The riddle of orange-red luminescence in Bismuth-doped silica glasses

1 Vol.:(0123456789) Scien i ic Repo s | (2021) 11:7766 | h ps://doi.o g/10.1038/s41598-021-87290-z www.na u e.com/scien i ic epo s The iddle o o ange– ed luminescence in Bismu h‑doped silica glasses Oleksii V. Lagu a1* & Igo M. Razdob ee 2 Fo o e he pas wo decades i has been belie ed ha he in ense o ange‑ ed pho oluminescence in Bismu h‑doped ma e ials o igina es om Bi 2+ ions. Based on he esul s om magne ic ci cula pola iza ion expe imen s, we demons a e ha his hypo hesis ails o Bismu h‑doped silica glasses. Ou indings con adic he gene ally accep ed s a emen ha he o ange‑ ed luminescence a ises om 2 P3 / 2(1 ) → 2 P1 /2 ansi ion in a di alen Bismu h ion. The deg ee o magne ic ci cula pola iza ion o his luminescence exhibi s non‑mono onic empe a u e and ield dependencies, as well as sign e e sal. This complex beha iou canno be explained unde he assump ion o a single Bi 2+ ion. The de ailed analysis enables us o cons uc a consis en diag am o ene gy le els in ol ed in he magne o‑op ical expe imen s and p opose a new in e p e a ion o he na u e o o ange‑ ed luminescence in Bismu h‑doped silica glass. A cen e esponsible o his no o ious pho oluminescence mus be an e en‑elec on sys em wi h an in ege o al spin, p esumably a dime o Bismu h ions o a complex consis ing o Bi 2+ and an oxygen acancy. The o ange- ed pho oluminescence (ORPL) in Bismu h-doped ma e ials has been no ed since he wo k o Lecoq de Boisbaud an1, who obse ed his luminescence in MSO 4 :Bi (M = Zn, Cd and Pb) sul a es which we e subjec ed o he elec ical discha ge. Ne e heless, only a he end o he 20 h cen u y did Blasse e al.2,3 a ibu e he o igin o his unusual ORPL in Bismu h-doped s on ium e abo a e S B 4 O 7 :Bi and some alkaline-ea h-me al sul a es o a 2P3/2 → 2P1/2 ansi ion in a di alen Bismu h ion. In he ollowing yea s, a wide ange o Bi-doped ma e ials exhibi ing he ORPL was syn hesized – ba ium bo a es4, phospha es5–7, luo ides8, ba i e9, g ossi e10. The main eason o such inc eased in e es in hese ma e ials is he po en ial use o he ORPL o imp o e he e iciency and colou empe a u e o whi e ligh -emi ing diodes11–14. No ably, he au ho s o he abo e-men ioned wo ks assigned he o ange- ed pho oluminescence o a Bi 2+ ion based exclusi ely on he esemblance o he abso p ion/ emission spec a o hose obse ed by Blasse e al.2. The elec onic con igu a ion o a Bi 2+ ion, 6s2 6 p1 , in p inciple, sugges s ha an EPR signal should be obse ed. Howe e , e en a liquid helium empe a u e, he EPR signal could no be obse ed in any o he s udied com- pounds. De Jong e al.15 ha e ecen ly e-examined he ORPL in S B 4 O 7 :Bi in o de o asce ain he eason why he EPR signal is no obse ed. Acco ding o he au ho s, he e y low uppe limi o he Bi 2+ concen a ion, which was es ima ed o be 20 ppm, is he main eason ha he EPR signal in his ma e ial canno be de ec ed. I is wo h no ing, howe e , ha in hei analysis, i was a p io i pos ula ed ha Bi 2+ is esponsible o he ORPL. In his ega d, i should be no ed ha a leas wo Bi-doped ungs a es, namely, CdWO 4 16 and PbWO 4 17, a e known in which Bi 2+ ela ed EPR was in es iga ed. The EPR signal in hese single c ys als appea s only a e i adia ion (X- ays, me cu y o xenon lamps) a low empe a u es (77 – 100 K) and i comple ely disappea s a oom empe a u e. Un o una ely, bo h compounds exhibi in insic ORPL, which is obse ed in hese ma e i- als unde exci a ion in he ange o 340 – 370 nm ( o PbWO 4 see, o ins ance e .18), while he appea ance o new addi ional pho oluminescence (PL) o abso p ion bands di ec ly connec ed o Bi 2+ ions was no epo ed and emains ques ionable. Ano he class o ma e ials ha exhibi ORPL ela ed o Bismu h doping a e silica-based glasses. The in e es in hese ma e ials a ose due o he disco e y o Bismu h- ela ed b oadband nea in a ed pho oluminescence (NIR PL)19,20. Since silica-based glasses a e compa ible wi h mode n ib e op ic sys ems, emendous p og ess has been made o e he pas 20 yea s in he de elopmen o bismu h-based ib e lase s and ampli ie s21–31. Ne e he- less, despi e his echnological p og ess, he na u e o NIR PL and lasing cen es con inues o be discussed32–34. OPEN 1Cen al Eu opean Ins i u e o Technology, CEITEC BUT, Pu kyňo a 656/123, 61200 B no, Czech Republic. 2UMR 8523 - PHLAM - Physique des Lase s A omes e Molécules, CERLA, Uni . Lille, CNRS, 59000 Lille, F ance. *email: oleksii.lagu a@cei ec. u b .cz 2 Vol:.(1234567890) Scien i ic Repo s | (2021) 11:7766 | h ps://doi.o g/10.1038/s41598-021-87290-z www.na u e.com/scien i ic epo s/ The in ense ORPL obse ed unde UV o blue-g een exci a ion is also one o he in iguing ea u es o Bismu h-doped ib es. This PL was demons a ed o he i s ime in bulk sin e ed ib e p e o ms ab ica ed om nano-po ous xe ogels35. La e , his luminescence was a ibu ed o a single Bi 2+ ion by analogy wi h he wo k o Blasse e al.2 and a p esen , o he bes o ou knowledge, his is a gene ally accep ed in e p e a ion28,36–38. The ORPL in Bi-doped silica glasses is obse ed a e high- empe a u e sin e ing o he po ous hos ma ix o a e he mel ing o aw compounds, and i s eco ding does no equi e any i adia ion. A he same ime, all ou a emp s o de ec he EPR signal a e i adia ion a low empe a u es (down o 5 K) had ailed. Fo his eason, he iden i ica ion o ORPL in Bismu h-doped silica glass as a ansi ion 2P3/2 → 2P1/2 in a di alen Bismu h ion was doub ul. In ou ecen pape 39, we in es iga ed he magne ic ci cula dich oism (MCD) in bulk ib e p e o ms SiO 2 :Bi, in which bo h cen es, ORPL and NIR PL (lasing cen e), coexis . I was shown ha he eco ded MCD bands, mos p obably, should be assigned o he ORPL ela ed cen e. Fu he mo e, he MCD beha iou as a unc ion o magne ic ield and empe a u e can be explained only by assuming he e en-elec on na u e o a single emi ing cen e. Thus, he la e canno be iden i ied as a Bi 2+ ion. Un o una ely, e en a e y low doping le els Bismu h- doped glasses a e always cha ac e ized by he mul iplici y o op ical cen es and he unambiguous co espondence be ween he abso p ion and emission bands is no ob ious. Fo his eason, in he p esen pape , we epo on he measu emen s o magne ic ci cula pola iza ion o luminescence (MCPL) in bulk Bismu h-doped silica glass. MCPL di ec ly cha ac e izes he magne ic p ope ies o he exci ed s a es and o e s such ad an ages o e he s anda d EPR as high sensi i i y and selec i i y. In con as o ou ea lie s udy40, in which we in es iga ed he NIR PL o he lasing cen e, he e we analyse he ORPL and accompanying PL in he egion o 850 nm. The esul s o he p esen wo k cons i u e di ec and unequi ocal e idence ha he ORPL in Bi-doped silica glass canno be iden i ied as a 2P3/2 → 2P1/2 ansi ion in a Bi 2+ ion. Also, based on he esul s ob ained om MCPL and MCD expe imen s, we ha e cons uc ed a consis en ene gy le el diag am o he ORPL cen e and discuss i s na u e. Resul s Pho oluminescence and exci a ion spec a. In Fig.1 we show he PL spec a eco ded in he ange o 460 – 900 nm unde exci a ion a 450 and 532 nm. The spec a we e eco ded a wo empe a u es: 1.47 and 13 K. Al hough all he spec a shown in Fig.1 we e eco ded in a magne ic ield o 6T, we did no e eal any ield dependence o he PL spec um. I is seen ha he PL spec um a bo h exci a ion WLs consis s o he in ense ORPL band accompanied by a long-wa eleng h ail. The s ong exci a ion wa eleng h dependence o he ORPL is due o i s non-homogeneous b oadening. The maximum in ensi y o he ORPL band is a 585 and 650 nm, while i s ull wid h a hal maximum is app oxima ely 3000 and 2600 cm −1 unde exci a ion a 450 and 532 nm, espec i ely. The change o empe a u e only sligh ly a ec s he shape and posi ion o he ORPL, being wi hin he expe imen al accu acy. I is e y impo an o no e ha unde exci a ion a 532 nm a ela i ely weak and b oad PL band NIR1 wi h he maximum a ound 850 nm accompanies he ORPL. As i will be shown below, his NIR1 and he ORPL bands belong o he same luminescen cen e. Like he ORPL band, he NIR1 band also exhibi s s ong non-homogeneous b oadening. The addi ional ela i ely na ow band NIR2 a 830 nm, which is obse ed unde exci a ion a 450 nm, belongs o he lasing cen e and should be assigned o he 2−→ 1+ ansi ion in a Bi + ion, which has been discussed by us p e iously34. Such an assignmen is also suppo ed by he measu emen s o exci a ion spec a shown in Fig.2. I is seen in Fig.2a ha he exci a ion spec a eco ded a 580 and 830 nm a e e y di e en . The exci a ion spec um o he ORPL band eco ded a 580 nm shows wo well-sepa a ed peaks (P1 and P4). The spec um eco ded a 5 4 3 2 1 0 In ensi y (cps, x10 4 ) 900 800700600500 Wa eleng h (nm) T = 1.47 K λ exc = 450 nm λ exc = 532 nm T = 13 K λ exc = 450 nm λ exc = 532 nm ORPL NIR1 NIR2 (Bi + ) Figu e1. Pho oluminescence spec a o SiO 2 :Bi a di e en exci a ion wa eleng hs and empe a u es in he magne ic ield 6 T. 3 Vol.:(0123456789) Scien i ic Repo s | (2021) 11:7766 | h ps://doi.o g/10.1038/s41598-021-87290-z www.na u e.com/scien i ic epo s/ 830 nm consis s o wo s ong peaks (labelled P2 and P3) which a e caused by he ansi ions 1+−→ 3+ (P2) and 1+−→ 3 (P3) in Bi + 34. The sho - and long-wa eleng h ails, hough, seem o coincide wi h P1 and P4 peaks o he ORPL spec um, and a e due o he o e lapping o NIR2 and NIR1 luminescence. Finally, he exci a ion spec um eco ded a 855 nm shown in Fig.2b can be conside ed as a weigh ed sum o he wo p e ious spec a. I s shape is again due o he o e lapping o he luminescence spec a o wo di e en cen es: he Bi + ion and he cen e unde he ques ion esponsible o he ORPL. Howe e , a 855 nm he ela i e PL in ensi ies o hese cen es a e compa able because in Bi + a his pa icula wa eleng h he ac i e ansi ion is 2+−→ 1+ , which is much weake han he 2−→ 1+ ansi ion a 830 nm (see Fig.1 and Table1 in Re .34). The p esence o he addi ional spec al band NIR1, which belongs o he ORPL cen e, becomes e en mo e ob ious when s udying he MCPL. Magne ic ci cula pola iza ion o luminescence. In Fig.3a,b we show he MCPL spec a ( I+−I− ) along wi h spec a o he o al in ensi y ( (I++I−)/2 ) eco ded in a magne ic ield o 6 T a wo empe a u es: 1.47 and 13 K. The mos no iceable ea u e is he di e en sign o ci cula pola iza ion in he ORPL band a hese empe a u es; namely, i is nega i e a he lowe empe a u e and posi i e a he highe . Addi ionally, a he lowe empe a u e he peak o MCPL is blue-shi ed ela i e o he PL band o bo h exci a ion WLs. This shi becomes impe cep ible a T = 13 K. I is also seen ha he MCPL in he long-wa eleng h ail o ORPL is nega i e a bo h empe a u es and i does no co ela e wi h he na ow PL band obse ed a 830 nm. I ollows ha his b oad PL should be a ibu ed o a sepa a e band, which is no ela ed o he luminescence o Bi + ions, and, on he con a y, belongs o he ORPL cen e. In he s udy o MCPL, he mos use ul quan i y is he so-called emission aniso opy ac o o he deg ee o ci cula pola iza ion (DCP), which is de ined as DCP =(I+−I−)/(I++I−) 41,42. Simila ly o MCD, his quan- i y can be p esen ed as a sum o h ee e ms: A , B and C . Fo he simples model o an isola ed Gaussian band, only he A - e m exhibi s a spec al dependence: A∼ (ν −ν 0 ) . Then, he obse ed DCP spec um, as a ule, akes he o m o an inclined ( A = 0 ) o a ho izon al ( A=0 ) s aigh line41. A e y use ul p ope y o he A - e m is ha i akes ze o alue a he maximum in ensi y o he PL band43. In he as majo i y o cases, A - and B - e ms a e empe a u e independen ; he e o e, measu emen s o DCP a he maximum PL band in ensi y allow he di ec s udy o he pa amagne ic e m C . The expe imen al spec a o DCP eco ded in he magne ic ield o 6 T a he exci a ion WLs 450 and 532 nm, a e shown in Fig.3c,d, espec i ely. I is seen ha a he low empe a- u e he DCP spec um o he main ORPL band can be well app oxima ed by a single inclined s aigh line. The pa ame e s ha co espond o he pa amagne ic C and diamagne ic A e ms ob ained om he i a e as ollows: a) C=− 5.99 ( 1 )× 10 −2 and A=− 8.74 ( 1 )× 10 −6 , and b) C=− 7.52 ( 2 )× 10 −2 and A=− 1.14 ( 1 )× 10 −5 o he abo e exci a ion WLs, espec i ely. A a empe a u e o 13 K, he pa amagne ic e ms become posi i e and a e e y close o bo h exci a ion WLs: C= 3.5 ( 1 )× 10 −2 , while he A - e m becomes negligible. This s ange beha iou o he A - e m (by he de ini ion i is empe a u e insensi i e) can appea only i he MCPL o igina es om wo o mo e magne ic mul iple s, and i is well explained in ou model p esen ed in he nex sec ion. The e is ano he impo an ea u e o he obse ed DCP ha mus be poin ed ou . I is seen ha he long-wa eleng h ail o luminescence always exhibi s nega i e DCP. The sha p peak, which can be seen in Fig.3c a 12050 cm −1 (830 nm), can be also obse ed unde exci a ion a 375 nm, whe e PL, and consequen ly posi i e DCP, om Figu e2. Exci a ion spec a. (a) PL exci a ion spec a o he ORPL and NIR2 bands. (b) PL exci a ion spec um o he NIR1 band eco ded a 855 nm and i s mul i-peak decon olu ion. All he spec a we e eco ded a 1.47 K and ze o magne ic ield. 4 Vol:.(1234567890) Scien i ic Repo s | (2021) 11:7766 | h ps://doi.o g/10.1038/s41598-021-87290-z www.na u e.com/scien i ic epo s/ Bi + ions become dominan 40. This peak is absen in he DCP spec um eco ded a exc = 532 nm (panel (d) o Fig.3), while he nega i e DCP in his spec al egion p ac ically ep oduces ha in panel (c) o Fig.3. Thus, he measu emen s o MCPL con i m ha he PL NIR2 is due o Bi + ions and i appea s due o he pa ial o e lap o i s exci a ion spec um and ha o ORPL, as i was poin ed ou p e iously in his wo k. Fu he de ailed in es iga ions o he MCPL in he ORPL band a a a iable empe a u e and magne ic ield (VTVH) we e pe o med a 650 and 585 nm, co esponding o he maximums o ORPL a he lowes empe a u e unde exci a ion a 532 and 450 nm, espec i ely. The expe imen al da a on VTVH-MCPL ob ained a 585 nm VTVH-MCPL a e shown in Fig.4a,b. In he le (a) panel he da a a e p esen ed so as o show he empe a u e dependence a a ixed magne ic ield. On he con a y, in he igh panel Fig.4b, he da a is shown as a unc ion o he magne ic ield a a ixed empe a u e. I is seen ha a any ixed alue o he magne ic ield, DCP beha iou is non-mono onic. The deg ee o ci cula pola iza ion inc eases wi h he empe a u e dec ease and eaches i s maximum alue DCP =+0.047 a T ≈ 10 K in he magne ic ield o 6 T. Then, he DCP dec eases, changes i s sign in he icini y o T=6 K, and sa u a es. Ano he in e es ing de ail is ha he MCPL sa u a es o di e en alues o each magne ic ield magni ude. This is a dis inguishing ea u e o an e en-elec on sys em40,44. Sa u a ion cu es eco ded a 650 nm unde exci a ion a 532 nm exhibi a a he simila empe a u e beha iou , and can be ound in Fig.S1 o he Supplemen a y Ma e ials. We also pe o med VTVH measu emen s o DCP a he wa eleng h o 850 nm upon exci a ion a 532 nm o make only he ORPL cen e ac i e. As i is shown in Fig.4c, he DCP exhibi s a simple mono onic empe a u e dependence a any magne ic ield. The pola iza ion deg ee inc eases wi h empe a u e and sa u a es below 5 K. Simila ly o he ORPL band, he alue o he sa u a ion pa ame e inc eases wi h he inc easing magne ic ield. In Fig.4d we show he magne ic ield dependences a ixed empe a u es. I is seen ha he magne iza ion cu es exhibi he nes ed (o an-ou ) beha iou cha ac e is ic o a ze o- ield spli pa amagne ic ini ial s a e. Thus, bo h ep esen a ions o DCP beha iou shown in Fig.4c,d di ec ly indica e he e en-elec on sys em na u e o he ORPL cen e, whe eas Bi 2+ has only one unpai ed elec on. When analysing MCPL measu ed in he expe imen s wi h con inuous wa e (CW) exci a ion, he spin elaxa- ion p ocess mus be aken in o accoun . I is c ucial ha he popula ion dis ibu ion eaches i s he modynamic Figu e3. Compa ison o MCPL ( I+−I− ) and o al in ensi y spec a eco ded upon exci a ion a 450 and 532 nm ( (a) s (b) and (c) s (d)). Compa ison o DCP ( I + −I −)/( I + +I −) spec a eco ded a empe a u es o 1.47 and 13 K. Magne ic ield is 6 T in all panels. Ve ical dash-do lines indica e he band maximum posi ion. 5 Vol.:(0123456789) Scien i ic Repo s | (2021) 11:7766 | h ps://doi.o g/10.1038/s41598-021-87290-z www.na u e.com/scien i ic epo s/ equilib ium be o e he elec ons elax om he exci ed s a e. This means ha he spin-la ice elaxa ion ime mus be much sho e han he adia i e decay ime. I his condi ion is no sa is ied, he ob ained alues o g- ac o and ZFS will be inco ec . To ensu e ha in ou expe imen s he abo e-men ioned condi ion is sa is ied, we pe o med he measu emen s o ime- esol ed DCP45. In his echnique, one eco ds sepa a ely he ime- esol ed decays o σ+ and σ − componen s o pho oluminescence. Then he subsequen sub ac ion and no maliza ion allow he ob ainmen o he ime e olu ion o DCP. Howe e , expe imen s wi h nanosecond ime esolu ion did no allow us o obse e he ini ial build-up o DCP a e pulsed exci a ion, as i is shown in Fig.S2. In all ou expe imen s DCP eached i s equilib ium alue ins an ly. This indica es a e y as spin elaxa ion, mos p obably a he picosecond scale, so ha i does no in luence he esul s o CW DCP measu emen s. MCPL analysis and ene gy le el diag am. The unusual beha iou o DCP desc ibed abo e could be explained wi h he assump ion ha he e a e wo independen e en-elec on cen es, each o which con ib- u es o luminescence. Howe e , ou ea lie esul s on magne ic ci cula dich oism con adic his assump ion39. Ano he possibili y is o assume ha he e is only one kind o luminescen cen e, in which case he ORPL is caused by he simul aneous ansi ions om a leas wo closely spaced exci ed s a es (o sub-s a es). The MCPL due o hese exci ed s a es is opposi e in sign and, in addi ion, he absolu e alues o hei ze o- ield spli ing (ZFS) mus be signi ican ly di e en . In Fig.5 we show he ene gy le el diag am o he ORPL cen e, which sa is ies all he obse ed MCPL ea u es. Figu e5a gi es a gene al ou line o le els. Due o he s ong inhomogeneous b oadening, bo h s a es |2� o |3� can be exci ed simul aneously, bu in he cen es om di e en sub-se s. A e he exci a ion, which is ollowed by as non- adia i e elaxa ion, he ORPL a ises as a ansi ion om hese s a es o he g ound s a e |0� , while he PL band a ound 850 nm a ises due o he ansi ion om he s a e |3� o an in e media e s a e |1� . Ob iously, Figu e4. DCP da a eco ded a 585 and 850 nm upon exci a ion a 450 and 532 nm, espec i ely. (a) Da a a e shown as a unc ion o empe a u e a ixed magne ic ields. Black do ma ke s— expe imen al da a, ainbow- colou ed su ace— heo e ical DCP calcula ed using pa ame e s om he da a i o Eq.(1) wi h i=2 . The ed colou plane indica es he ze o pola iza ion le el. b) Magne ic ield dependence o DCP a ixed empe a u es. (c) Da a eco ded a 850 nm a e shown as a unc ion o empe a u e a ixed magne ic ields. Ma ke s— expe imen al da a, lines— heo e ical DCP calcula ed om he da a i o Eq.(1) wi h i=1 . (d) DCP da a shown as a unc ion o magne ic ield a ixed empe a u es. 6 Vol:.(1234567890) Scien i ic Repo s | (2021) 11:7766 | h ps://doi.o g/10.1038/s41598-021-87290-z www.na u e.com/scien i ic epo s/ i he ene gy ba ie s E 23 and E 32 be ween he s a es |2� and |3� a e su icien ly high (E 23 , E 32 ≫1/kT ), hen he DCP measu ed in he ORPL band can be ep esen ed as a sum o indi idual con ibu ions. Assuming ha he MCPL is caused only by xy-pola ized elec ic dipole ansi ions, and g�≫g⊥ , he pa a- magne ic C - e m o DCP is40: whe e he summa ion is o e all he ansi ions om di e en exci ed s a es con ibu ing a he gi en WL, n=cos(θ) , A sa i a e he sa u a ion cons an s, i a e he co esponding ZFS’s be ween wo magne ic sub-le els due o he low symme y componen o he c ys al ield, ˜gi a e he e ec i e g- ac o s, which, o example, a e equal o 4g,i o 2g,i o a spin quin e ( S=2 ) and a spin iple ( S=1 ), espec i ely. This exp ession is alid o any se o s ongly aniso opic and non-in e ac ing ( e m B=0 ) double s, K ame s o non-K ame s. I some pa - icula ansi ion in ol es hal -in ege spin s a es, hen he co esponding ZFS ene gy i igo ously equals ze o. The wo-dimensional i o he expe imen al da a se DCP(T,B) o he Eq. (1) is shown in Fig.4a o WL 585 nm. I can be seen ha ou simpli ied model simula es he empe a u e and magne ic ield dependences eason- ably well, hough, i can be u he imp o ed by aking in o accoun possible ansi ions be ween s a es |2� and |3� . The pa ame e s o spin Hamil onian o he co esponding exci ed s a es a e gi en in Table1, which, o he sake o comple eness, is supplemen ed by he esul s ob ained om he MCD measu emen s (ORPL cen e) and MCPL in he lasing cen e (p esumably, Bi + ion)34,39,40. I is wo h no ing, ha wi h he ZFS pa ame e ixed o ze o, i was impossible o ge any sa is ac o y i . This allows us o make he unequi ocal conclusion ha a single Bi 2+ ion o , as a ma e o ac , any o he sys em wi h a hal -in ege spin canno cause his o ange– ed luminescence. The igh panel in Fig.5 shows he ine s uc u e o ene gy le els. This model depic s he simples case wi h he iple g ound and second exci ed s a es since nei he MCD no MCPL gi es he exac alue o he e ec i e spin S. One can see ha he nega i e MCPL componen o igina es om he exci ed s a e wi h he ZFS =4.5 cm −1 , while he posi i e one om he s a e wi h =21 cm −1 . The e ec i e ˜g - ac o o he exci ed s a e wi h a high ZFS (see Table1) is small and i has a la ge i unce ain y. This is no su p ising, since he main con ibu ion o (1) C (T,B)= N � i=1 Asa i�1 0 ˜giµBBn4 � �2 i+(˜giµBBn)2 anh  � �2 i+(˜giµBBn)2 2kT  dn, Figu e5. Ene gy le els diag am o he ORPL cen e in SiO 2 :Bi. (a) Gene al layou . Wa y a ows co espond o as non- adia i e elaxa ion. Dashed a ow: possible, bu unobse ed ye ansi ion. (b) Magne ic suble el s uc u e. All suble els a e labelled by he quan um numbe s MS co esponding o he quan ized p ojec ions o he spin S. Table 1. Ze o ield spli ings (ZFS), sa u a ion cons an s and e ec i e ˜g - ac o s o ORPL and NIR PL (Bi + ion) cen es in SiO 2 :Bi glass. The exci a ion WL o he MCPL is gi en in pa en heses. Bi + ORPL cen e Asa ˜g ZFS (  ) Wa eleng h, nm Wa eleng h, nm cm −1 MCD – 465 ( 2.8 ±0.03 ) ×10 −3 2.16 ±0.04 1.22 ±0.11 MCPL 580 (450) −1.8 ±0.3 1.4 ±0.16 5.1 ±0.4 11 ±0.6 0.44 ±6 17 ±2 640 (532) −1.4 ±0.2 1.28 ±0.14 4.5 ±0.3 7.6 ±0.5 0.56 ±8 21 ±2.5 850 (532) −7.6 ±0.8 0.86 ±2 21 ±3 830 (375) 1.26 ±0.08 1.4 ±10 97 ±6 1440 (375) −1.38 ±0.06 4.28 ±0.17 6.18 ±0.24 7 Vol.:(0123456789) Scien i ic Repo s | (2021) 11:7766 | h ps://doi.o g/10.1038/s41598-021-87290-z www.na u e.com/scien i ic epo s/ he spli ing o double sub-le els is due o ZFS, while he Zeeman e m in he a ailable ange o magne ic ields is no su icien ly la ge o he p ecise g- ac o de e mina ion. I is also shown in Table1 ha he se o pa am- e e s ob ained om he DCP da a o NIR1 band (850 nm) is p ac ically iden ical o ha o he posi i e DCP componen o ORPL. This is e y s ong e idence ha hese ansi ions ha e he same ini ial exci ed sub-s a e and di e en e mina ing s a es esul ing in opposi e DCP signs. Acco ding o he model o Fig.5b, he NIR1 band is caused only by he ansi ion om he s a e |3� o he i s (in e media e) exci ed s a e |1� . The subsequen ansi ion om |1� o he g ound s a e, i i exis s, should p oduce NIR luminescence beyond 2 µ m, and pe haps his is he eason why i has no been epo ed so a . The p oposed ene gy le els diag am nicely explains he obse ed ze oing o he A - e m a 13 K in he ORPL band. We ecall ha he A- e m is empe a u e independen acco ding o he s anda d heo y41–44. Fo his eason, we belie e ha he obse ed A - e m should be conside ed as a pseudo A - e m, he appea ance o which is associ- a ed wi h wo o e lapping C - e ms o opposi e signs46. Fi s , acco ding o Fig.5b, only he ansi ion |2�−→|0� p oduces a genuine A - e m as a esul o he g ound s a e spli ing. Howe e , due o he ela i ely small alue o his spli ing, W , on he one hand, and he la ge wid h o he luminescence band, �ν , on he o he hand, we would expec he diamagne ic e m o be negligibly small e en a high magne ic ields since A≃�W/�ν . Second, in ou model, he s a es |2� and |3� ha e he same exci a ion p obabili ies, meaning ha hey a e equally popula ed a low empe a u es, and he ansi ions o he g ound s a e a e o compa able in ensi ies. Howe e , he absolu e deg ee o pola iza ion is highe o emission om he s a e wi h a small ZFS, i. e. s a e |2� . This is because he DCP is in e sely p opo ional o he ze o- ield spli ing  a low and in e media e magne ic ields ( ˜gµBB≤� ). As he esul , he DCP is nega i e and has a non-ze o slope a 1.47 K. As he empe a u e ises, he C - e m o s a e |2� dec eases as e han ha o s a e |3� , educing he pseudo A - e m. Addi ionally, he possible elaxa ion |2�−→|3� o e he ba ie E 23 con ibu es o he edis ibu ion o he exci ed s a es’ popula ions in a ou o he ansi ion |3�−→|0� . This p ocess can also dec ease he con ibu ion o he s a e |2� o he DCP, which leads o ze oing o he pseudo A - e m. Acco ding o he selec ion ules, he sign o MCPL is de ined by he change o he magne ic momen ’s p o- jec ion µz=−µBgMS du ing he ansi ion be ween he ene gy le els. Since in ou model he inal s a e is he same and has he ze o p ojec ion ( MS=0 ), he wo componen s o MCPL can ha e he opposi e signs only i he co esponding exci ed s a es ha e he opposi e momen s µz . Consequen ly, he nega i e MCPL, e.g. he le -hand pola ized componen σ − , o igina es om he s a e wi h a posi i e µz , and he σ+ componen om he s a e wi h a nega i e µz . In e ms o he spin Hamil onian i means ha he non-K ame s double |±1� o he s a e |3� has a nega i e g- ac o . The appea ance o a nega i e g- ac o is possible in a sys em wi h la ge spin S and in e ac ions ha in oduce non-diagonal e ms o he spin Hamil onian. A simple example is a Yb 3+ ion in he oc ahed al en i onmen 47. Chibo a u and Ungu 48 discussed possible mechanisms o nega i e g- ac o s in lan hanide and ansi ion me al complexes. They concluded ha he necessa y condi ion o a nega i e g- ac o is he p esence o s ong ou h- and highe -o de pe u ba ions, namely he ZFS e ec s and/o s ong spin-o bi coupling. Re u ning o Bi 2+ , i indeed exhibi s a e y s ong spin-o bi in e ac ion, which spli s he 2 P e m o he 6p elec on o he g ound s a e 2 P 1/2 and he i s exci ed s a e 2 P 3/2 . Howe e , he ou h- and highe -o de pe u ba ion e ec s do no a ise o he s a es o qua e and lowe mul iplici y47,49. This is ye ano he ac excluding Bi 2+ as he o igin o he ORPL in Bismu h-doped silica glass. Ano he class o sys ems ha can demons a e a nega i e g- ac o is clus e s o ions. The exchange coupling mixes he s a es o indi idual ions p oducing simila e ec s as high-o de spin-o bi and c ys al ield in e ac ions. Piligkos e al.50 in es iga ed ansi ion me al dime s wi h a weak exchange coupling using MCD spec oscopy. Among di e en model sys ems hey obse ed a e e se o he MCD signal sign in an an i e omagne ically coupled C (III)Ni(II) dime a he empe a u es abo e ≃ 15 K. The g ound s a e o he sys em consis s o h ee sub-le els wi h he lowes spin s a e S = 1/2. The o he wo spin s a es o he sys em, S = 3/2 and S = 5/2, lie a 20.1 and 53.6 cm −1 abo e he lowes one, espec i ely. Magne ic momen s o he exci ed s a es a e opposi e o he magne ic momen o he S = 1/2 g ound double . Abo e 15 K he popula ion o exci ed s a es becomes su - icien ly high o p oduce a majo con ibu ion o MCD, which e e ses i s sign. Taking in o accoun he abo e conside a ions, i should be clea ha he o ange– ed luminescence can no be assigned o a single Bi 2+ ion. This ac i e cen e mus ha e a mo e complex elec onic s uc u e. The simples objec s ha could sa is y he ea u es obse ed in he MCD and MCPL expe imen s a e, o example, homo- nuclea dime s. I is wo h ecalling he e ha Bismu h ions a e cha ac e ized by ex ended ou e p-o bi als, which enhances a endency o clus e ing. On his basis, i can be assumed ha di alen Bismu h ions compose a dime [Bi 2+ — Bi 2+ ] ha is esponsible o he o ange– ed pho oluminescence. Al e na i ely, one can assume ha some kind o a s able complex o med by a Bi 2+ ion and an oxygen acancy ([Bi – V O ] n+ ) is he o igin o he ORPL. Un o una ely, MCPL is no capable o dis inguishing hese wo cases. As a solu ion, he op ically de ec ed nuclea magne ic esonance can be used. This echnique can p o ide he same in o ma ion as he con en ional NMR bu i is mo e selec able since i p obes he nuclei o he luminescence cen e only51,52. On he o he hand, ORPL is obse ed e en a e y low doping le els o Bismu h ( ∼ 1 ppm) a which o ming o dime s is qui e imp obable, and he e o e he second assump ion seems o be mo e ealis ic. Discussion and conclusions De ailed in es iga ions o magne ic ci cula pola iza ion o he o ange– ed pho oluminescence band in Bis- mu h-doped silica glass wi hou o he co-dopan s we e pe o med a a iable empe a u e and magne ic ield. The expe imen s show a he complex beha iou o he deg ee o ci cula pola iza ion, which exhibi s a non- mono onic empe a u e and ield dependence, and sign e e sal. Such beha iou canno be explained by he gene ally accep ed assump ion o Bi 2+ being he o igin o he ORPL. The analysis o VTVH cu es e ealed 8 Vol:.(1234567890) Scien i ic Repo s | (2021) 11:7766 | h ps://doi.o g/10.1038/s41598-021-87290-z www.na u e.com/scien i ic epo s/ ha he exci ed s a es o he ORPL cen e a e magne ic mul iple s wi h la ge ze o- ield spli ing. We p oposed a consis en model o ene gy le els o he ORPL cen e, which explains all he peculia i ies obse ed in MCPL and MCD expe imen s. We assume ha he sou ce o his he mally s able o ange– ed pho oluminescence can be homonuclea Bismu h dime s ([Bi 2+ – Bi 2+ ], o ins ance) o complexes consis ing o a Bi 2+ ion and an oxygen acancy [Bi – V O ] n+ , and each o hese cen es mus be a sys em wi h an e en numbe o elec ons. Me hods Ma e ials. In he p esen in es iga ion, wo ypes o ma e ial we e used. The i s ype o he samples was manu ac u ed using he mel ing o SiO 2 and Bi 2 O 3 powde s in silica ubes53. The size o pa icles was 100–200 and 1–5 µ m o silica and bismu h oxide, espec i ely. The mass a io o he oxides was chosen so ha he a omic a io Bi/Si ob ained in he glass was abou o 5000 ppm. The de ailed p ocedu e o sample ab ica ion was e y close o ha desc ibed p e iously by I. Bu e o e al.54. The second ype o ma e ial used in ou expe imen was a Bismu h-doped silica glass p e o m manu ac u ed om nano-po ous silica xe ogels. The xe ogels we e doped wi h bismu h by soaking in an ace one solu ion o Bi-con aining p ecu so complex. A e he dehyd oxyla ion, he xe ogels we e sin e ed a 1300 ◦ C in a helium a mosphe e. A mo e de ailed desc ip ion o his ma e ial ab ica ion was epo ed elsewhe e35,55,56. The samples p epa ed om his ma e ial we e he same ha we p e iously used in he MCD expe imen s39, and we ecall he e ha he a omic a io Bi/Si es ima ed om he s anda d elec on p obe mic oanalysis (EPMA) was abou 400 ppm. Bo h ypes o samples we e cu and polished o ge he dimensions o abou 2×4×5 mm 3 . The samples showed e y simila esul s. We p esen ed he e he da a ob ained om he second ype o ma e ial only since i was used in he MCD measu emen s epo ed be o e39. Measu emen s. Expe imen s in he empe a u e ange 1.4–300 K we e pe o med in he closed cycle mag- ne o-op ical c yos a (Spec omagPT, Ox o d Ins .). The he mal s abili y o he samples a ached o he holde o he a iable empe a u e inse was abou 0.01 K, excep he ange om 4.2 o 10 K, whe e he he mal s abili y was abou 0.05 K. Lase diodes and equency doubled Nd:YAG lase s we e used o he exci a ion a di e en wa eleng hs. In he expe imen s wi h con inuous wa e exci a ion he qua e -wa e e a da ion ( /4 ) was in oduced by he pho oelas ic modula o (I/FS-20, Hinds Ins umen s) a he equency 20.077 kHz. The esul ing PL emission in he ange o 460–900 nm was analysed by he ixed uncoa ed Glan-Thompson pola ize , hen il e ed by a monoch oma o and de ec ed wi h he cooled GaAs pho omul iplie (R943-02, Hamama su Inc.). The pho on coun ing echnique (P7887 scale , Fas ComTec) was used o eco d he signal because o i s ou s anding signal o noise a io. In he expe imen s wi h pulsed exci a ion he pho oelas ic modula o was eplaced wi h a supe - ach oma ic /4 wa epla e. O he equipmen and de ails o he expe imen s we e desc ibed elsewhe e40. The spec al esolu ion in he in es iga ed WL ange was 1.5 nm. Da a a ailabili y Expe imen al da a shown in he g aphs h oughou his a icle a e a ailable om he co esponding au ho upon easonable eques . Recei ed: 27 Janua y 2021; Accep ed: 26 Ma ch 2021 Re e ences 1. Boisbaud an, M. Fluo escence des composés du bismu h soumis à l’e lu e élec ique dans le ide. C. R. Acad. Sci. Pa is 103, 629–621 (1886). 2. Blasse, G., Meije ink, A., Nomes, M. & Zuidema, J. Unusual bismu h luminescence in s on ium e abo a e S B 4 O 7 :Bi. J. Phys. Chem. Solids 55, 171–174 (1994). 3. Hams a, M. A., Folke s, H. F. & Blasse, G. Red bismu h emission in alkaline-ea h-me al sul a es. J. Ma e . Chem. 4, 1349–1350 (1994). 4. de Jong, M. & Meije ink, A. Colo uning o Bi 2+ luminescence in ba ium bo a es. Jou nal o Luminescence 170, 240–247 (2016). 5. S i as a a, A. Luminescence o di alen bismu h in M 2+ BPO 5 (M 2+ = Ba 2+ , S 2+ and Ca 2+ ). Jou nal o Luminescence 78, 239–243 (1998). 6. Cao, R. e al. Syn hesis and luminescence p ope ies o no el ed-emi ing R 3 P 4 O 13 :Bi 2+ (R = S and Ba) phospho s. Jou nal o Alloys and Compounds 661, 77–81 (2016). 7. Li, L., Cao, J., Viana, B., Xu, S. & Peng, M. Si e Occupancy P e e ence and An i he mal Quenching o he Bi 2+ Deep Red Emission in β -Ca 2 P 2 O 7 : Bi 2+ . Ino g. Chem. 56, 6499–6506 (2017). 8. Cao, R., Zhang, F., Liao, C. & Qiu, J. Yellow- o-o ange emission om Bi 2+ -doped RF 2 (R = Ca and S ) phospho s. Op . Exp ess 21, 15728 (2013). 9. Ga , M. e al. The luminescence o Bi, Ag and Cu in na u al and syn he ic ba i e BaSO 4 . Op ical Ma e ials 16, 279–290 (2001). 10. Puchalska, M., Zych, E. & Bolek, P. Luminescences o Bi 3+ and Bi 2+ ions in Bi-doped CaAl 4 O 7 phospho powde s ob ained ia modi ied Pechini ci a e p ocess. Jou nal o Alloys and Compounds 806, 798–805 (2019). 11. Peng, M., Da, N., K olikowski, S., S iegelschmi , A. & Wond aczek, L. Luminescence om Bi 2+ -ac i a ed alkali ea h bo ophos- pha es o whi e LEDs. Op . Exp ess 17, 21169 (2009). 12. Cao, R., Peng, M. & Qiu, J. Pho oluminescence o Bi 2+ -doped BaSO 4 as a ed phospho o whi e LEDs. Op . Exp ess 20, A977 (2012). 13. K asniko , A., Mihoko a, E., Nikl, M., Zazubo ich, S. & Zhydache skyy, Y. Luminescence spec oscopy and o igin o luminescence cen e s in bi-doped ma e ials. C ys als 10, 208 (2020). 14. Xu, W., Peng, M., Ma, Z., Dong, G. & Qiu, J. A new s udy on bismu h doped oxide glasses. Op . Exp ess 20, 15692 (2012). 15. De Jong, M., Meije ink, A., Go don, R., Ba andia án, Z. & Seijo, L. Is Bi 2+ esponsible o he ed-o ange emission o Bismu h- doped S B 4 O 7 ? J. Phys. Chem. C 118, 9696–9705 (2014). 9 Vol.:(0123456789) Scien i ic Repo s | (2021) 11:7766 | h ps://doi.o g/10.1038/s41598-021-87290-z www.na u e.com/scien i ic epo s/ 16. Mu phy, H. J. e al. Op ical and EPR cha ac e iza ion o poin de ec s in bismu h-doped CdWO 4 c ys als. Radia . E . De ec . S. 149, 273–278 (1999). 17. Vazhenin, V. A., Po apo , A. P., Asa yan, G. R. & Nikl, M. Pho osensi i e bismu h ions in lead ungs a e. Phys. Solid S a e 55, 803–806 (2013). 18. Lecoq, P. e al. Lead ungs a e PbWO 4 scin illa o s o LHC EM calo ime y. Nucl. Ins um. 365, 291–298 (1995). 19. Mu a a, K., Fujimo o, Y., Kanabe, T., Fuji a, H. & Naka suka, M. Bi-doped SiO 2 as a new lase ma e ial o an in ense lase . Fusion Eng. Des. 44, 437–439 (1999). 20. Fujimo o, Y. & Naka suka, M. In a ed luminescence om Bismu h-doped silica glass. Jpn. J. Appl. Phys. 40, L279–L281 (2001). 21. Diano , E. M. e al. CW bismu h ib e lase . Quan um Elec on. 35, 1083–1084 (2005). 22. Razdob ee , I. e al. E icien all- ibe bismu h-doped lase . Appl. Phys. Le . 90, 031103 (2007). 23. Bu e o , I. & Diano , E. Bi-doped ibe lase s. Lase Phys. Le . 6, 487–504 (2009). 24. Chamo o skiy, A., Rau iainen, J., Ran amäki, A., Golan , K.M. & Okho niko , O.G. semiconduc o disk lase . Op . Exp ess 19, 6433–6438 (2011). 25. Kali a, M. P., Yoo, S. & Sahu, J. K. In luence o cooling on a bismu h-doped ibe lase and ampli ie pe o mance. Appl. Op . 48, G83–G87 (2009). 26. Diano , E.M. Bismu h-doped op ical ibe s: A challenging ac i e medium o nea -IR lase s and op ical ampli ie s. Applica ions 1, 1–7 (2012). 27. Gumenyuk, R. e al. ope a ing in anomalous and no mal dispe sion egimes. Op . Le . 38, 4005–4007 (2013). 28. Bu e o , I. e al. Bi-doped op ical ibe s and ibe lase s. IEEE J. Sel. Top. Quan . 20, 0903815 (2014). 29. Thippa apu, N. K., Jain, S., Umniko , A. A., Ba ua, P. & Sahu, J. K. 1120 nm diode-pumped Bi-doped ibe ampli ie . Op . Le . 40, 2441–2444 (2015). 30. Fi s o , S. V. e al. A 23-db bismu h-doped op ical ibe ampli ie o a 1700-nm band. Sci. Repo s 6, 28939 (2016). 31. Thippa apu, N. K. e al. 40 dB gain all ibe bismu h-doped ampli ie ope a ing in he O-band. Op . Le . 44, 2248–2251 (2019). 32. Diano , E. M. Na u e o Bi- ela ed nea IR ac i e cen e s in glasses: S a e o he a and i s eliable esul s. Lase Phys. Le . 12, 095106 (2015). 33. Romano , A., Haula, E., Shashkin, D., V yu ina, D. & Ko chak, V. On he o igin o nea -IR luminescence . J. Lumin. 183, 233–237 (2017). 34. Lagu a, O. V. & Razdob ee , I. M. O igin o nea -in a ed luminescence in Bismu h-doped silica and ge manosilica e glass ibe s: C ys al ield modeling. Op . Ma e . 84, 103–108 (2018). 35. Razdob ee , I. e al. Op ical spec oscopy o Bismu h-doped pu e silica ibe p e o m. Op . Le . 35, 1341–1343 (2010). 36. Sun, H.-T., Zhou, J. & Qiu, J. Recen ad ances in bismu h ac i a ed pho onic ma e ials. P og. Ma e . Sci. 64, 1–72 (2014). 37. Zhou, S. e al. Mul i unc ional Bismu h-doped nanopo ous silica glass: F om blue-g een, o ange, ed, and whi e ligh sou ces o ul a-b oadband in a ed ampli ie s. Ad . Func. Ma e . 18, 1407–1413 (2008). 38. Vebe , A., Cicconi, M. R., Pu i, A. & de Ligny, D. Op ical p ope ies and bismu h edox in Bi-doped high-silica Al-Si glasses. J. Phys. Chem. C 122, 19777–19792 (2018). 39. Lagu a, O. V., El Hamzaoui, H., Bouazaoui, M., A ion, V. B. & Razdob ee , I. M.: On he na u e o pho oluminescence in Bismu h- doped silica glass. Sci. Rep. 7, 3178 (2017). 40. Lagu a, O., El Hamzaoui, H., Bouazaoui, M., A ion, V. B. & Razdob ee , I. Magne ic ci cula pola iza ion o luminescence in Bismu h-doped silica glass. Op ica 2, 663–666 (2015). 41. Zapasskii, V. S. & Feo olo , P. P. De elopmen o pola iza ion magne o-op ics o pa amagne ic c ys als. So . Phys. Usp. 18, 323–342 (1975). 42. Riehl, J. P. & Richa dson, F. S. Theo y o magne ic ci cula ly pola ized emission. The Jou nal o Chemical Physics 66, 1988–1998 (1977). 43. Scha z, P. N., Mowe y, R. L. & K ausz, E. R. 1978 M.C.D./M.C.P.L. sa u a ion heo y wi h applica ion o molecules in D ∞h and i s subg oups. Mol. Phys. 35, 1537–1557. 44. Solomon, E. I., Pa el, E. G., Loeb, K. E. & Campochia o, C. Magne ic ci cula dich oism spec oscopy as a p obe o he geome ic and elec onic s uc u e o non-heine e ous enzymes. Coo dina ion Chemis y Re iews 44, 369–460 (1995). 45. Liu, F. e al. Spin dynamics o nega i ely cha ged exci ons in CdSe/CdS colloidal nanoc ys als. Phys. Re . B 88, 035302 (2013). 46. Mason, W. R. A P ac ical Guide o Magne ic Ci cula Dich oism Spec oscopy (John Wiley & Sons, 2007). 47. Ab agam, A. & Bleaney, B. Elec on pa amagne ic esonance o ansi ion ions (Cla endon P ess, Ox o d, UK, 1970). 48. Chibo a u, L. F. & Ungu , L. Nega i e g ac o s, Be y phases, and magne ic p ope ies o complexes. Phys. Re . Le . 109, 246403 (2012). 49. B own, J. M. & Ca ing on, A. Ro a ional spec oscopy o dia omic molecules (Camb idge Uni e si y P ess, 2003). 50. Piligkos, S. e al. Magne ic ci cula dich oism spec oscopy o weakly exchange coupled ansi ion me al dime s: A model s udy. Coo . Chem. Re . 253, 2352–2362 (2009). 51. Mlynek, J., Wong, N. C., DeVoe, R. G., Kin ze , E. S. & B ewe , R. G. Raman he e odyne de ec ion o nuclea magne ic esonance. Phys. Re . Le . 50, 993–996 (1983). 52. Wong, N. C., Kin ze , E. S., Mlynek, J., DeVoe, R. G. & B ewe , R. G. Raman he e odyne de ec ion o nuclea magne ic esonance. Phys. Re . B 28, 4993–5010 (1983). 53. Ne , M., Romano, V. & Lü hy, W. Me al-doped ib es o b oadband emission: Fab ica ion wi h g anula ed oxides. Op . Ma e . 31, 247251 (2008). 54. Bu e o , I. A. e al. Op ical p ope ies o ac i e bismu h cen es in silica ib es con aining no o he dopan s. Quan um Elec on. 40, 639–641 (2010). 55. El Hamzaoui, H. e al. F om po ous silica xe ogels o bulk op ical glasses: The con ol o densi ica ion. Ma e . Chem. Phys. 121, 83–88 (2010). 56. El Hamzaoui, H. e al. F om molecula p ecu so s in solu ion o mic os uc u ed op ical ibe : A sol-gel polyme ic ou e. Op . Ma e . Exp ess 1, 234–242 (2011). Acknowledgemen s We hank H. El Hamzaoui and M. Bouazaoui o he syn hesis o nano-po ous silica xe ogels and sample p epa- a ion, V. B. A ion o he syn hesis o Bismu h p ecu so s. We also wan o hank P o . M. Baye and P o . D. Yako le om Technische Uni e si ä Do mund (Ge many) o he oppo uni y o conduc compa a i e measu emen s in hei labo a o y. This wo k has been suppo ed by “Agence Na ionale de la Reche che” ia ANR g an “BOATS” 12BS04-0019-01. OL is suppo ed by he Eu opean Social Fund unde he p ojec CZ.02. 2.69/0.0/0.0/19_074/0016239. Au ho con ibu ions Bo h au ho s con ibu ed equally o his wo k.