J. MATER. CHEM., 1994,4( l), 139-144 Electroluminescence of Ho3+ Ions in Semiconducting Polycrystalline Zinc Oxide Electrodes in Contact with Aqueous Electrolyte Jean-Claude Ronfard-Haret, Kazufumi Azuma, Sai'd Bachir, Djakaridia Kouyate and Jean Kossanyi Laborafoire de Photochirnie Solaire, CNRS, 2-8, rue Henri Dunant, 94320 Thiais, France Sintered semiconducting zinc oxide has been doped with varying amounts of Ho3+ ions and its electroluminescence has been studied under various polarization conditions. Under cathodic polarization and in the presence of persulfate ions, the emission of the ZnO matrix only was observed. Conversely, under anodic under anodic polarization, character- istic emission bands of the Ho3+ ions were generated. The intensity of these bands was studied as a function of both the polarization potential and the rare earth doping level.In the 9.8-10.2 V vs. SCE range, the log of the emission intensity varied inversely with the square root of the applied potential, according to the Alfrey-Taylor relationship. It showed a maximum for a Ho3+ doping level of ca. 0.5 atom%. Analysis of the samples' polycrystalline structure using scanning electron microscopy (SEM) and energy dispersive spectroscopy (EDS) techniques indicates that the rare- earth ions create islands on the surface of ZnO which induce dislocations in the polycrystalline structure. The electroluminescence of rare earth (RE3 +)-doped semi- conductors is attractive because of its numerous applications.' Surprisingly, of the RE3 + -doped 11-VI semiconductors, most literature deals with sulfides, selenides or tellurides, and only do our recent articles report the electroluminescence of the RE3+ ions inserted into semiconducting zinc o~ide.~-~ Rare earth-doped zinc oxide electrodes show two types of electroluminescence depending upon the polarization of the electrode in the electrochemical cell. Under negative bias the light originates from the recombination of electron-hole pairs (e-/h+).The spectrum shows two features: a sharp band centered at 390 nm corresponding to the direct band to band radiative recombination, and a broad band centred around 550 nm which corresponds to recombinations at self-activated centres intrinsic to ZnO. This electroluminescence observed under negative bias is very similar to the photoluminescence of zinc oxide obtained by 390 nm ex~itation.~,~The only difference lies (i) in the way in which the holes are created in the valence band: by injection from a redox couple in solution for the electroluminescence process and by the promotion of an electron from the valence band to the conduction band in the case of photoluminescence; (ii) in the intensity of the reabsorption, by the rare earth incorporated into the matrix, of the light emitted by zinc o~ide.~?~?~ Under positive bias, the characteristic emission of the RE3+ ions is observed in addition to the luminescence originating from zinc oxide.8 Under such conditions, the luminescent centres of zinc oxide are excited by two different processes: direct electron impact excitation or impact ionization of the medium, while excitation of the RE3+ ions occurs only by a direct electron impact-excitation process.The above fundamental results which differ with the polariz- ation mode have been interpreted in terms of the absence of coupling between the energy levels of the semiconducting matrix and those of the lanthanide. This is corroborated subsequently by SEM which shows that the RE3 + ions induce dislocations inside the zinc oxide microcrystalline structure5 as it does inside zinc sulfide.' This result lead us to question the method of rare earth insertion into zinc oxide. In a previous paper,, we reported some preliminary results concerning the electroluminescence of Ho3+ and Sm3 ions+ when inserted in semiconducting zinc oxide.In the present article we report a systematic study of the electroluminescence of the Ho3+ ions inserted in ZnO, and the results obtained by SEM and EDS techniques carried out upon polycrystal- line samples. Experimental Sintered RE3 +-doped polycrystalline ZnO pellets were pre- pared by mixing intimately 0.04-5.0 atom% of the lanthanide (99.9% pure from Rhbne-Poulenc) to zinc oxide (99.99% pure from Koch-Light) in the presence of a small amount of ethanol. The mixture of the two oxides was pressed (4 tons cmP2) in a Specac press, then the pellets were heated for 5 h at 1100"C, under atmospheric pressure in an Adamel Lhomargil furnace. The electrodes were prepared by covering one face of the pellet with the In-Ga eutectic (both In and Ga metals from Prolabo). A copper wire was fixed to this ohmic contact by silver resin (Eco-Bond, Emerson and Cummings).The resulting preparation was then glued to a glass support by a chemically inert epoxy resin (Araldite). The electrodes were polished and etched with 1mol dm-3 HCI. The electrochemical cell used was a classical three-electrode system with a saturated calomel electrode (SCE) as reference and a platinum counter-electrode. Potential polarizations were generated by a P.A.R. model 173 potentiostat monitored by a P.A.R. model 175 programmer system. For the electrofumi- nescence measurements, the electrochemical cell replaced the usual cell holder of a Perkin-Elmer MPF-44B spectrofluori-meter. The photocurrent spectra and the photoresponse of the electrodes were measured using an irradiation set-up built in the laboratory with a 450Watts XBO Xenon lamp (Cunow) and an H25 monochromator (Jobin-Yvon). In order to pre-vent rapid ageing of the electrode, pulsed (instead of continu- ous) polarizations, between 0 V and the working potential, were applied for recording the electroluminescence spectra.Characterization of the electroluminescence under cathodic polarization was performed with an NaCl (1 mol dmP3) + Na,S208 (0.5 mol dmP3) aqueous solution; for the anodic polarization measurements, the electrolyte was a 1mol dmP3 NaCl aqueous solution sometimes containing 0.3 rnol dm -3 Na,S,03.For time-resolved experiments, the duration of the anodic bias pulse and the relaxation time were adjusted separately to achieve a 95% signal decay. The synchronized electrical signals one originating from the potentiostat and the other from the photomultiplier, were recorded on a Tektronix 7834 storage oscilloscope. The charge resistance was adapted to the resolution needed. The energy dispersive detector (Tracor Northern system) was mounted on a scanning electron microscope (JEOL JSM 840 A). The major components of the sample were identified by superimposing labeled cursors to each element over a recorded spectrum. The chemical analysis of the spots of micrographs was determined by looking at the energy disper- sive X-ray spectra obtained under electronic irradiation. X-ray fluorescence spectrometry is a valuable technique for element analysis at the submicrometre level.The electron beam focussed onto the sample generates X-rays which correspond to the electronic relaxation between the initial and final excited states of the atom and is then analysed by the spectrometer. Results and Discussion Semiconducting properties The semiconducting properties of the Ho3 +-doped ZnO elec- trodes were verified by photocurrent measurements in a neutral medium (pH 7, phosphate buffer) in the presence of 0.4 mol dmP3 Na,SO,. Despite the absorption of light by the Ho3+ ions around 450 and 550 nm6,7, the photocurrent action spectra display only, as in the case of Er3+-doped ZnO electrode^,^ the characteristic response of the band to band absorption of zinc oxide below 390 nm.The current-voltage response of the electrodes is typical of an n-type semicon- ductor. As observed here and already reported for the Sm3+- doped electrode^,^ the cyclic voltammograms indicate that the semiconducting properties of the RE3+-doped electrodes are less marked than those of the undoped ones. Electroluminescence measurements Cathodic Polarization In the absence of persulfate ions and for applied potentials ranging from 0 to -4 V us. SCE, no light emission is observed which would originate from the electrode. When persulfate ions are added to the solution a white-greenish light emission is observed the spectrum of which is given in Fig.1. As reported previously for other RE3+ -doped ZnO electrodes, this spectrum exhibits two patterns: a sharp band centred at 390 nm which corresponds to band-to-band e-/h+ recombina- tions, and a broad band centred around 550 nm corresponding to e-/h+ recombinations at ZnO intrinsic self-activated centres. The mechanisrn6,l0,l' put forward to explain the electroluminescence phenomenon involves the injection of holes into the valence band of the semiconductor. These holes are produced by the second reaction of the two-step reduction of the persulfate ions: I I I 400 500 600 700 wavelengthhm Fig. 1 Electroluminescence spectrum under cathodic polarization of a 0.6 atom% Ho3+-doped ZnO electrode pulsed between 0.0 and -4.0 V in 1 mol dm-3 aqueous Na,S,O, electrolytic solution J.MATER. CHEM., 1994, VOL. 4 where the subscripts CB and VB refer to conduction and valence bands, respectively. The redox potential of the second reaction lies below the valence band edge." Therefore, although electron/hole pairs are created in the semiconductor, neither emission from the rare-earth ion, nor any emission originating from a new emitting centre could be detected. The only difference with the electroluminescence of pure zinc oxide comes from the weak reabsorption around 560nm by the Ho3+ ions of the light emitted by ZnO, as observed previously under the same conditions when Er3+ (ref. 3) and Nd3+ (ref. 4) ions are the dopant. Anodic Polarization In 1 mol dm-3 NaC1, the etched Ho3+-doped polycrystalline semiconducting zinc oxide electrodes show a weak green light emission when submitted to a polarization potential higher than +5.5 V us.SCE (Fig. 2). An anodic current is also detected. Both reach a plateau between 7 and 9.7 V us. SCE. Increasing the applied potential above 9.7 V us SCE induces a strong increase of both the emitted light and the anodic current. The variation of the anodic current is similar to the one already reported by Kiess12 for pure zinc oxide monocrys- talline electrodes. The only difference is the value of the second potential threshold found for the anodic current which occurs after the plateau: 9.7 V us. SCE as compared to the ca. 50 V reported by Kiess12 for monocrystalline ZnO.The spectrum obtained is characteristic of the Ho3+ ions (Fig. 3). By analogy with published spectra13p15 of the Ho3+/ZnS electro-luminescent system, we have attributed the series of bands centred around 500, 550, 650 and 760nm to the 5F3+518, 5s2+518, 5F3-+517, transitions of and 5s2-+s17 the Ho3+ion, respectively. In the 5.5-9.7 V t's. SCE potential range, only the most intense emission at 550 nm, correspond- ing to the 5S2-+518 transition, could be detected. If the electrode has not been etched or if it has been purposefully aged, the electroluminescence and the anodic current can only be detected when the applied potential reaches ca. +9.7 V us. SCE. However, in a pure solution of NaC1, the light emission is unstable, and its intensity decreases with time.A partial dissolution of the electrode and an oxygen evolution are observed. As already disc~ssed,~-~ the addition of thiosulfate ions to the electrolytic solution stabilizes the emission of light and prevents both the dissolution of the electrode and the evolution of gas. But once the electrode has been aged, the addition of thiosulfate ions to the electrolytic solution does not allow recovery from the loss of both current and luminescence observed previously at low voltage; only 0.51 1 I, I "'I' I, ~ E 0.30.41 0.21 I I lloo 0 24 6 8 10 12 VN vs. SCE Fig.2 Dependence of the anodic current i (solid line) and of the emitted light B (points) observed at 550nm i:ersus the applied potential V vs.SCE for an etched 0.5 atom% Ho3+-doped ZnO electrode. Electrolyte: 1 mol dm-3 NaCl aqueous solution. J. MATER. CHEM., 1994, VOL. 4 55, -51, 5s2-51, 5F3-518 A A 500 600 700 800 wavelengthlnm Fig. 3 Electroluminescence spectra under anodic polarization of a 0.6 atom% Ho3+-doped ZnO electrode pulsed between 0.0 and 9.88 V (us. SCE) in a 1 mol dmP3 NaCl and 0.3 mol dm-, Na,S,O, aqueous electrolytic solution fresh etching creates the conditions necessary for observation of the current and the luminescence below +9.7 V. This behaviour suggests that the oxidation of an intrinsic ZnO surface state, evidenced by Kiess,12 does occur. The variations of both the emitted light and the anodic current around 5V us.SCE are too weak to be analysed correctly, but in the 9.7-10.0 V us. SCE potential range, the analysis of the variation of the emitted light intensity B as a function of the applied potential V shows that the studied electrodes follow an Alfrey-Taylor-type relationship (Fig. 4)16: B=B, exp(b/V1/2) which indicates clearly that the light emitted by the electrodes is controlled by a Mott-Schottky type potential barrier, the width of which varies with the square root of the applied potential. In previous paper^,^.^' two different mechanisms for gener- ating hot electrons at the ZnO/electrolyte interface were proposed: (i) electrons originate from the oxidation of a surface state of ZnOs O:;f-+@2 +2e~~ (3) 10 ooor 1000?-100-10: t 11 I I I I I I 0.31 0.312 0.314 0.316 0.318 0.32v-'/2 Fig.4 Alfrey-Taylor plot of the electroluminescence intensity observed at 550 nm for an unetched 0.6 atom% Ho3+-doped ZnO electrode, in a 1mol dm-, NaCl and 0.3 M Na,S,O, aqueous electrolytic solution and depending on the relative position of the 0fuTf and of the electrolyte redox potentials, either the surface of the electrode or the electrolyte is decomposed: ZnOsurf+Zn2+ +402+2e- (4) H20surf+2H++302+2e-(5) Addition of thiosulfate ions to the solution results in the replacement of eqn. (4)and (5) by 2S20:-440g-+2e-(6) and consequently prevents the degradation of the electrode. Due to the proximity of the redox potential of the S20i-/S40g-couple (Eo=0.09 V us.normal hydrogen elec- trode)18 to that of ZnO surface states8,I2 (which act as an electron source in the absence of thiosulfate ions), these surface states catalyse the electron exchange at the electrode s~l-face.'~ (ii) Electrons are originating from the valence band by a band-to-band tunnelling process." This process generates holes in the valence band of ZnO at the surface of the electrode which must induce an electrochemical dissolution of ZnO and an oxygen evolution ZnO+2h+-+Zn2+ +to2 (7) H20+2h++2H+ +to2 (8) Holes are able to react not only with the electrode surface or with water but also with any redox system present in the electrolyte. Thus with thiosulfate ions, the reaction (9) equival- ent to (6) occurs rather than (7) and (8) 2S20i- +2h+-+S4O2- (9) Once injected into the conduction band, the electrons are accelerated by the high electric field until they gain enough energy to impact-excite the rare-earth ions.The concomitant variation of both the luminescence and the anodic current agrees with the electron impact-excitation mechanism pro- posed in our previous st~dies.~-~ That only the emission characteristic of the transitions between the 4f levels of the Ho3+ ions inserted into ZnO is observed, led us to disregard the hypothesis20,21 of new emissive donor levels created by the RE3+ ions in the gap of the semiconducting ZnO; instead, a direct impact-excitation process of the RE3+ ions, induced by the high energy electrons as a result of their acceleration by the applied potential, is proposed.If the excitation of the rare earth were the result of an energy transfer from the host semiconductor to the RE3' ions, then the same results should have been obtained what- ever the excitation mode (anodic or cathodic). Therefore the I I I0.10.001 0.01 0.1 1 10 N (atom%) Fig.5 Dependence of the relative values of Vphys upon the Ho3+ doping level for different electrodes observed at 550 nm in d 1 rnol dm-, NaCl and 0.3 mol dm-3 Na2S,03 aqueous electrolytic holution Fig.6 SEMs of the ZnO samples doped with (a) 0.4,(h) 1.0, (c) 3.0 atom% Ho3+.S and B refer to the EDS spectra of Fig. 7. Note that the magnification of (a) is 1.5 x greater than (b)and (c). emission observed in both polarization modes should be identical.This is not strictly the case here, since the only emission observed is characteristic of either ZnO or Ho3+ depending upon the polarization mode of the electrode. Hence, no coupling can exist between the energy levels of the RE3+ ions and those of the semiconductor. Furthermore, in the emission spectra obtained under both polarizations, no new structure originating from a new species could be observed and all the observed luminescence bands could be attributed to transitions originating from ZnO itself or from HO~+ions. This verifies the absence of any donor level created by the RE3+ions inside the gap of ZnO. The technical efficiency qtech of an electrode2' is given by where Pelecrefers to the electrical power consumed in the J.MATER. CHEM., 1994, VOL. 4 Zn Zn Fig. 7 EDS spectra (a) recorded over the entire Fig. 6(c) micrograph; (b)recorded over the area S, of Fig. 6(c) micrograph; (c) recorded on 6(a)micrograph: solid line area B,, dotted line area S, electrode and Plumto the power emitted by the same electrode. But instead of qtech the physical yield qphys. is often used for experimental ease and for conceptual simplicity: number of photons emitted out qphys = number of charges transferred and this physical yield has been decomposed into three parts23 rphys =qexc qrad vopt where qexc is the ratio of the number of luminous centres excited over the number of charges transferred; qrad is the number of luminous centres which deactivate radiatively over the number of excited luminous centres and rIopt is the number of photons which emerge from the electrode over the number of luminous centres which do deactivate radiatively.Using the same method as that described previously for evaluating the luminescence of Er3+ (ref. 3) and Sm3+ (ref. 5) in ZnO, relative values of qphys have been measured for several +electrodes of different RE3 doping levels ranging from 3.0 x to 3.0 atom%. The results shown in Fig. 5 show a maximum efficiency for an Ho3+-doping level around 0.5 atom%. Below this value qphys varies linearly with the Ho3+-doping level. As such a dependence is expected for qexc,23'24both qrad and qOptmust remain constant. As already de~cribed,~,'qopt, which depends upon the experimental con- ditions, was found to be constant for all the observations.The invariance of qrad was verified by measuring the luminescence decay of the 5S2+518transition observed at 550nm for two different Ho3+-doping levels, i.e. 0.04 and 0.6 atom%. In both J. MATER. CHEM., 1994, VOL. 4 cases the luminescence created by a short polarization pulse decays exponentially with a rate constant equal to 1.10kO.2 x lo4SKI.Thus, the dependence of qphys upon Ho3+ is limited to the variations of qexc. Scanning Electron Microscopy and Energy Dispersive Spectroscopy The SEMs of Ho3+-doped ZnO are given in Fig. 6. They show that an increase of the Ho3+ concentration induces a decrease of the average size of the grains and an increase of the number of holes at the grain boundaries.As reported previously for Sm3+ ions,5 the presence of the rare earth prevents the growth of the grains, limits their microcrystalline structure, and induces dislocations in the crystal lattice of ZnO. But in addition a new feature appears [Fig. 6(a), 6(b) and (c)] as clear irregular spots. The EDS analysis of the samples of Fig. 6 is shown on Fig. 7. The spectrum shown in Fig. 7(a)is recorded over the entire 3 atom% sample [Fig. 6(c)] whereas that shown in Fig. 7(h)is recorded over a selected area centred around the white spot, S,, on the right side of Fig. 6(c). The difference between these two spectra arises from the intensity of the signals of Ho atoms. This intensity, higher on the spectrum of Fig.7(h)than on that of Fig. 7(a),indicates a higher Ho concentration on the area centred around the spot. The EDS analysis of the SEM shown in Fig. 6(a)is reported in Fig. 7(c). Two spectra are presented. The first is recorded over an area B,, centred on the left side of Fig. 6(a) and excluding any spot. The second is recorded over an area S, centred around the white spot near the centre of Fig. 6(a). Clearly both spectra present the same features corresponding to Zn atoms whereas only the second one shows the patterns of Ho atoms. The same results are observed on Fig. 6(b).The patterns of Ho atoms are present only on the spectrum recorded on an area including a white spot. Hence the white spots on the SEMs are attributed to rare earth aggregates, the presence of which has been postulated in previous on the basis of luminescence kinetics measurements, and already characterized in zinc oxide varistors.26 Furthermore the number of these clear spots on a given constant area increases with the Ho3+-doping level.The EDS analysis of the surroundings of the holes at the grain boundaries shows also a higher Ho concentration in the area centred around such holes. Fig. 8 reports a micro- graph of an Ho3+-doped ZnO sample and its EDS analysis. The signals arising from Ho atoms are present only on the spectrum recorded over the area H centred around the hole near the centre of the micrograph. This last result can be interpreted as the fact that the rare earth oxide acts as an inhibitor of grain growth.Conclusion The large bandgap of semiconducting zinc oxide enables the observation of the electroluminescence of the inserted hol- mium ions, as in the case of zinc sulfide. The yield of this electroluminescence reaches a maximum value when the con- centration of the dopant is ca. 0.5 atom%. The totally different behaviour of the electrode when changing the polarization mode from anodic to cathodic indicates that the energy levels of the rare earth are not coupled with those of the host. Such lack of coupling is corroborated by the SEM results which show that the incorporated rare-earth does not substitute zinc atoms, but creates islands and dislocations in the ZnO crystal lattice.This results in a progressive decrease of the semicond- ucting properties of the host. The rare-earth-doped zinc oxide has to be taken as a composite material. The electrochemical Fig. 8 (a) SEM and (b) EDS spectra of a 0.4 atom% doped ZnO sample. Solid line: spectrum recorded over the area B, centred on a ZnO monocrystal. Dotted line: spectrum recorded over the area H, centred around the hole at the grain boundaries near the centre of the micrograph. technique used to induce the electroluminescence of the rare earth needs rather low polarization voltages compared with those usually needed for all solid devices. The presence in the electrolyte solution of an electron source external 10 the semiconductor, and of a judiciously chosen redox potzntial, suppresses considerably the electrode decomposition.from Rhone-Poulenc recherches (France) for SEM and EDS measurements, and for fruitful discussions. PA -. qhinnnva and H Knhavachi <i>rinoer1 Fl~rtml~~min~cr~nr~ -..---. J..Y...,-""I.."."r..rr.r""""..cr) .,-. ',""'b.,' Verlag, Berlin, Heidelberg, 1989. 2 J. C. Ronfard-Haret, D. Kouyate and J. Kossanyi, SoliJ State Comm., 1991,79,85. 3 D. Kouyate, J. C. Ronfard-Haret and J. Kossanyi, J. Electroanal. Chem., 1991,319, 145. 4 D. Kouyate, J. C. Ronfard-Haret and J. Kossanyi, J.Luininesc., 1991,50,205. 5 D. Kouyate, J. C. Ronfard-Haret and J. Kossanyi, J. Muter. Chem., 1992,2,727. 6 J. Kossanyi, D. Kouyate, J. Pouliquen, J-C. Ronfard -Haret, P. Valat, D. Oelkrug, U.Mammel, G. P. Kelly and F. Willcinson, J. Luminesc., 1990, 46, 17. 7 D. Kouyate, J.-C. Ronfard-Haret, P. Valat, J. Kossanyi, U. Mammel and D. Oelkrug, J. Luminesc., 1990, 46,329. 8 D. Fichou and J. Kossanyi, J. Electrochem. SOC.,1986, 133, 1607. 9 P. K. Patil, J. K. Nandgave and R. D. Lawangar-Pawar, Solid State Commun., 1990,76, 571. 10 B. Pettinger, H. R. Schoeppel and H. Gerischer, Ber.Bur:senges. Phys. Chem., 1976,80,849. 144 J. MATER. CHEM., 1994, VOL. 4 11 T. Yamase and H. Gerischer, Ber. Bunsenges. Phys. Chem., 1983, 20 S. Bhushan, B. R. Kaza and A. N. Pandey, Phys. Status. Solidi. A, 87, 349. 1978,49, K167. 12 H. Kiess, J. Phys. Chem. Solids, 1970,31, 2379. 21 S. Bhushan, A. N. Pandey and B. R. Kaza, J. Luminesc., 1979, 13 E. W. Chase, R. T. Hepplewhite, D. C. Krupka and D. Kanhg, 20,29. 14 J. Appl. Phys., 1969,40,2512. R. Mach and G. 0.Mueller, Phys. Status Solidi A, 1982,69, 11. 22 23 R. Mach and G. 0.Mueller, Phys. Status Solidi A, 1984,81,609. J. W. Allen, J. Luminesc., 1984,31132, 665. 15 16 17 18 G. Boulon, Rev. Phys. Appl., 1986,21,689. G. F. Alfrey and J. B. Taylor, Proc. Phys. SOC. London, 1955, 68B, 775. B. Pettinger, H. R. Schoeppel, T. Yokoyama and H. Gerischer, Ber. Bunsenges. Phys. Chem., 1974,78,1024. Handbook of Chemistry and Physics, ed. R. C. Weast, CRC Press, 24 25 26 A. Krier and F. J. Bryant, Phys. Status Solidi 4,1984,83, 315. D. Kouyate, J. C. Ronfard-Haret and J. Kossanyi, J. Luminesc., 1993,55209. P. Williams, 0. L. Krivanek, G. Thomas and M. Yodogawa, J. Appl. Phys., 1980,51 3930. 19 Boca Raton, 1985, 65th edn., Table D155-62. H. Gerischer, Faraday Discuss., 1980,70, 137. Paper 3/03322H; Receired 9th June, 1993