is sufficiently long. In the literature,waveform-selective metasurfaces are integrated with anantenna design to tailor antenna characteristics, includingradiation patterns in accordance with the pulse durationeven at a constant frequency22,50,51. To date, the conceptof such waveform selectivity has been introduced into, forinstance, sensors22, RISs27,52, IoT tags53 and signal pro-cessing31,54. However, unlike conventional studies, ourmetasurfaces allow us to control time-varying responsesvia changes in the circuit values of lumped components,as mentioned above. Note that our machine learningapproach itself is quite ordinary and readily availablefrom, for instance, the built-in functions of Python. Thismachine learning approach cannot predict physicalquantities without our metasurfaces, which indicates thatthe uniqueness of our study lies in our metasurface designbut not in the machine learning approach.Our metasurface-based sensing approach was experi-mentally validated to be capable of estimating more thanone physical quantity by breaking the harmonic oscilla-tion of metasurfaces in the time domain and using onlyone frequency component. Our approach is rationalbecause the use of frequency resources is strictly deter-mined in practice to avoid electromagnetic interferenceissues19–21. This approach may apply to one of theindustrial, scientific and medical (ISM) bands, wherefrequency resources are readily available without rigorouslicense issues for the use of radio-frequency (RF) waves.Moreover, our approach is useful for managing anddesigning future IoT systems. Conventionally, IoT sensorswere designed to obtain multidimensional and multi-modal information for realizing next-generation IoT sys-tems, including cyber-physical space, digital twins,autonomous driving, smart cities and VR/AR1,2,5. How-ever, this conventional approach requires regularreplacement of internal batteries to maintain commu-nication with external internet/cloud systems. Because thenumber of IoT sensors is increasing rapidly, maintainingall IoT sensors manually and replacing their batteries willsoon be unrealistic. Here, our metasurface-basedapproach does not require the use of batteries but per-mits the sensing of multiple physical quantities, addres-sing an emerging issue in the design of future IoTsystems. In particular, although this study was limited tosensing two physical quantities as a proof of concept, ourapproach can be extended to sensing more than twophysical quantities by adding additional circuit compo-nents that react with other physical quantities and char-acterize the time-domain response of metasurfaces37,41,42.For more practical use as IoT sensors, furtherimprovements of our metasurface-based sensors arerequired. For example, the metasurfaces demonstrated inthis study rely on commercial diodes that require a largeamount of input power to rectify incoming signals andvary the time-domain response associated with physicalquantities. Therefore, reducing the power level by usingcustomized low-power diodes helps design battery-freemetasurface-based multifunctional sensors in an energy-efficient manner. Additionally, the successful imple-mentation of our approach in realistic environmentsdepends on how reflected (or scattered) waveforms areassociated with physical quantities. This study addressedthis issue by exploiting random forest regression, whichprovided excellent determination coefficients larger than0.96. Although other sensor technologies may providegreater accuracy, our determination coefficient can befurther improved by adopting other machine learningtechniques suited for individual application scenarios(refer to Supplementary Note 4 in the SupplementaryInformation). Moreover, cost-effectiveness is an impor-tant point in our study since the proposed metasurface-based sensors are maintenance-free and do not requirehuman resources to replace the batteries of conventionalIoT sensors. Therefore, the proposed approach onlyrequires the fabrication cost of the metasurfaces. Thiscost-effectiveness can be further evaluated and discussedin future studies. The present study evaluated ourmetasurfaces with plane waves but not with spatiallymodulated waves55–57, which can be exploited as anadditional degree of freedom to sense physical quantities.At the same time, however, the use of spatial modulationmay narrow the applicable sensing scenarios in terms ofthe angles of the incident wave and the metasurface.Finally, connecting our sensing approach to IoT/cloudsystems is an important issue that is not fully demon-strated in this study. In particular, real-time feedback isneeded for IoT systems in cyber-physical space, smartcities, autonomous driving, farming and health care58–61,where wireless networks are expected to reduce theTashiro et al. NPG Asia Materials           (2024) 16:55 Page 7 of 10    55 estimation time for physical quantities by using, forexample, simplified learning models and fast-speed cal-culation approaches.ConclusionIn conclusion, we present a metasurface-based sensordesign that achieves multifunctional sensing without theneed for multiple frequencies or an external powersupply. Our metasurface showed variable reflectanceprofiles in the time domain, which were independentlydetermined by lumped circuit parameters that wereresponsive to the two physical quantities of interest,specifically, temperature and light intensity. Addition-ally, we introduced an approach to estimate the tem-perature and light intensity from the reflected waveformof the metasurface via random forest regression. Thus,the temperature and light intensity were successfullydetected, with determination coefficients of 0.9861 and0.9610, respectively. Our study affords new possibilitiesfor sensing multiple physical quantities without the needfor an external power supply or several frequencies,which facilitates the design of next-generation wirelesscommunication systems.MethodsSimulationsNumerical simulations were performed via a cosimu-lation method in ANSYS Electronics Desktop (versionR2) 2022. This method models metasurfaces in anelectromagnetic solver (HFSS). Importantly, all of thediscrete circuit components were replaced with lumpedports. The scattering parameters of the metasurfaceswere then used in a circuit simulator (Circuit) as circuitmodels. In these circuit simulations, lumped ports wereconnected to the actual circuit components used, whichwas equivalent to directly including the circuit compo-nents in electromagnetic simulations. However, thiscosimulation method facilitated the entire simulationprocess for readily obtaining the final simulation resultscompared with stand-alone electromagnetic simulationapproaches62. Short-pulse simulations (as shown by thered curve in Fig. 2b) were conducted via 50-ns pulses. Inthis case, the total reflected energy was compared withthe total incident energy to calculate the reflectance.Additionally, CW simulations (as shown by the blackcurve in Fig. 2b) were performed via the harmonic bal-ance approach, where the steady-state response wasdirectly obtained. Here, we calculated the reflectance bydividing the reflected energy by the incident energyduring 2 cycles. Moreover, we also calculated the tran-sient reflectance in the time domain (as shown inFig. 2c). In this case, the time-varying reflectance wasobtained by calculating the moving average of thereflected energy for 250 ns with discretized 100 pstime steps and comparing it with the moving average ofthe incident energy. The detailed design parameters ofthe simulation models are given in the SupplementaryInformation (Supplementary Note 2).Measurement samplesOur metasurface measurement samples consisted of aground plane, a dielectric substrate (Rogers 3003) and aperiodic array of square conducting patches with minortrimming to connect discrete circuit components. Thesedesign parameters are fully summarized in the Supple-mentary Information (Supplementary Note 2). Thediodes used were provided by Avago (HSMS-286x series).The temperature-dependent capacitors and photocellswere produced by Murata Manufacturing Co.(RDEF51H013Z0P1H03B) and Luna Innovations (NSL-19M51), respectively.Measurement methodAlthough detailed measurement setups are illustrated inthe Supplementary Information (Supplementary Note 2), tocharacterize frequency-domain profiles (as shown inFig. 3b), we used not only a vector network analyzer (VNA)(Keysight Technologies, N5249A) but also an amplifier(Ophir, 5193RF) to sufficiently increase the input powerlevel and turn on the diodes loaded on the metasurfaces.For time-domain profiles (e.g., those shown in Fig. 3c toFig. 3e), we used a signal generator (Anritsu, MG3692C) asa signal source. Similarly, the abovementioned amplifierwas used to ensure that the input power level was suffi-ciently large. Additionally, an isolator was used to protectthe amplifier and the signal generator from excessivereflection. Part of the incident wave was sent to an oscil-loscope (Keysight, DSOX6002A), while most of the energywas radiated to the metasurfaces through a standard hornantenna (Schwarzbeck Mess-Elektronik, BBHA9120D).Importantly, the surrounding light intensity and the tem-perature of the metasurfaces were controlled by a lightsource (Safego, C36W-FL) and a programmable hot plate(AS ONE, ND-2A). These light and heat sources werearbitrarily controlled by pulse width modulation (PWM)and proportional–integral–derivative (PID) control,respectively. Owing to these two sources, the metasurfacesvaried the reflected waveforms that were received byanother horn antenna and measured by the oscilloscope. Asmentioned in the above “Simulations” subsection, thetransient reflectance varying in the time domain wasobtained by comparing the reflected energy with the inci-dent energy.Estimation methodsThe reflected waveform was used to estimate theactual temperature and light intensity. This study usedPython program codes based on random forestTashiro et al. NPG Asia Materials           (2024) 16:55 Page 8 of 10    55 regression to obtain the results shown in Fig. 4. Here, weused a built-in function of Phyton (specifically, “split”)to select training datasets at random from all of themeasurement datasets (2290 in total), while theremaining datasets were used as test datasets. Forexample, in Fig. 4a, b, 11 datasets were selected atrandom as training datasets, whereas the remaining2279 datasets (=2290 to −11) were used as test datasets.In Fig. 4c, d, the number of training datasets wasincreased to 458, whereas that of the test datasets wasreduced to 1832. In the Supplementary Information(Supplementary Note 4), the relationship between thenumber of training datasets and the number of testdatasets was varied to demonstrate how these numbersinfluence the estimation performance. Ridge regres-sion63 was also alternatively applied to estimate thetemperature and light intensity in the SupplementaryInformation (Supplementary Note 4).AcknowledgementsThis work was supported in part by the Japan Science and Technology Agency(JST) under Fusion Oriented Research for Disruptive Science and Technology(FOREST) and under Precursory Research for Embryonic Science andTechnology (PRESTO) Nos. JPMJPR193A, JPMJPR1932 and JPMJPR2134 and theNational Institute of Information and Communications Technology (NICT),Japan, under commissioned research No. 06201.Author details1Department of Engineering, Nagoya Institute of Technology, Nagoya, Aichi466-8555, Japan. 2Precursory Research for Embryonic Science and Technology(PRESTO), Japan Science and Technology Agency (JST), Kawaguchi, Saitama332-0012, Japan. 3Research Center for Materials Nanoarchitechonics (MANA),National Institute for Materials Science (NIMS), Tsukuba, Ibaraki 305-0044,Japan. 4Faculty of Pure and Applied Physics, University of Tsukuba, Tsukuba,Ibaraki 305-8577, Japan. 5Graduate School of Science and Technology, KeioUniversity, Yokohama, Kanagawa 223-8522, Japan. 6Graduate School ofInformation Science and Technology, Osaka University, Suita, Osaka 565-0871,JapanAuthor contributionsH.W. conceived the concept of the project. M.T., K.I., and K.A. conducted thesimulations and measurements and analyzed the results. H.W., S.I., Y.S., and A.U.supervised these simulations and measurements and supported their dataanalysis. H.W. wrote the manuscript, and all the authors commented on it.Data availabilityThe data that support the findings of this study are available from thecorresponding author upon request.Competing interestsThe authors declare no competing interests.Publisher’s noteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.Supplementary information The online version contains supplementarymaterial available at https://doi.org/10.1038/s41427-024-00574-4.Received: 7 June 2024 Revised: 25 August 2024 Accepted: 4 September2024References1. Gubbi, J., Buyya, R., Marusic, S. & Palaniswami, M. Internet of things (IoT): avision, architectural elements, and future directions. Futur. Gener. Comput. Syst.29, 1645–1660 (2013).2. Al-Fuqaha, A., Guizani, M., Mohammadi, M., Aledhari, M. & Ayyash, M. Internetof things: a survey on enabling technologies, protocols, and applications. IEEECommun. Surv. Tutor. 17, 2347–2376 (2015).3. Dang, S., Amin, O., Shihada, B. & Alouini, M.-S. What should 6G be? Nat.Electron. 3, 20–29 (2020).4. Wilson, J. S. Sensor Technology Handbook (Elsevier, 2004).5. Tao, F., Zhang, H., Liu, A. & Nee, A. Y. C. Digital twin in industry: state-of-the-art.IEEE Trans. Ind. Inform. 15, 2405–2415 (2018).6. Liu, N., Mesch, M., Weiss, T., Hentschel, M. & Giessen, H. Infrared perfectabsorber and its application as plasmonic sensor. Nano Lett. 10, 2342–2348(2010).7. Schueler, M., Mandel, C., Puentes, M. & Jakoby, R. Metamaterial inspiredmicrowave sensors. IEEE Microw. Mag. 13, 57–68 (2012).8. Ebrahimi, A., Withayachumnankul, W., Al-Sarawi, S. & Abbott, D. High-sensitivitymetamaterial-inspired sensor for microfluidic dielectric characterization. IEEESens. J. 14, 1345–1351 (2013).9. Sreekanth, K. V. et al. Extreme sensitivity biosensing platform based onhyperbolic metamaterials. Nat. Mater. 15, 621–627 (2016).10. Hu, X. et al. Metamaterial absorber integrated microfluidic terahertz sensors.Laser Photonics Rev. 10, 962–969 (2016).11. Yu, N. et al. Light propagation with phase discontinuities: generalized laws ofreflection and refraction. Science 334, 333–337 (2011).12. Pfeiffer, C. & Grbic, A. Metamaterial Huygens’ surfaces: tailoring wave frontswith reflectionless sheets. Phys. Rev. Lett. 110, 197401 (2013).13. Yu, N. & Capasso, F. Flat optics with designer metasurfaces. Nat. Mater. 13,139–150 (2014).14. Driscoll, T. et al. Memory metamaterials. Science 325, 1518–1521 (2009).15. Liu, M. et al. Terahertz-field-induced insulator-to-metal transition in vanadiumdioxide metamaterial. Nature 487, 345–348 (2012).16. Tao, H. et al. Reconfigurable terahertz metamaterials. Phys. Rev. Lett. 103,147401 (2009).17. Ren, Z. et al. Leveraging of MEMS technologies for optical metamaterialsapplications. Adv. Opt. Mater. 8, 1900653 (2020).18. Lin, W. & Ziolkowski, R. W. Wirelessly powered light and temperature sensorsfacilitated by electrically small omnidirectional and Huygens dipole antennas.Sensors 19, 1998 (2019).19. The European table of frequency allocations and applications in the frequencyrange 8.3 khz to 3000 ghz (eca table) Approved November 2020 (2020).20. FCC online table of frequency allocations 47 c.f.r. §2.106. Revised on February1, 2021. (2021)21. MIC: The Radio Use Web Site. MIC Frequency Assignment Plan, September2021. https://www.tele.soumu.go.jp/e/adm/freq/search/share/plan.htm (2021).22. Ushikoshi, D. et al. Pulse-driven self-reconfigurable meta-antennas. Nat.Commun. 14, 633 (2023).23. Tibshirani, R. Regression shrinkage and selection via the Lasso. J. R. Stat. Soc.Ser. B: Stat. Methodol. 58, 267–288 (1996).24. Breiman, L. Random forests. Mach. Learn. 45, 5–32 (2001).25. Wu, Q. & Zhang, R. Towards smart and reconfigurable environment: intelligentreflecting surface aided wireless network. IEEE Commun. Mag. 58, 106–112(2020).26. Zhang, L. et al. Space-time-coding digital metasurfaces. Nat. Commun. 9, 4334(2018).27. Fathnan, A. A. et al. Unsynchronized reconfigurable intelligent surfaceswith pulse-width-based design. IEEE Trans. Veh. Technol. 72, 15103–15108(2023).28. Caloz, C. et al. Electromagnetic nonreciprocity. Phys. Rev. Appl. 10, 047001(2018).29. Nagulu, A., Reiskarimian, N. & Krishnaswamy, H. Non-reciprocal electronicsbased on temporal modulation. Nat. Electron. 3, 241–250 (2020).30. Barbuto, M. et al. Metasurfaces 3.0: a new paradigm for enabling smartelectromagnetic environments. IEEE Trans. Antennas Propag. 70, 8883–8897(2022).31. Takimoto, K. et al. Perfect pulse filtering under simultaneous incidence at thesame frequencies with waveform-selective metasurfaces. APL Mater. 11,081116 (2023).32. Liu, M., Powell, D. A., Zarate, Y. & Shadrivov, I. V. Huygens’ metadevices forparametric waves. Phys. Rev. X. 8, 031077 (2018).Tashiro et al. NPG Asia Materials           (2024) 16:55 Page 9 of 10    55 https://doi.org/10.1038/s41427-024-00574-4https://www.tele.soumu.go.jp/e/adm/freq/search/share/plan.htm33. Taravati, S. & Eleftheriades, G. V. Full-duplex reflective beamsteering metasur-face featuring magnetless nonreciprocal amplification. Nat. Commun. 12, 4414(2021).34. Moussa, H. et al. Observation of temporal reflection and broadband frequencytranslation at photonic time interfaces. Nat. Phys. 19, 863–868 (2023).35. Wakatsuchi, H., Kim, S., Rushton, J. J. & Sievenpiper, D. F. Waveform-dependentabsorbing metasurfaces. Phys. Rev. Lett. 111, 245501 (2013).36. Wakatsuchi, H., Long, J. & Sievenpiper, D. F. Waveform selective surfaces. Adv.Funct. Mater. 29, 1806386 (2019).37. Takeshita, H. et al. Frequency-hopping wave engineering with metasurfaces.Nat. Commun. 15, 196 (2024).38. Sievenpiper, D., Zhang, L., Broas, R. F. J., Alexopolous, N. G. & Yablonovitch, E.High-impedance electromagnetic surfaces with a forbidden frequency band.IEEE Trans. Microw. Theory 47, 2059–2074 (1999).39. Munk, B. A. Frequency Selective Surfaces: Theory and Design (John Wiley & Sons,2005). https://doi.org/10.1002/0471723770.40. Asano, K., Nakasha, T. & Wakatsuchi, H. Simplified equivalent circuit approachfor designing time-domain responses of waveform-selective metasurfaces.Appl. Phys. Lett. 116, 171603 (2020).41. Wakatsuchi, H. et al. Waveform selectivity at the same frequency. Sci. Rep. 5,9639 (2015).42. Wakatsuchi, H. Time-domain filtering of metasurfaces. Sci. Rep. 5, 16737 (2015).43. Baena, J. D. et al. Equivalent-circuit models for split-ring resonators andcomplementary split-ring resonators coupled to planar transmission lines. IEEETrans. Microw. Theory Tech. 53, 1451–1461 (2005).44. Zhou, L., Liu, L. & Shen, Z. High-performance energy selective surface based onthe double-resonance concept. IEEE Trans. Antennas Propag. 69, 7658–7666(2021).45. Bakshi, S. C., Mitra, D. & Teixeira, F. L. Wide-angle broadband rasorber forswitchable and conformal application. IEEE Trans. Microw. Theory 69,1205–1216 (2021).46. Qian, C. et al. Deep-learning-enabled self-adaptive microwave cloak withouthuman intervention. Nat. Photonics 14, 383–390 (2020).47. Wen, E., Yang, X. & Sievenpiper, D. F. Real-time 2-D beamforming with rota-table dielectric slabs enabled by generative neural network. IEEE Trans.Antennas Propag. 70, 8360–8367 (2022).48. Liu, C. et al. A programmable diffractive deep neural network based on adigital-coding metasurface array. Nat. Electron 5, 113–122 (2022).49. Tsiftsis, T. A., Valagiannopoulos, C., Liu, H., Boulogeorgos, A.-A. A. & Miridakis, N.I. Metasurface-coated devices: a new paradigm for energy-efficient and secure6G communications. IEEE Veh. Technol. Mag. 17, 27–36 (2022).50. Vellucci, S., Monti, A., Barbuto, M., Toscano, A. & Bilotti, F. Waveform-selectivemantle cloaks for intelligent antennas. IEEE Trans. Antennas Propag. 68,1717–1725 (2020).51. Barbuto, M. et al. Waveguide components and aperture antennas with fre-quency- and time-domain selectivity properties. IEEE Trans. Antennas Propag.68, 7196–7201 (2020).52. Fathnan, A. A., Homma, H., Sugiura, S. & Wakatsuchi, H. Method for extractingthe equivalent admittance from time-varying metasurfaces and its applicationto self-tuned spatiotemporal wave manipulation. J. Phys. D: Appl. Phys. 56,015304 (2023).53. Tashiro, M., Fathnan, A. A., Sugiura, Y., Uchiyama, A. & Wakatsuchi, H. Meta-surface‐inspired maintenance‐free Internet of things tags characterised in bothfrequency and time domains. Electron. Lett. 58, 937–939 (2022).54. Imani, M. F. & Smith, D. R. Temporal microwave ghost imaging using areconfigurable disordered cavity. Appl. Phys. Lett. 116, 054102 (2020).55. Cui, T. J., Qi, M. Q., Wan, X., Zhao, J. & Cheng, Q. Coding metamaterials, digitalmetamaterials and programmable metamaterials. Light Sci. Appl. 3, e218–e218(2014).56. Tsilipakos, O., Koschny, T. & Soukoulis, C. M. Antimatched electromagneticmetasurfaces for broadband arbitrary phase manipulation in reflection. ACSPhotonics 5, 1101–1107 (2018).57. Valagiannopoulos, C., Tsiftsis, T. A. & Kovanis, V. Metasurface-enabled inter-ference mitigation in visible light communication architectures. J. Opt. 21,115702 (2019).58. Celesti, A. et al. An IoT cloud system for traffic monitoring and vehicularaccidents prevention based on mobile sensor data processing. IEEE Sens. J. 18,4795–4802 (2018).59. Plageras, A. P., Psannis, K. E., Stergiou, C., Wang, H. & Gupta, B. B. Efficient IoT-based sensor BIG Data collection–processing and analysis in smart buildings.Futur. Gener. Comput. Syst. 82, 349–357 (2018).60. Sood, K. et al. Accurate detection of IoT sensor behaviors in legitimate, faultyand compromised scenarios. IEEE Trans. Dependable Secur. Comput. 20,288–300 (2023).61. Harb, H., Mansour, A., Nasser, A., Cruz, E. M. & de la Torre Díez, I. A sensor-baseddata analytics for patient monitoring in connected healthcare applications.IEEE Sens. J. 21, 974–984 (2021).62. Wakatsuchi, H., Anzai, D. & Smartt, C. Visualization of field distributions ofwaveform-selective metasurface. IEEE Antennas Wirel. Propag. Lett. 15, 690–693(2016).63. Hoerl, A. E. & Kennard, R. W. Ridge regression: biased estimation for non-orthogonal problems. Technometrics 42, 80 (2000).Tashiro et al. NPG Asia Materials           (2024) 16:55 Page 10 of 10    55 https://doi.org/10.1002/0471723770 Metasurface-enabled multifunctional single-frequency sensors without external power Introduction Results and discussion Fundamental design theory Numerical demonstration Experimental demonstration Estimation of physical quantities Discussion Conclusion Methods Simulations Measurement samples Measurement method Estimation methods Acknowledgements