# Fileset

[2020_Yakabe_J._Phys._Commun.pdf](https://mdr.nims.go.jp/filesets/7b63077f-ac07-4c11-b5ad-bb87ce116857/download)

## Creator

[Taro Yakabe](https://orcid.org/0000-0002-2244-5890), [Gaku Imamura](https://orcid.org/0000-0002-3130-7190), [Genki Yoshikawa](https://orcid.org/0000-0002-9136-8964), [Masahiro Kitajima](https://orcid.org/0000-0001-9584-190X), [Akiko N Itakura](https://orcid.org/0000-0001-5783-141X)

## Rights

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

## Other metadata

[Hydrogen detection using membrane-type surface stress sensor](https://mdr.nims.go.jp/datasets/e72e3865-c1ae-46cb-8222-af8fbd439f52)

## Fulltext

Hydrogen detection using membrane-type surface stress sensorJournal of Physics CommunicationsPAPER • OPEN ACCESSHydrogen detection using membrane-type surface stress sensorTo cite this article: Taro Yakabe et al 2020 J. Phys. Commun. 4 025005 View the article online for updates and enhancements.This content was downloaded from IP address 144.213.253.16 on 23/03/2020 at 09:24https://doi.org/10.1088/2399-6528/ab7319J. Phys. Commun. 4 (2020) 025005 https://doi.org/10.1088/2399-6528/ab7319PAPERHydrogen detection usingmembrane-type surface stress sensorTaroYakabe1 , Gaku Imamura2, Genki Yoshikawa3,MasahiroKitajima1 andAkikoN Itakura11 ResearchCenter for AdvancedMeasurement andCharacterization, National Institute forMaterials Science (NIMS), 1-2-1 Sengen,Tsukuba, Ibaraki, 305-0047, Japan2 International Center forMaterialsNanoarchitectonics, National Institute forMaterials Science (NIMS), 1-1Namiki, Tsukuba, Ibaraki,305-0044 Japan3 Center for Functional Sensor&Actuator,National Institute forMaterials Science (NIMS), 1-1Namiki, Tsukuba, Ibaraki, 305-0044 JapanE-mail: yakabe.taro@nims.go.jpKeywords: hydrogen,MSS, langmuir, adsorption, absorptionAbstractThis study shows a possibility of the application of amembrane-type surface stress sensor (MSS)with aPdfilm to a hydrogen sensor. It was able to detect hydrogen concentrations from5 to 40000 ppm in anitrogen gasmixture. In the case of a conventional sensor using a hydrogen-occludingmaterial, it isnecessary towait for a state of saturation. In contrast, the proposedmethod can detect hydrogenquickly by the initial rate of hydrogen absorption. The relationship between the initial absorption rateand hydrogen concentration is explained by considering the two-step reaction kinetics of hydrogenabsorption into bulk Pd via Langmuir dissociative adsorption on surface.1. IntroductionHydrogen is known as a clean and renewable energy carrier. The demand for hydrogen is expected to increaseworldwide every year.Hydrogen is produced by reforming natural gas, or the electrolysis of water, biomass, abyproduct of steelworks, and so on. It is supplied to fuel cells, hydrogen automobiles, and power generations.The transport and storage of hydrogenmust be performed in a safemanner. Therefore, it is also indispensable todevelop hydrogen sensors that detect hydrogen leaks. Various techniques of hydrogen detection have beendeveloped so far; these include catalytic, thermal conduction, electrochemical, resistance-based, work-functionbased,mechanical, optical, and acoustic techniques [1]. Here are some details on themechanicalmethodassociatedwith our study.Hydrogen absorptionwould lead to stress formation of somematerials. The stressmeasurements of them are applied to hydrogen sensing. The advantage of sensors by the stressmeasuring is thatthey can be applied to various gas sensors with selecting sensitivefilms. Attempts to usemicro-cantilevers, whichwere known as atomic forcemicroscope probes [2], as stress sensors began in the 90’s [3], and thefirst trial fordetecting hydrogen by using a cantilever sensor was executedwith its adsorption on a Ptfilm [4]. The cantileversensorwas improved by using hydrogen-storagematerials [5, 6]. Hydrogen penetrates into the bulk afterdissociation at the surface [7]. For example, Pd expands by absorbing hydrogen; this hydrogen-absorbed Pd haslow- and high-hydrogen-concentration phases (α andα′, respectively). The lattice constants of Pd, Pd-H (α),and Pd-H(α′) are 0.38874 nm, 0.3895 nm, and 0.4025 nm, respectively [7]. Amembrane-type surface stresssensor (MSS)was developed by optimizing a standard piezoresistive cantilever [8]. It has two-order highersensitivity than a standard piezoresistive cantilever [9]. As shown infigure 1(a), the outline of theMSS design isas follows. The stress of the sensitive film on the disk is transmitted to the four narrowpiezoresistive parts thatsupport the disk and constitute aWheatstone bridge circuit. Thus, the sensor can observe signals according tothe stress with high sensitivity. Studies have used the finite element analysis to study themechanism ofMSS forvarious gas detections through experiments, theoreticalmodels, and numerical simulations [10–12].In this paper, we report on hydrogen-concentrationmeasurements by usingMSSwith a Pd thinfilm.Furthermore, in this study, by focusing on the initial process of hydrogen absorption, we solved theconventional problemof the utilization of a large amount of time formeasuring the saturation state. In addition,the relationship between hydrogen concentration and initial reaction rate is explained simply by the two-stepreaction kinetics of hydrogen absorption into bulk Pd via Langmuir dissociative adsorption on the surface.OPEN ACCESSRECEIVED12November 2019REVISED3 February 2020ACCEPTED FOR PUBLICATION5 February 2020PUBLISHED13 February 2020Original content from thisworkmay be used underthe terms of the CreativeCommonsAttribution 4.0licence.Any further distribution ofthis workmustmaintainattribution to theauthor(s) and the title ofthework, journal citationandDOI.© 2020TheAuthor(s). Published by IOPPublishing Ltdhttps://doi.org/10.1088/2399-6528/ab7319https://orcid.org/0000-0002-2244-5890https://orcid.org/0000-0002-2244-5890mailto:yakabe.taro@nims.go.jphttps://crossmark.crossref.org/dialog/?doi=10.1088/2399-6528/ab7319&domain=pdf&date_stamp=2020-02-13https://crossmark.crossref.org/dialog/?doi=10.1088/2399-6528/ab7319&domain=pdf&date_stamp=2020-02-13http://creativecommons.org/licenses/by/4.0http://creativecommons.org/licenses/by/4.0http://creativecommons.org/licenses/by/4.02. ExperimentalmethodsFigure 1(a) showsMSSwith a Pd film. Thefilmwas deposited by an electron beambombardment, the target wasa 99.9%Pdnugget, and the deposition ratewas approximately 0.1 nm s−1. Thefilm thickness was 20 nm,whichwasmeasured using a quartz crystal thicknessmonitor during evaporation. Figure 1(b) shows our experimentalset up, in which two kinds of sensing-gas sources of hydrogen and nitrogenwere used in cylinders, the hydrogenconcentrations of whichwere 40000 ppm (4.00%)±100 (0.01%) ppmand 100 ppm±10 ppm, respectively.Each of the sensing gas lines was further dilutedwith pure nitrogen (99.999%) from the other regulating gas line.The hydrogen concentration in theMSS chamber was controlled by twomass-flow controllers (MFC, SEC-N112MGMofHoriba STEC). The amount of the flowing gaswas 30 sccm.We varied the concentration ofhydrogen from2000 ppm to 40000 ppmby using the 40000-ppmhydrogen in one cylinder and from5ppm to100 ppmby using the 100-ppmhydrogen in the other cylinder. TheMSSwith a Pd filmwasmounted in theMSSchamber sealedwithO-rings. TheMSS chamber and gas linewere placed in a heated incubatorwith ribbonheaters (orange-colored region infigure 1(b)). The temperaturewas controlled at 333.15 K±0.02 K. For eachconcentration of hydrogen, one experimental procedure consisted of three cycles of 1-h hydrogen injection and3-h pure nitrogen purge, with the stress beingmeasured using theMSS. The Pdfilm immediately afterdeposition showed large stress and low reproducible signals during hydrogen exposure. The low reproducibilitywas due to the hysteresis of hydrogen absorption and desorption [13]. Beforemeasuring the hydrogenconcentration, we performed absorption and desorptionmore than 10 times by using the 40000-ppmhydrogengas to suppress hysteresis. The stress signals were read as electrical voltages ofMSS in the sampling time of 1 s (i.e., 1-Hzmeasurement).3. Results and discussionsFigure 2 shows theMSS signals for hydrogen concentration from2000 to 40000 ppm. Infigure 2(a), thehorizontal axis represents time t, and the vertical axes represent the signals ofMSS (V ) and the concentration ofhydrogen (C) under theMFC regulation. These curves show the hydrogen concentrations of 2000 ppm (black),10000 ppm (red), 20000 ppm (blue), and 40000 ppm (green). The signals showed noticeable increase just after10800 s, 25200 s, and 39600 s, which corresponded to regulated hydrogen injection times. The signals decreasedFigure 1. (a)Opticalmicroscope image ofMSSwith a Pdfilm. The diameter of the disk is 300μm. (b) Schematic of our experimentalsystem.Hydrogen concentrationwith nitrogen is adjusted by twomass-flow controllers (MFCs).Figure 2.MSS signals for hydrogen concentrations of 2000 ppm (black), 10000 ppm (red), 20000 ppm (blue), and 40000 ppm (green).(a)Observed voltage signals ofMSS (V ) and concentration of hydrogen (C)with time (t). (b)Numerical differentiation (dV/dt) of thevoltage signal curves in (a). (c)Enlargement around the upward convex peaks in (b).2J. Phys. Commun. 4 (2020) 025005 TYakabe et aljust after 14400 s, 28800 s, and 43200 s, which are the nitrogen purging times. In the observation of the changesin the signals of the hydrogen injections for 1 h, the signal for 40000 ppm seemed to approach saturation but theother signals for 20000 ppm, 10000 ppm, and 2000 ppmneededmore time to saturate. It was difficult to usethese signals directly for determining the concentration because saturation took�1 h and gradual changes wereobserved in the baseline. Therefore, we focused on the sharp changes over time andmade numericaldifferentiations, as shown infigure 2(b), by using a standard data analysis and graphing software. The numericalsmoothingwas performed using a second-order polynomial and 101 adjacent points, corresponding to±50 s.The change of theMSS signal is proportional to the stress of the sensitive film [8, 9]. which in turn is proportionalto the amount of absorbed hydrogen [5, 6]. Therefore, the time differentiation of theMSS signal corresponds tothe hydrogen-absorption rate. The upward and downward convex peaks are clearly observed immediately afterthe hydrogen injection and nitrogen purge, respectively. Figure 2(c) shows enlargements around the upwardconvex peaks in fugure 2(b); these enlargements correspond to the changes in the hydrogen absorption rate inthe initial process. The peak in the green curve (40000-ppmhydrogen) is the largest, while those in the blue(20000 ppm), red (10000 ppm), and black (2000 ppm) curves decreased. The peak heights immediately after thefirst injection of hydrogenwere approximately 20%higher than those after the second and third injections.Weconsider that this difference is caused by the remaining hysteresis of hydrogen absorption and desorption in Pd.Furthermore, we confirmed that theMSS detected signals at lower hydrogen concentrations. Figure 3 showthe signals ofMSS for hydrogen concentration from5ppm to 100 ppm. Infigure 3(a), the horizontal and verticalaxes are the same as those infigure 2(a). These curves show the hydrogen concentrations of 5 ppm (black), 25ppm (red), 50 ppm (blue), and 100 ppm (green). The signals do not show sudden increase and decrease as thoseinfigure 2(a), but slowly increasedwith hydrogen injections. Figure 3(b) shows numerical differentiation byusing the samemethod and the same number of adjacent points as that used in figure 2(b). The peaks at 100ppm, 50 ppm, and 25 ppmare clearly defined and their heights are reproducible. The peak at 5-ppmhydrogeninjection is not clear and appears buried in noise.Whenwe adopted 1001 adjacent points, remarkable peakswere defined, as shown infigure 3(c). These show a trade-off between the detectable region size of the hydrogenconcentrations and the brevity of the necessary time required for the concentrationmeasurement. From apractical point of view, suitable number of adjacent points should be selected.Figure 4 shows the relation between hydrogen concentrations and the peak heights of differentiations infigures 2(c) and 3(b). The peak height at each concentration from25 to 40000 ppmupon hydrogen injection isplotted. The peak heights immediately after the first, second, and third hydrogen injections are represented byblack squares, red circles, and blue triangles, respectively, infigures 2(c) and 3(b). These points arefitted by theequation (6), and this physicalmechanism is explained as follows. Assuming a two-step reaction process infigure 5, hydrogenmolecules are dissociated into atoms on the Pd surface and each atom is absorbed into thebulk [14–16]. The process is represented by the following chemical reaction formula:( ) ⟶ ( ) ⟶ ( )+ -¬ ¾¾ ¬ ¾¾-+-+HPd s12H Pd s PdH, 1k k2k k1122where Pd(s)-H, PdH, and ki (I=1+, 1−, 2+, 2−) imply the status ofH-atom adsorption on the Pd surface, thestatus ofH-atom storage in Pd bulk, and the reaction-rate constants, respectively (figure 5).We assume that (i)reaction 1 ismuch faster than reaction 2 (i.e., reaction 2 is the rate-determining step), (ii)Pd(s)-H is Langmuirdissociated adsorption, and (iii) the concentration of PdH is zero (i.e., [PdH]∼0) in the initial state. Fromassumptions (i) and (ii),Figure 3.MSS signals for hydrogen concentrations of 5 ppm (black), 25 ppm (red), 50 ppm (blue), 100 ppm (green). (a)Observedvoltage signals ofMSS (V ) and hydrogen concentration (C)with time (t). (b)Numerical differentiation (dV/dt) in (a). (c)Numericaldifferentiation of 5-ppm curves in (a).More adjacent points were adopted.3J. Phys. Commun. 4 (2020) 025005 TYakabe et al[ ( ) ] ( )- µ+K CK CPd s H1, 2whereC is the concentration of hydrogen in themixture gas with nitrogen and is proportional to the partialpressure of hydrogen, andK=k1+/k1− [17]. Based on the rate equation of reaction 2 and assumption (iii),[ ] [ ( ) ] [ ][ ( ) ] ( )= - --+ -+ddtk kkPdHPd s H PdHPd s H . 32 22As PdH is used to input stress to theMSS and the change in theMSS signal (V ) is proportional to the stress [11],[ ] ( )- µV V PdH , 40Figure 4. Square root of hydrogen concentration (C1/2) versus localmaximumpeak height of differential (dV/dt [max]). Themaximumvalue of the derivative at each concentration from25 ppm to 40000 ppmupon hydrogen injection is plotted. The peakheights after thefirst (black square), second (red circle), and third (blue triangle)hydrogen injections in figures 2(c) and 3(b). Thesepoints arefitted by the Langmuir dissociative adsorptionmodel.Figure 5.Model schematic: hydrogenmolecules are adsorbed on the surface and dissociated atoms are absorbed into the Pdfilm. Theadsorbed hydrogens expand the lattice and the stress is caused. The blue andwhite atoms representH and Pd, respectively.4J. Phys. Commun. 4 (2020) 025005 TYakabe et alwhereV0 is the initial signal. By differentiating bothmembers,[ ] ( )µdVdtddtPdH. 5From equations (2)), and (5), we get( )=+dVdtaK CK C1. 6Thefitting curves of equation (6) are presented infigure 4, parameterK has a common value of 0.0080 foreach curve, a =0.0011 in the solid curve representing the first hydrogen injection, and a =0.00091 in thedashed curve representing second and third hydrogen injections. The relation between the hydrogenconcentrations and localmaximumpeak heights of the differentiations of signals is explained using the kineticmodel of hydrogen absorption to Pd bulk via Langmuir dissociative adsorption on the Pd surface.4. ConclusionWeperformed hydrogen sensing by using theMSSwith a 20-nmPdfilm. The absorption kinetics could beelucidated by the two-step reaction kinetics of hydrogen absorption into bulk Pd via Langmuir dissociativeadsorption on the surface. It was possible to analyze the concentration of hydrogen quantitatively from25 to40000 ppm. Themerit of using the differentiations to determine hydrogen concentration is the rapid detection,which does not require waiting for signal saturation. The hydrogenMSS sensor could be improved using lowhysteresismaterials instead of Pd. AsMSS is a precise stress sensor, it will be useful for development of hydrogenmaterials.Moreover,MSS is a good tool to study kinetics on a surface or bulk.AcknowledgmentsWeare grateful to C.Nishimura for scientific discussions andA.Ohi, T.Ohki, andN. Ikeda for technicalsupports inNational Institute forMaterials Science. This work supported byMEXT/JPSJ KAKENHIGrantNumber JP 18H03849 and JST-Mirai ProgramGrantNumber JPMJMI18A3.ORCID iDsTaro Yakabe https://orcid.org/0000-0002-2244-5890References[1] Hubert T, Boon-Brett L, BlackG andBanachU 2011Hydrogen sensors—A review Sensors and Actuators B-Chemical 157 329–52[2] BinnigG,Quate C F andGerber C 1986Atomic forcemicroscope Phys. Rev. Lett. 56 930[3] Gimzewski J K,Gerber C,Meyer E and Schlittler RR1994Observation of a chemical-reaction using amicromechanical sensorChem.Phys. Lett. 217 589[4] LangHP et al 1998A chemical sensor based on amicromechanical cantilever array for the identification of gases and vaporsAppliedPhysics a-Materials Science&Processing 66 S61–4[5] Okuyama S,Mitobe Y,OkuyamaK andMatsushita K 2000Hydrogen gas sensing using a Pd-coated cantilever Japanese Journal ofApplied Physics Part 1-Regular Papers Short Notes&Review Papers 39 3584–90[6] ChouY I, ChiangHC andWangCC2008 Study onPd functionalization ofmicrocantilever for hydrogen detection promotion SensorsandActuators B-Chemical 129 72–8[7] Manchester FD, San-Martin A and Pitre JM1994TheH-Pd (hydrogen-palladium) SystemJ.M. J. Phase Equilibria 15 62[8] YoshikawaG, AkiyamaT, Gautsch S, Vettiger P andRohrerH2011Nanomechanicalmembrane-type surface stress sensorNano Lett.11 1044–8[9] YoshikawaG, AkiyamaT, Loizeau F, ShibaK, Gautsch S,NakayamaT, Vettiger P, de Rooij N F andAonoM2012Twodimensionalarray of piezoresistive nanomechanicalmembrane-type surface stress sensor(MSS)with improved sensitivity Sensors 12 15873–87[10] ImamuraG, ShibaK andYoshikawaG2016 Smell identification of spices using nanomechanicalmembrane-type surface stress sensorsJapan. J. Appl. Phys. 55 1102B3[11] ImamuraG, ShibaK andYoshikawaG2016 Finite element analysis on nanomechanical detection of small particles : toward virusdetection Frontiers inMicrobiology 7 488[12] ImamuraG, ShibaK, YoshikawaG andWashio T 2018Analysis of nanomechanical sensing signals; physical parameter estimation forgas identificationAIPAdv. 8 075007[13] WangD, Flanagan TB andKuji T 2002Hysteresis scans for Pd-H and Pd-alloy-H systems Phys. Chem. Chem. Phys. 4 4244–54[14] KayBD, PedenCHF andGoodmanDW1986Kinetics of hydrogen absorption by PD(110)Physical Review B 34 817–21[15] FayeO and Szpunar J A 2018An efficient way to suppress the competition between adsorption of h2 and desorption of nH2–Nbcomplex fromgraphene sheet: a promising approach toH2 storage J. Phys. Chem. C 122 28506–17[16] FayeO, Szpunar J A, Szpunar B andBeye AC2017Hydrogen adsorption and storage on Palladium—functionalized graphenewithNH-dopant: afirst principles calculationAppl. Surf. Sci. 392 362–745J. Phys. Commun. 4 (2020) 025005 TYakabe et alhttps://orcid.org/0000-0002-2244-5890https://orcid.org/0000-0002-2244-5890https://orcid.org/0000-0002-2244-5890https://orcid.org/0000-0002-2244-5890https://doi.org/10.1016/j.snb.2011.04.070https://doi.org/10.1016/j.snb.2011.04.070https://doi.org/10.1016/j.snb.2011.04.070https://doi.org/10.1103/PhysRevLett.56.930https://doi.org/10.1016/0009-2614(93)E1419-Hhttps://doi.org/10.1007/s003390051100https://doi.org/10.1007/s003390051100https://doi.org/10.1007/s003390051100https://doi.org/10.1143/JJAP.39.3584https://doi.org/10.1143/JJAP.39.3584https://doi.org/10.1143/JJAP.39.3584https://doi.org/10.1016/j.snb.2007.07.072https://doi.org/10.1016/j.snb.2007.07.072https://doi.org/10.1016/j.snb.2007.07.072https://doi.org/10.1007/BF02667685https://doi.org/10.1021/nl103901ahttps://doi.org/10.1021/nl103901ahttps://doi.org/10.1021/nl103901ahttps://doi.org/10.3390/s121115873https://doi.org/10.3390/s121115873https://doi.org/10.3390/s121115873https://doi.org/10.7567/JJAP.55.1102B3https://doi.org/10.3389/fmicb.2016.00488https://doi.org/10.1063/1.5036686https://doi.org/10.1039/B201271Phttps://doi.org/10.1039/B201271Phttps://doi.org/10.1039/B201271Phttps://doi.org/10.1103/PhysRevB.34.817https://doi.org/10.1103/PhysRevB.34.817https://doi.org/10.1103/PhysRevB.34.817https://doi.org/10.1021/acs.jpcc.8b09498https://doi.org/10.1021/acs.jpcc.8b09498https://doi.org/10.1021/acs.jpcc.8b09498https://doi.org/10.1016/j.apsusc.2016.09.032https://doi.org/10.1016/j.apsusc.2016.09.032https://doi.org/10.1016/j.apsusc.2016.09.032[17] Masel R I 1996Principles of Adsorption andReaction on Solid Surfaces (Wiley Interscience) p. 2446J. Phys. Commun. 4 (2020) 025005 TYakabe et al 1. Introduction 2. Experimental methods 3. Results and discussions 4. Conclusion Acknowledgments References