Non-Ohmic Phenomena in Mn-doped BaTiO3

Tí ulo a ículo / Tí ol a icle: Non-ohmic phenomena in Mn-doped BaTiO3 Au o es / Au o s Ma a P ades Héc o Bel án Eloisa Co doncillo Pablo J. Alonso Nahum Masó An hony R. Wes Re is a: Phys. S a us Solidi A 209, No. 11, 2267–2272 (2012 Ve sión / Ve sió: P e-p in Ci a bibliog á ica / Ci a bibliog à ica (ISO 690): PRADES, Ma a, BELTRÁN, Héc o , CORDONCILLO, Eloisa, ALONSO, Pablo J., MASÓ, Nahum, WEST, An hony R. Non-ohmic phenomena in Mn-doped BaTiO3. Pha macology Biochemis y and Beha io , No embe 2012, Issue 11. u l Reposi o i UJI: h p://hdl.handle.ne /10234/67863 1 Non-Ohmic Phenomena in Mn-doped BaTiO3 Ma a P ades1, Héc o Bel án1, Eloisa Co doncillo1, Pablo J. Alonso2, Nahum Masó3 and An hony R. Wes *,3 1 Depa amen o de Química Ino gánica y O gánica, Uni e si a Jaume I, A da. Sos Bayna s/n, 12071 Cas ellón, Spain. 2 Ins i u o de Ciencia de Ma e iales de A agón, ICMA, Uni e sidad de Za agoza-C.S.I.C., C/ Ped o Ce buna 12, E-50009 Za agoza, Spain. 3 Depa men o Ma e ials Science and Enginee ing, Uni e si y o She ield, Mappin S ee , S1 3JD She ield, Uni ed Kingdom. * E-mail: a. .wes @she ield.ac.uk Keywo ds: Non-Ohmic Phenomena, Ba ium Ti ana e, Ce amics, Elec ical Conduc i i y. We epo he e a no el e ec in which he esis ance o a semiconduc ing oxide ce amic inc eases on applica ion o a small dc bias. The ce amic conduc s a high empe a u es by an n- ype hopping mechanism. On applica ion o a dc bias, conduc ion elec ons a e apped a su ace s a es and he esis ance inc eases. On emo al o he dc bias, he apped elec ons a e eleased and he sample egains i s o iginal s a e. This e ec is he mi o image o ha seen wi h simila ce amics ha conduc by a p- ype mechanism whose esis ance dec eases e e sibly on applica ion o a small dc bias. These wo phenomena oge he o e he possibili y o no el swi ching de ices and mem is i e applica ions, especially i he swi ching imes can be educed. In he absence o in e acial e ec s, non-me allic ma e ials such as oxide ce amics show linea ol age/cu en , V/I, beha iou a small applied ol ages and hei esis ance, R, is he e o e cons an , as gi en by Ohm’s law. This is because pa ame e s ha con ol conduc i i y, ie he numbe o mobile ca ie s and hei mobili y o hopping a e, a e unin luenced by small applied ol ages. 2 Non-linea low- ield beha iou has been obse ed ecen ly[1–3] in p- ype, accep o -doped BaTiO3 ce amics in which he esis ance dec eased by 1–2 o de s o magni ude on applica ion o a small bias ol age, ypically in he ange 1–10V, ac oss pelle s o hickness 1–2mm; he esis ance dec ease was ully e e sible on emo al o he dc bias and was no associa ed wi h in e acial e ec s such as Scho ky ba ie s. The a e o change o esis ance wi h ime was e y empe a u e-dependen . I ook se e al hou s o achie e a bias-dependen s eady s a e a 200–300 °C whe eas he changes we e comple e in a ew minu es a 600–800 °C. The enhanced conduc i i y was a ibu ed o he p esence o unde bonded oxygen a oms su ounding accep o dopan s in he BaTiO3 c ys al la ice; acile ionisa ion o hese unde bonded oxygens gene a ed holes, o O– ions, and he inc ease in hole concen a ion was esponsible o he enhanced conduc i i y. The d i ing o ce o ionisa ion was p o ided by he dc bias which caused he ac i a ion o elec on ap s a es a he sample su ace[4]. The e ec epo ed he e is he mi o image o he beha iou o accep o -doped sys ems and is obse ed in Mn-doped BaTiO3 which, when ligh ly educed, is an n- ype semiconduc o . The e ec o a small dc bias is also o ap some o he conduc ion elec ons which in his case, leads di ec ly o a esis ance inc ease. On emo al o he dc bias, he apped elec ons a e eleased and he conduc i i y egains i s o iginal alue. A key o unde s anding his, and he ea lie phenomenon wi h accep o -doped ma e ials, is he obse a ion ha he conduc i i y o bo h p- ype and n- ype doped BaTiO3 can be modi ied in a simila way by wo independen me hods, ei he by changing he oxygen pa ial p essu e, 2 O P , in he a mosphe e su ounding he sample du ing conduc i i y measu emen s[5,6] o by applica ion o a dc bias in an a mosphe e o cons an 2 O P .[1–4] When O2 molecules abso b on a ce amic su ace, he molecules ionise by apping elec ons; dissocia ion o he molecules may also occu . Consequen ly, elec ons a e deple ed om a egion o he ce amic close o 3 aps a he sample su ace. I he conduc i i y inc eases, he sample is p- ype and ice e sa, i is n- ype i he conduc i i y dec eases. I hus appea s ha applica ion o a dc bias p o ides a second mechanism o inc ease he apping o conduc ion elec ons a su ace s a es. The sample o Mn-doped BaTiO3 used he e was an n- ype conduc o . I s esis ance inc eased as conduc ion elec ons we e wi hd awn om he sample, ei he by applying a dc bias o by inc easing 2 O P in he a mosphe e su ounding he sample. These esul s a e o possible signi icance o se e al easons. Fi s , hey demons a e ha he pa ially ionised oxygens which mus exis in many s anda d ce amic samples, especially a sample su aces[7], can eadily ac as ei he elec on aps o elec on sou ces. Second, i he sample is an n- ype conduc o , he apping o elec ons unde he ac ion o a small dc bias leads o an inc ease in sample esis ance. Thi d, i he ma e ial is a p- ype semiconduc o hen, unde ce ain ci cums ances, as wi h accep o -doped BaTiO3, ionisa ion o unde bonded oxide ions associa ed wi h he accep o dopan s may occu , p o iding elec ons ha a e subsequen ly apped a he su ace s a es; his hen leads o an inc ease in hole concen a ion and he e o e a dec ease in sample esis ance. Fou h, no el non-linea phenomena a e obse ed in which, depending on he pa icula sys em and conduc i i y mechanism, he sample esis ance may ei he inc ease o dec ease on applica ion o a small dc bias ol age. Fi h, gi en he sensi i i y o sample esis ance o ol age, no el senso and swi ching applica ions may be possible, especially i he swi ching speed can be inc eased, and a lowe empe a u es. The samples used we e sin e ed pelle s o BaTiO3 in which 0.5% o Ti was eplaced by Mn o gi e he nominal composi ion BaTi0.995Mn0.005O3-δ. Samples we e p epa ed by sol-gel 4 syn hesis o alkoxide p ecu so s using me hods desc ibed p e iously.[8] Resul ing powde s we e pelle ed and i ed in s eps wi h inal hea ing a 1400 °C o 12 h in ai a e which samples we e slowly cooled o oom empe a u e in he u nace. Fo elec ical p ope y measu emen s, elec odes we e ab ica ed om o gano P pas e which was applied o opposi e pelle aces, d ied and ha dened by hea ing o 900 °C. Samples wi h elec odes a ached we e e u ned o he u nace a 1400 °C o 2 h in ai and hen quenched in liquid N2. Pelle densi ies we e ~90 %. Samples we e placed in o he conduc i i y jig and elec ical p ope y da a eco ded using an Agilen 4294A impedance analyse o e he equency ange, 40 Hz o 13 MHz and o e he empe a u e ange, oom empe a u e o 900ºC. Impedance da a we e co ec ed o he o e all pelle geome y and o he blank capaci ance o he jig. Resis ance and capaci ance da a a e, he e o e, epo ed in uni s o cm and Fcm–1, espec i ely. The oxida ion s a e o Mn was s udied by Elec on Pa amagne ic Resonance (EPR) a oom empe a u e using an Elexys E580 spec ome e om B uke , wo king in he X- band. The magne ic ield was measu ed wi h a B uke ER035M gaussme e . Powde ed samples we e packed in used qua z ubes (707-SQ om Wilmad-LabGlass). Impedance da a a e shown a one empe a u e, 488 ºC, o a sample quenched om 1400 °C, in Fig 1. A single a c was obse ed in he Z* complex plane (a) and single peaks in he Z''/M'' spec oscopic plo s (b) which showed ha he sample bulk domina ed he impedance da a wi hou signi ican g ain bounda y o elec ode con ac e ec s. The pe mi i i y as a unc ion o empe a u e (c) showed a maximum a he Cu ie empe a u e and Cu ie-Weiss beha iou a empe a u es abo e Tc (d). Bulk conduc i i y da a p esen ed in A henius o ma in (e), showed ha he quenched sample had highe conduc i i y and lowe ac i a ion ene gy han he same sample which had been slow-cooled om 1400 °C. The conduc i i y inc ease was 5 a ibu ed o a small amoun o in insic oxygen loss om he sample a 1400 °C which was no egained du ing apid quench, unlike eoxida ion ha occu ed du ing slow cool. The sample quenched om 1400ºC was e ec i ely in a ozen-in s a e and suscep ible o eoxida ion in ai , a a a e ha was dependen on empe a u e, leading o a dec ease in conduc i i y (no shown) as a consequence o wi hd awal o elec ons om he sample acco ding o he eac ion: x OO2 O2e4V2O     . This indica ed ha elec ons we e he main cha ge ca ie s and, he e o e, he conduc ion mechanism was p edominan ly n- ype. Consequen ly, o a oid sample oxida ion du ing impedance measu emen s, da a on he quenched sample we e eco ded in an a mosphe e o N2. Impedance da a o he quenched sample a 488ºC in lowing N2 a e shown in Fig 1( ). O e a pe iod o 24h, he esis ance showed a g adual inc ease o ~9% a ibu ed o p esence o esidual O2 in he N2 a mosphe e. This g adual oxida ion was on a much longe imescale han he o he changes epo ed he e. The e ec o a 10 V dc bias on he o al conduc i i y o he quenched sample is shown in Fig 2(a) as unc ion o ime. I dec eased apidly a i s , ollowed by a much mo e g adual dec ease o e longe imes. On emo al o he dc bias, he conduc i i y inc eased apidly, al hough did no ully a ain i s o iginal s a e, indica ing ha he changes induced by he dc bias we e essen ially e e sible. The di e ence be ween ini ial and inal conduc i i ies in Fig 2(a) (ie be o e and a e applica ion/ emo al o he dc bias) is a ibu ed o a pa ial eoxida ion which was supe posed on he e e sible dec ease in conduc i i y on applica ion o he dc bias. Simila esul s o hose shown in Fig 2(a) we e ob ained a o he empe a u es abo e ~200 °C bu he a e o change was e y empe a u e-dependen ; o ins ance, he esis ance con inued o inc ease o e a pe iod o se e al days a 400 °C. The esis ance changes shown in Fig 2(a) o n- ype BaTiO3, a e he mi o image o hose seen wi h accep o -doped, p- ype BaTiO3, as shown o one example in Fig 2(b). These di e ences in 6 beha iou we e independen o ce amic mic os uc u e: as shown in Fig 3, g ain sizes we e compa able, ~5–30 μm o (a) and ~20–60 μm o (b). Fu he con i ma ion ha he conduc ion mechanism in he quenched sample was n- ype was ob ained by conduc i i y measu emen s made in di e en a mosphe es in a sample ha was pa ially oxidised, Fig 4a. The conduc i i y inc eased e e sibly wi h dec easing 2 O P in he measu ing a mosphe e and, he e o e, elec ons we e he p incipal cha ge ca ie s. EPR spec a o Mn cen es (g ≈ 2) o e he magne ic ield ange 300–400 mT a e shown o slow-cooled and quenched samples in Fig 4b and c, espec i ely. In he slow-cooled sample, a b oad esonance and a single a g = 1.97 we e obse ed, Fig 4b. No signals associa ed wi h Mn cen es could be iden i ied al hough he b oad esonance could be ela ed o agg ega ion o Mn cen es. The signal a g = 1.97 was a ibu ed o an una oidable ace (ppm’s) o C 3+ (Re . 9) in he s a ing ma e ials. By con as , in he spec um o he quenched sample, wo sex e s (ma ked as * and + in Fig 4c) and a single a g = 1.97 ela ed o C 3+ we e iden i ied. Bo h sex e s we e associa ed wi h 55Mn cen es (nuclea spin I = 5/2 and 100 % na u al abundance) wi h g- ac o s and hype ine coupling cons an s (A): g = 2.05(1) and A = 200(6) MHz (signals labelled as *) and g = 2.00(1) and A = 240(6) MHz (signals labelled as +). The Mn-sex e s we e assigned o *:Mn2+ and +:Mn4+ (Re s 10, 11) al hough al e na i ely, bo h sex e s may be associa ed wi h Mn2+, as epo ed in S TiO3 (Re . 12). In addi ion, samples could also con ain Mn3+ since his canno usually be de ec ed by EPR.[10,11] The e o e, he oxida ion s a e o Mn in BaTiO3 could be some combina ion o 2+, 3+ and 4+, depending on p ocessing condi ions. In he quenched sample, bo h Mn2+ and Mn4+ a e p esen . I he Mn4+ was indeed p esen a 1400ºC, and did 7 no o m du ing eoxida ion, hen Mn3+ would also p obably be p esen , gi ing Mn2+,3+,4+ mix u es. Ce amic samples, such as pu e and doped BaTiO3, unde equilib ium condi ions a high empe a u e, gene ally show a p- ype o n- ype change in beha iou on educing oxygen pa ial p essu e [13,14]. In he p- ype egion, holes a e gene a ed by he idealised eac ion:   h2OVO21 x OO2 (1) The loca ion o he holes in ‘pu e’ BaTiO3 is o en no clea bu is equen ly a ibu ed o unspeci ied and una oidable accep o impu i ies such as Fe3+. In he n- ype egion, he e e se eac ion occu s, e2VO21O O x O   (2) and he libe a ed elec ons a e p esumed o be associa ed wi h ca ions, ie Ti o eg Mn. In Mn-doped BaTiO3, se e al s udies on he 2 O P dependence o conduc i i y ha e been epo ed [13,15,16]. These show a p- ype o n- ype c osso e wi h dec easing 2 O P and a displacemen o he c osso e egion o highe 2 O P wi h inc easing empe a u e. Nei he s udies epo ed da a a 1400ºC bu , by app oxima e ex apola ion, n- ype beha iou could be expec ed a 1400ºC. This is consis en wi h ou expe imen al conduc i i y da a in di e en a mosphe es which show n- ype beha iou and also by EPR measu emen s, Fig 4(b), which show he p esence o Mn2+ species in he quenched samples. The e ec o a dc bias, leading o a dec ease in conduc i i y, Fig 2, was simila o ha o exposu e o he quenched sample o an oxygen- ich a mosphe e, Fig 4a. In bo h cases, n- ype ca ie s we e emo ed om he sample bulk leading o an inc ease in esis ance. Va ia ions in conduc i i y o his na u e as a consequence o changing a mosphe e a e e y well 8 es ablished and, i hey occu , a e used as a s ong indica o as o whe he he ca ie s a e p incipally n- ype o p- ype. Howe e , we belie e ha his is he i s ime ha a simila e ec has been ecognised on applica ion o a small dc bias leading o deple ion o n- ype ca ie s om he sample bulk. The ap s a es a e no necessa ily he same o he wo cases o oxygen abso p ion and applica ion o a dc bias: se e al possibili ies a e indica ed in he ollowing sequence o successi e s eps, h ee o which, s eps 2, 3 and 5, in ol e a educ ion p ocess: (3) In he oxygen abso p ion case, he aps may be eshly adso bed oxygen molecules om he su ounding a mosphe e which pick up elec ons om he sample o o m supe oxide ions, O2–, s ep 2. The i s s ep in ol es adso p ion o oxygen molecules on he ce amic su ace bu his would no be de ec ed by impedance measu emen s since no educ ion o he O2 molecules is in ol ed. These adso bed O2 molecules may ac as elec on aps, s ep 2, by picking up elec ons om he sample o o m supe oxide ions, O2–. The supe oxide ions may hen pick up a second elec on o o m pe oxide ions, O22–, s ep 3, which may o may no be s able o dissocia ion, s ep 4. The O– ions o med on dissocia ion may ha e some s abili y in an unde bonded en i onmen a sample su aces o may pick up an ex a elec on, s ep 5, o o m O2– ions, which a e hen able o di use in o he sample in e io . A p esen , we canno commen on he ela i e likelihood o hese possibili ies du ing changes o 2 O P in he a mosphe e and on applica ion o a dc bias bu no e ha , in he case o accep o -doped ma e ials, dc bias e ec s we e obse ed e en when he sample was e acua ed. This he e o e indica es ha p e-exisi ing aps a e p esen , spon aneously, a any ime, on he su ace o such oxide ce amics. O2(ads)O2 (g) s ep 1 O2 s ep 2 e-O2 s ep 3 e-22O s ep 4 2e- O2(ads)O2 (g) s ep 1 O2 s ep 2 e-O2 s ep 3 e-22O s ep 4 2e- 2 O2(ads)O2 (g) s ep 1 O2 s ep 2 e-O2 s ep 3 e-22O s ep 4 2e- O2(ads)O2 (g) s ep 1 O2 s ep 2 e-O2 s ep 3 e-22O s ep 4 2e- 2 5 O2(ads)O2 (g) s ep 1 O2 s ep 2 e-O2 s ep 3 e-22O s ep 4 2e- O2(ads)O2 (g) s ep 1 O2 s ep 2 e-O2 s ep 3 e-22O s ep 4 2e- O2(ads)O2 (g) s ep 1 O2 s ep 2 e-O2 s ep 3 e-22O s ep 4 2e- O2(ads)O2 (g) s ep 1 O2 s ep 2 e-O2 s ep 3 e-22O s ep 4 2e- 15 Fig 3. SEM o he pelle su ace o (a) BaTi0.995Mn0.005O3- and (b) BaTi0.99Ca0.01O2.99. (a) (b) 16 Fig 4. (a) Impedance complex plane plo , Z* a 490ºC, o he pa ial eoxidised sample (s eady s a e), measu ed in di e en a mosphe es; EPR spec a o he slow cooled (b) and quenched sample (c) a oom empe a u e. (a) (b) (c)