scieee Open visual document viewer

A novel vacuum pressure sensor using a thermoelectric device

Beltrán-Pitarch, Braulio; García-Cañadas, Jorge

Abstract

We present the proof-of-concept of a new vacuum pressure sensor based on a new operating principle. The new sensor is formed by the simple combination of a thermoelectric (TE) module that is contacted at both sides by a bent copper plate. The vacuum pressure is related to the change in the thermal contact resistance that exists between the outer ceramic surfaces of the TE module and the copper plate, since heat transfer through the ceramic/copper interface is found to be influenced by the amount of air present in the interface gaps (vacuum pressure). The variations of the thermal contact resistance produce a change of the TE module voltage when a fixed current is applied to it. By monitoring this voltage simultaneously to the response of a commercial pressure sensor at different vacuum pressures inside a vacuum chamber, a calibration equation was identified, which enables obtaining the vacuum pressure from the voltage signal. Random errors lower than 10% were found in the 0.1 to 250 mbar range, which is the pressure range that the sensor can properly sense. This new device is inexpensive, simple to fabricate and integrate, and benefits from the high stability of TE modules.

Full text

1 A no el acuum p essu e senso using a he moelec ic de ice B aulio Bel án-Pi a ch* and Jo ge Ga cía-Cañadas Depa men o Indus ial Sys ems Enginee ing and Design, Uni e si a Jaume I, Campus del Riu Sec, 12071 Cas ellón, Spain *e-mail: bel [email protected] We p esen he p oo -o -concep o a new acuum p essu e senso based on a new ope a ing p inciple. The new senso is o med by he simple combina ion o a he moelec ic (TE) module ha is con ac ed a bo h sides by a ben coppe pla e. The acuum p essu e is ela ed o he change in he he mal con ac esis ance ha exis s be ween he ou e ce amic su aces o he TE module and he coppe pla e, since hea ans e h ough he ce amic/coppe in e ace is ound o be in luenced by he amoun o ai p esen in he in e ace gaps ( acuum p essu e). The a ia ions o he he mal con ac esis ance p oduce a change o he TE module ol age when a ixed cu en is applied o i . By moni o ing his ol age simul aneously o he esponse o a comme cial p essu e senso a di e en acuum p essu es inside a acuum chambe , a calib a ion equa ion was iden i ied, which enables ob aining he acuum p essu e om he ol age signal. Random e o s lowe han 10% we e ound in he 0.1 o 250 mba ange, which is he p essu e ange ha he senso can p ope ly sense. This new de ice is inexpensi e, simple o ab ica e and in eg a e, and bene i s om he high s abili y o TE modules. Keywo ds: Vacuum gauge, Pel ie de ice, he moelec ic module, he mal con ac esis ance. I. INTRODUCTION A he moelec ic (TE) de ice, also known as a TE module o Pel ie de ice, can gene a e a empe a u e di e ence ac oss i when cu en is applied o i (Pel ie e ec ). In addi ion, he de ice ol age a open ci cui is p opo ional o he empe a u e di e ence ac oss he TE ma e ials (Seebeck e ec ). TE modules a e o med by al e na ed p- ype and n- ype semiconduc o legs elec ically connec ed in se ies and he mally in pa allel [1,2]. This assembly o TE legs is sandwiched by wo elec ically insula ing pla es ( ypically a ce amic ma e ial). Due o he Seebeck e ec , TE de ices a e a ec ed by he he mal p ocess occu ing in hei su oundings. This o e s he possibili y o sense he mal phenomena by he di ec use o elec ical signals. Taking ad an age o his ea u e di e en applica ions o TE de ices can be encoun e ed in he li e a u e, such as hei use as he mal conduc i i y and he mal di usi i y senso s [3,4], as de ices able o measu e he con ec ion hea ans e coe icien (h) [5], he emissi i y (ε) [6], and also he he mal con ac esis ance [7]. Rela ed o ou s udy on he he mal con ac esis ance [7], i can be seen om elec ical impedance measu emen s pe o med o a comme cial TE de ice (Fig. 1) ha he he mal con ac esis ance be ween he TE de ice and coppe blocks is hugely in luenced by a ia ions 2 in he ai p essu e ( acuum le el). When he module was sandwiched be ween wo coppe blocks wi hou using a he mal in e ace ma e ial (TIM), he size o he semici cle esponse ob ained expe ienced a huge inc ease, which was only due o he a ia ion o he ai p essu e (see measu emen s om he con ac ed sys em in Fig. 1). Fig. 1. Impedance spec oscopy measu emen s, a ambien p essu e and in acuum (<10-3 mba ), o a he moelec ic de ice unde wo di e en condi ions: suspended and con ac ed by coppe blocks, as schema ically shown in he inse . Rema kably, he esponse ound in acuum was nea ly he same as ha om he module suspended in ai , i.e. wi hou being in con ac wi h any solid a all, which highligh s he poo hea conduc ion h ough he coppe -TE de ice he mal con ac when no TIM is p esen . In addi ion, i is e idenced ha he ai illing he gaps in he he mal con ac is he pa ame e go e ning he hea ans e h ough he in e ace. A u he inc ease in he semici cle esponse wi h espec o he TE module measu ed suspended in ai can be also obse ed when he module was measu ed suspended in acuum, which indica es he low hea losses om he TE de ice p oduced by he na u al con ec ion. These obse a ions poin o he possibili y o ab ica ing a acuum p essu e senso based on a he mal con ac c ea ed on a TE de ice, which ep esen s a no el ope a ing p inciple wi h espec o he exis ing echnologies, such as he Pi ani [8] and he he mopile senso s [9]. In bo h es ablished echnologies, a piece o ma e ial ( ypically a pla e o a wi e) is hea ed up by and elec ical cu en (Joule e ec ) and he empe a u e ha i eaches is eco ded. The empe a u e ise in hese echnologies is go e ned by con ec ion losses, which depend on he acuum p essu e o he gas p esen in he su oundings o he hea ed piece. In addi ion o hese wo common acuum senso s, o he ype o senso s ha e been de eloped which a e capable o de e mine he acuum p essu e by a change in elec ical capaci y [10] o he amoun o ionized gas [11,12], e en hough hey a e mo e complex and no ypically used [13]. 3 Fu he mo e, he e a e o he senso s ha measu e he acuum p essu e using a he mocouple as a hea e and empe a u e senso simul aneously [14,15], o he signal o a mic o- he mocouple o TE de ice wi h oscilla ing hea ing and cooling p ocesses [4,16]. Howe e , none o hem ha e a he mal con ac as hei key pa and a e no based on he mal changes occu ing by he amoun o gas enclosed in he he mal con ac . The main objec i e o his wo k is o demons a e he p oo -o -concep o a no el acuum p essu e senso which is based on he changes expe ienced in he elec ical signals o a TE module he mally con ac ed by a solid a bo h sides. The a ia ions o he elec ical signals a e p oduced by changes in he amoun o gas ( acuum p essu e) p esen in he gaps o he he mal con ac esis ance a he TE module/me al in e aces. Fi s , he undamen al o he ope a ing p inciple o he new acuum senso is explained. Then, he concep is p o ed expe imen ally by ab ica ing he senso and es ing i s ol age a ia ions a di e en acuum p essu es in a acuum chambe . II. OPERATING PRINCIPLE The ol age di e ence ac oss a TE module is gi en by [17], 𝑉 =𝑅Ω𝐼+𝑆∆𝑇 (1) Whe e RΩ is he o al ohmic esis ance o he TE module (accoun ing o he elec ical esis ance o wi es, con ac s, and semiconduc o ma e ials), I is he elec ic cu en applied o i , S he module Seebeck coe icien , and ∆T=T(L)-T(0) he empe a u e di e ence be ween he ends o he TE legs, being L he leng h o he TE legs (see Fig. 2). Bo h RΩ and S a e module’s p ope ies, which a e empe a u e dependen and hence can be conside ed cons an s i no signi ican changes occu in he ambien empe a u e. Unde his assump ion and i a cons an cu en is supplied o he TE module, only changes in ∆T will a ec he TE de ice ol age, as s a ed in Eq. (1). I a TE module is he mally con ac ed a bo h sides by a ben coppe pla e wi hou any he mal in e ace ma e ial (as shown in Fig. 2 and Fig. 3), plen y o ai gaps will be p esen in he he mal con ac in e aces (see inse o Fig. 2). No e ha a he he mal con ac be ween wo solids he eal con ac a ea may be as low as 1-2% o he o al a ea [18]. When he module is suspended in ai and a cons an cu en is applied unde cons an ambien empe a u e, i s ΔT signi ican ly depends on he he mal con ac esis ance a he ce amic pla es/coppe in e aces, as schema ically shown in Fig. 2. The cons an cu en low p oduces he ΔT due o he Pel ie e ec , i.e. hea is emo ed a one end o he legs and eleased a he o he . Fo an in ini e he mal con ac esis ance, he ΔT will be he highes possible, since he Pel ie hea emo ed/ eleased a he junc ions will no be e acua ed h ough he coppe pla e (do ed line case o Fig. 2). Howe e , i he he mal con ac esis ance is dec eased, hea is able o c oss he in e ace and hence o be ex ac ed h ough he coppe pla e, educing he ΔT (dashed line case o Fig. 2). Fo an ideal con ac (no he mal con ac esis ance), all he hea eaching he in e ace will be ex ac ed and hus could no be main ained in 4 he TE de ice, consequen ly he empe a u es a he Cu-TE module in e aces do no change and a e simila o he si ua ion wi hou cu en low (solid line case in Fig. 2). Fig. 2. Schema ic iew o he empe a u e p o iles ob ained in a TE module con ac ed by a ben coppe pla e when a cons an cu en is applied. Fo simplici y, he TE module is ep esen ed by only one leg. The p o iles when cu en is no applied ( hin g ey ho izon al line), o an ideal pe ec con ac (black solid line), o a eal con ac (dashed line), and o an in ini e he mal con ac esis ance (do ed line) a e indica ed. Tempe a u e in he coppe pla e is always he same as he ini ial. The inse shows a schema ic iew o he ai gaps and he eal con ac poin s in he he mal con ac . No e ha he ac o ha ing he ben coppe pla e p o okes ha he empe a u e in his pla e do no change in any o he cases, since due o i s high he mal conduc i i y and he symme y o he sys em, he same amoun o hea added a he ho side, is quickly cancelled by he same amoun o hea emo ed a he cold side. Unde he condi ions desc ibed, and as obse ed in Fig. 1, changes in he ol age (and hus in ΔT) a e only possible i he amoun o ai is a ied in he ai gaps ( a ia ions in he acuum p essu e), which o e s he possibili y o link his pa ame e o he acuum p essu e. III. EXPERIMENTAL PROOF-OF-CONCEPT In o de o expe imen ally p oo he concep o he new acuum p essu e senso , a coppe pla e o 10 mm x 20 mm x 0.3 mm dimensions was sligh ly polished wi h 400 g i size silicon ca bide sandpape (Re . 30-5208-012, Buehle ) and ca e ully ben manually in o de o he mally con ac he wo ou e ce amic su aces o a comme cial 7 mm x 7 mm Bi-Te TE module (In e m, e . CBM87), as shown in Fig. 3. In o de o ensu e a sui able and homogeneous he mal con ac along he whole con ac a ea se e al s eps we e ollowed as shown in Fig. 4. Fi s , he coppe ba was ben while being held by a me allic ba o sc ew o simila heigh o ha o he TE module a 5 he middle. Once he bending is comple ed, he dis ance be ween he coppe pla e ends should be somewha smalle han ha o he TE module. This is o gene a e some o e p essu e once he module is in oduced inside he ben coppe pla e which is he inal s ep.Fig. 4 Fig. 3. (a) Schema ic iew and (b) eal pic u e o he side iew o he TE module wi h he ben coppe shee con ac ed. Fig. 4. S eps ollowed o ab ica e he acuum senso . The ab ica ed senso was suspended inside a cubic acuum chambe o 400 mm side. A PGSTAT30 po en ios a (Me ohm Au olab B. V.) con olled by No a 1.11 so wa e was used o apply a 50 mA cons an cu en o he module and measu e i s ol age. I should be no ed ha any o he powe sou ce able o apply a cons an cu en and a mul ime e o measu e he ol age migh be used. Cables o he measu emen o he elec ical signals we e inse ed h ough a eed h ough in he acuum chambe . Simul aneously o he ol age measu emen , he acuum p essu e o he chambe was egis e ed using a comme cial p essu e senso (P ei e , e . PKR 251) di ec ly ins alled a one o he acuum chambe walls. Di e en acuum p essu e le els we e achie ed in he chambe using a high acuum pump uni (P ei e , HiCube 80 Eco). Be o e a ying he p essu e o he acuum chambe , he 50 mA cons an cu en was applied o he TE de ice, and i was wai ed un il a s eady s a e alue o he ol age signal was eached. The small cons an cu en alue was used o a oid signi ican Joule e ec , which may inc ease he empe a u e o he TE legs and change hei TE 6 p ope ies. Once he s eady s a e was eached, he acuum pump was ac i a ed and he ol age o he TE module was measu ed simul aneously o he acuum p essu e om he comme cial senso . The p essu e change was ins an aneously ollowed by he ol age. The expe imen was epea ed h ee imes. Resul s a e shown in Fig. 5. I can be obse ed ha he de ice ol age shows a signi ican change om ambien p essu e up o 0.1 mba . Howe e , a p essu e le els lowe han 0.1 mba he ol age esponse shows no signi ican change. A good epea abili y be ween he h ee expe imen s pe o med was also obse ed. Fig. 5. Measu ed ol age o he TE module a di e en alues o acuum p essu e o h ee di e en expe imen s. A calib a ion cu e (solid line) ob ained by a i ing o all he poin s o he h ee expe imen s is als shown. A i ing o all he poin s om he h ee expe imen s was pe o med using MATLAB, leading o he equa ion, 𝑃 =( 0.0762 𝑉−0.0862−5.58)1 0.734 (2) whe e P is he acuum p essu e (gi en in mba uni s), and V is gi en in V. The e o s o he ou i ed pa ame e s ob ained om he i ing can be ound in Table 1. All o hem a e lowe han 2%, p o ing he sui abili y o Eq. (2), which can be used o de e mine he acuum p essu e om a measu ed ol age alue o he new acuum p essu e senso . 7 Table 1. E o s om he di e en pa ame e s ob ained om he i ing o he 𝑃 =(𝑏 𝑉−𝑎 −𝑐)1 𝑑 equa ion. Pa ame e E o (%) a 0.08 b 1.79 c 1.99 d 1.43 In o de o calcula e he o al combined andom e o uc in he de e mina ion o P, which is he combined e o aking in o accoun all he con ibu ions om he e o s om he ou i ed pa ame e s (Table 1), he ollowing equa ion was used [19], 𝑢𝑐2=∑(𝜕𝑃 𝜕𝑥𝑖)2 𝑁 𝑖=1 𝑢(𝑥𝑖)+2∑ ∑ 𝜕𝑃 𝜕𝑥𝑖 𝜕𝑃 𝜕𝑥𝑗𝑢(𝑥𝑖,𝑥𝑗) 𝑁 𝑗=𝑖+1 , 𝑁−1 𝑖=1 (3) whe e N is he numbe o i ing pa ame e s (4 in ou case), xi=a, b, c, o d and u(xi) deno e he i ing pa ame e s and hei associa ed unce ain ies (shown in Table 1), espec i ely. Finally, u(xi,xj) is he unce ain y o he in e dependen i ed pa ame e s, which is di ec ly ob ained om he non-diagonal elemen s o he symme ic co a iance ma ix. This ma ix was p o ided by he i ing p ocedu e, which was pe omed using O igin 2018 so wa e. The o al combined andom e o can be seen in Fig. 6 o he p essu e ange whe e signi ican a ia ions o ol age occu wi h p essu e. The e o adop s di e en alues a di e en p essu es due o he non-linea i y o he esponse. I shows alues lowe han 10% om app oxima ely 0.1 o 250 mba , which can be conside ed he use ul p essu e ange o p ac ical p essu e measu emen s by he no el senso . 8 Fig. 6. Combined andom e o a di e en p essu e alues o he no el acuum p essu e senso . I should be no ed ha he ob ained calib a ion cu e [Eq. (2)] is only alid a he empe a u e a which he expe imen s we e pe o med (25.0 ± 1.5 °C), since he TE p ope ies o he TE module change wi h signi ican empe a u e a ia ions as p e iously discussed. In any case, he calib a ion can be epea ed a di e en empe a u es and ob ain he i ing pa ame e s as a unc ion o empe a u e o a p ecise measu emen a any ope a ional empe a u e. On he o he hand, he shape and dimensions o he in e ace gaps we e no con olled du ing senso ab ica ion, consequen ly, a me hodology o con ol his should be de eloped o ensu e epea abili y, o all he ab ica ed de ices should unde go an indi idual calib a ion. Finally, we would like o ema k he se e al ad an ages o he no el senso , such as low cos (a ound 35 €, which is he app oxima e p ice o he TE module and he coppe pla e), simple ab ica ion (i only in ol es he gen le polishing o he coppe pla e and i s bending o con ac he de ice ou e ce amic su aces apa om he ab ica ion o he de ice), and easy in eg a ion (only equi es he applica ion o a ixed cu en and egis e ing a ol age). IV. CONCLUSIONS The p oo -o -concep o a new acuum p essu e senso based on a he moelec ic module has been demons a ed. The senso elies on a new ope a ing p inciple in which he acuum p essu e is measu ed om he change in he he mal con ac esis ance ha exis s be ween he ou e ce amic su aces o he module and a he mally con ac ed me allic solid, since hea ans e h ough he ce amic/me al in e ace is in luenced by he amoun o gas p esen in he in e ace gaps ( acuum p essu e). The a ia ions o he he mal con ac esis ance p oduce a change 9 o he he moelec ic module ol age when a ixed cu en is applied o i unde cons an ambien empe a u e. The senso is buil by simply con ac ing bo h sides o he he moelec ic de ice by a ben coppe pla e. The ol age o he new senso and he acuum p essu e, measu ed using a comme cial senso , we e simul aneously eco ded se e al imes. A good epea abili y be ween expe imen s was ound, and a calib a ion equa ion, able o p o ide he acuum p essu e om he he moelec ic de ice ol age eading, was ob ained. To al combined andom e o s lowe han 10% we e ob ained om 0.1 o 250 mba , which is he p essu e ange ha he senso can p ope ly sense. This new de ice is inexpensi e, simple o ab ica e and in eg a e, and bene i s om he high s abili y o TE modules. ACKNOWLEDGMENTS The au ho s acknowledge inancial suppo om he Spanish Agencia Es a al de In es igación unde he Ramón y Cajal p og am (RYC-2013-13970), om he Gene ali a Valenciana and he Eu opean Social Fund unde he ACIF p og am (ACIF/2018/233), om he Uni e si a Jaume I unde he p ojec UJI-A2016-08, and he echnical suppo o Raquel Oli e Valls and José O ega He e os. REFERENCES [1] W. He, G. Zhang, X. Zhang, J. Ji, G. Li, X. Zhao, Recen de elopmen and applica ion o he moelec ic gene a o and coole , Appl. Ene gy. 143 (2015) 1–25. doi:10.1016/j.apene gy.2014.12.075. [2] D.K. Aswal, R. Basu, A. Singh, Key issues in de elopmen o he moelec ic powe gene a o s: High igu e-o -me i ma e ials and hei highly conduc ing in e aces wi h me allic in e connec s, Ene gy Con e s. Manag. 114 (2016) 50–67. doi:10.1016/j.enconman.2016.01.065. [3] B. Bel án-Pi a ch, L. Má quez-Ga cía, G. Min, J. Ga cía-Cañadas, Measu emen o he mal conduc i i y and he mal di usi i y using a he moelec ic module, Meas. Sci. Technol. 28 (2017) 045902. [4] C. Machu , E. Ga io , P. God s, S. Sakly, D. Lecle cq, A new pel ie senso o measu ing he he mal conduc i i y o luids, IEEE T ans. Ins um. Meas. 47 (1998) 398–402. doi:10.1109/19.744181. [5] B. Bel án-Pi a ch, J. Ga cía-Cañadas, In luence o con ec ion a ou e ce amic su aces on he cha ac e iza ion o he moelec ic modules by impedance spec oscopy, J. Appl. Phys. 123 (2018) 084505. doi:10.1063/1.5019881. [6] R. Mesalam, H.R. Williams, R.M. Amb osi, J. Ga cía-Cañadas, K. S ephenson, Towa ds a comp ehensi e model o cha ac e ising and assessing he moelec ic modules by impedance spec oscopy, Appl. Ene gy. 226 (2018) 1208–1218. doi:10.1016/j.apene gy.2018.05.041. [7] B. Bel án-Pi a ch, F. Vidan, J. Ga cía-Cañadas, Cha ac e iza ion o he mal con ac s be ween hea exchange s and a he moelec ic module by impedance spec oscopy, Appl. The m. Eng. (2019) 114361. doi:10.1016/J.APPLTHERMALENG.2019.114361. [8] M. Pi ani, , D sch. Phys. Gesellscha . 8 (1906) 24. [9] A.W. Van He waa den, D.C. Van Duyn, J. G oeneweg, Small-size acuum senso s based on silicon he mopiles, Senso s Ac ua o s A Phys. 27 (1991) 565–569. doi:10.1016/0924-4247(91)87052-5. [10] Y. Wang, M. Esashi, The s uc u es o elec os a ic se o capaci i e acuum senso s, Senso s Ac ua o s A Phys. 66 (1998) 213–217. doi:10.1016/S0924-4247(98)00037-5. [11] S. Wil e , C. Edelmann, Field emi e -based acuum senso s, Vacuum. 86 (2012) 556–571. doi:10.1016/J.VACUUM.2011.08.008. [12] T. G zebyk, A. Gó ecka-D zazga, MEMS ype ioniza ion acuum senso , Senso s Ac ua o s A Phys. 246