Study Guides, Notes, & Quizzes

A study of the degradation of poly(3-octylthiophene)-based light emitting diodes by Surface Enhanced Raman Scattering

A study of the degradation of poly(3-octylthiophene)-based light emitting diodes by Surface Enhanced Raman Scattering
of 7
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  DOI:10.1007/s00340-004-1592-5Appl. Phys.B 79, 603–609(2004) Lasers and Optics AppliedPhysicsB e.giorgetti 1, ✉ g.margheri 2 t.delrosso 2 s.sottini 1 m.muniz-miranda 3 m.innocenti 3 A study of the degradationof poly(3-octylthiophene)-based lightemitting diodes by Surface EnhancedRaman Scattering 1 INSTM e Istituto di Fisica Applicata “Nello Carrara”– CNR, Via Panciatichi 64, 50127 Firenze, Italy 2 Istituto di Fisica Applicata “Nello Carrara”– CNR, Via Panciatichi 64, 50127 Firenze, Italy 3 Dipartimento di Chimica, Università di Firenze, Via della Lastruccia 3, 50019, Sesto Fiorentino, Italy Received: 5 April 2004 / Revisedversion: 7 June 2004Published online: 11 August2004• © Springer-Verlag 2004 ABSTRACT  Organic light emitting diodes (OLED) based onpoly(3-octylthiophene) (P3OT) were studied by on-line SurfaceEnhanced Raman Scattering (SERS) experiments. A simultan-eous reduction of the emitted power and an increase of theSERS signal was observed above an 8V bias voltage in thecase of ITO / P3OT / Al structures. On-line measurements of thetemperature and independent tests with an atomic force micro-scope, performed on the Al cathode before and after currentflow, demonstrated that the observed enhancement of the Ra-man signal was partly thermal and partly electromagnetic insrcin, with clear evidence of cathode roughening induced bycurrent flow. In contrast, analogous tests performed on P3OT-based OLEDs containing a composite Al + Ag cathode showedthat the breakdown was related to electrode delamination. PACS  78.60Fi; 78.30Jw; 85.60Jb 1 Introduction Light-emitting diodes based on organic electro-luminescent materials (OLEDs) have been actively studiedsince the late 1980s [1]. These devices offer big advan-tages with respect to their inorganic counterparts in termsof processability, cost reduction, color tunability and lowvoltage operation. Moreover, the development of OLEDsbasedonpolymericcompounds,instead of smallluminescentmolecules, is expected to open the way towards the construc-tion of flexible and inexpensive roll-up displays. Althoughvery significant results have been obtained in the past tenyears, the behaviour of polymer LEDs still requires in-depthinvestigation. In particular, new electroluminescent materialsare currently being tested and the different mechanisms thatlimit device lifetime, both under storage and operating condi-tions,arewidely investigatedin theliterature.Up to now, many degradation and failure mechanismshave been identified [1–10]. In the first place, the interactionwith water and oxygen produces a quenching of the emis-sion and eventually the failure of devices. In order to impairthis effect, experiments on OLEDs are currently performedundervacuumorinertatmosphere. However,photo-oxidative ✉  Fax: +39-055-410-893, E-mail: degradation can still take place, due to oxygen diffusion fromindium-tin-oxide (ITO) anode into the active layer [1]. Solid-statemorphologyalsoplaysakeyroleindetermining thelife-time of a device [2–4]. Several authors observed that degra-dation andsubsequentfailureofOLEDswasaccompanied bythe formation of black spots, that were associated with localshort-circuits or morphological, physical and chemical mod-ifications of device structure. In some cases, black spots ledto delamination of the structure and final detachment of theelectrodes.Adetailedanalysisofthemechanismofblackspotgrowth and of their nature is reported in the literature in bothcasesofpolymer[5,6]andsmallmolecule[7–10]basedelec-troluminescent devices.In order to get a deeper understanding of the srcinand evolution of the degradation processes, OLEDs havebeen investigated by using a large number of different tech-niques, such as microscopy (scanning and transmission elec-tron microscopy, atomic force microscopy-AFM, scanningtunnelling microscopy, optical microscopy) [2–10], Augerelectron spectroscopy [6,10], secondary ion mass spectrom-etry [5], UV, and X-ray photoelectron spectroscopy in ultra-high vacuum [11], in situ infrared spectroscopy [12,13], andmicroRaman spectroscopy performed ondamaged areas afterbreakdown [14]. To the authors’ knowledge, there are no re-ports in the literature on the use of Surface Enhanced RamanScattering (SERS). SERS spectroscopy is extensively used toinvestigate the properties of molecules adsorbed on metals. Itis well known that molecules, which are in close vicinity toproperly roughened films or nanoparticles of silver, gold or,to a lesser extent, aluminum ( currently used as the cathodein OLEDs), can undergo huge magnification of the Ramanscattering that, in the case of silver, may be of the order of  ≈ 10 6 ÷ 10 14 , and is dramatically dependent on the morph-ology of the enhancing metal surface [15]. In fact, when theligand interacts with the metal surface, the Raman intensi-ties undergo changes on the basis of the surface selectionrules, which depend both on electromagnetic and chemicalmechanisms, by enhancement of the local electric field at themetal surface and of the dynamic polarizability of the ligand,respectively. Moreover, the spectral positions of the SERSbands can present frequency-shifts with respect to the Ra-man spectrum of the bulk, by changes of the force constants.As a consequence, SERS appears to be a convenient tool forthe characterization of the organic / metal interface in OLED  604 Applied Physics B – Lasers and Optics devices. In particular, during OLED operation, enhancementfactor, relative intensities and frequency-shifts of the Ramanbandscan exhibit variations associated with currentflow.TheaimofthispaperistoproposeSERSasatechniqueforon-linemonitoringofOLEDsand,inparticular,forefficientdetectionof several effects, such as heating, morphological changes atthe organic film / cathode interface, formation of new chem-ical species or of chemical bonds between metal layer andactive material [16], or simply degradation of the emittingmolecules.Inordertoassessthepotential ofSERSforOLEDcharac-terization, we performed some experiments on simple struc-tures, based on a commercial and well-characterized activematerial: poly(3-octylthiophene) (P3OT) [17,18]. Section 2contains the details of OLED preparation and electric char-acterization, while Sect. 3 is devoted to the description of theRamanexperimentsperformedwithP3OTindifferentphases.SomeAFMtests,reportedin Sect. 3,wereperformedto inde-pendently check theSERSresults. 2 OLED preparation All the devices used in this study contained a sin-gle active layer made of regioregular commercial P3OT pur-chased from Aldrich. Indium-tin-oxide (ITO) was used as theanode,whilethecathodewaseitherasingle Al layeroramix-ture of   Al  and  Ag . All the fabrication steps were performedunderclean roomconditions.ITO coated substrates were purchased by Thin Film De-vices. The ITO coating had nominal thickness and resistivityof   140nm  and  10 − 3 Ω cm , respectively. Before deposition of the P3OT layer, ITO surfaces were patterned by photoresist-based lithography. Each sample contained four active regionsand the typical dimensions of emitting areas were  4 × 5mm 2 .After the patterning, ITO coated substrates were washed intrichloroethylene, acetone,  5%  Decon mixture and finallyrinsed in deionized water. All these operations were per-formedin an ultrasonicbath.P3OT was dissolved in chlorobenzene and the films wereobtained by spin coating of   10g / l  solutions filtered with 0 . 45 µ m membranes.Thethicknessofthepolymerlayerswasmeasured with a stylus profilometer and was, typically, of theorder of 50 to  80nm . Figure 1 shows the electronic spectraof P3OT films obtained from chloroform and chlorobenzenesolutions. We preferred chlorobenzene to the currently usedchloroform [19], because it permits higher and faster solubil-ity and limits the recrystallization of films after the spin coat-ing, which can be a severe problem for VP3OT [20]. We alsoobserved that the use of chlorobenzene improves the qual-ity of the deposition, particularly when it is performed afteramoderate heating of substrateand solution, aswell.Immediately after thespin coating of P3OT,wefabricatedtop electrodes by electron gun evaporation. In general, weused aluminum electrodes with thickness of about  200nm .However, we also prepared some samples containing a two-layer cathode madeof  Ag and  Al . First,a 17nm thick layer of  Ag  was thermally evaporated onto the polymer at a slow rate( ≈ 0 . 2Å / sec ). The slow rate was aimed at producing a layerconsistingmainly ofsilverislandsandvoids.Then, a 180nm -thick  Al  layer was thermally evaporated onto silver at a rate FIGURE 1  Electronic spectra of P3OT films obtained from chlorobenzeneand chloroform solutions and spun on commercial microscope glass slides of  ≈ 10Å / sec . This double deposition was chosen not onlyto enhance the SERS signal, due to the presence of silvernanoclusters, but also to simulate possible unwanted effectsin those devices containing composite cathodes based on ex-tremely thin layers of metals with a low work-function, suchas  Ca  and Mg ,and athick coverlayer of  Ag [1].All the devices were stored and tested under a vacuumof  ≈ 10 − 2 Torr . Figure 2 shows a typical sample positionedwithin the vacuum chamber. When biased above ≈ 4V , thedevice emits orangelightand isclearly visibleunderstandardroomillumination conditions.Current–voltage(  I  – V  )characteristics oftheOLEDswererecorded in constantvoltage mode, by usinga DCpower sup-plier. Luminescence–voltage characteristics were measuredwith a calibrated silicon photodiode. The power efficiency of thedevices, defined as η = P tot  IV  × 100 ,  (1)was obtained from themeasured optical power  P  by calculat-ing the total emitted power  P tot  in the hypothesis of a Lam-bertian point source and by taking into account the distance FIGURE 2  One of our samples positioned inside the vacuum chamber usedfor all the experiments. The  arrow  indicates one of the four active areas  GIORGETTI et al. A study of the degradation of P3OT-based OLEDs by SERS 605 between emitting area and window of the vacuum chamberandthereflectivity ofthewindow.Figure 3 shows the  I  – V   characteristics of four of ourITO / P3OT / Al  samples and, in one case (crosses), of anITO / P3OT / Ag + Al  sample. The optical power-voltage char-acteristic of one of the ITO / P3OT / Al  samples is also re-ported in the inset of Fig. 3. The corresponding efficiencyevaluated in the peak of the curve was  η = 4 × 10 − 4 V  % . Asshown in Fig. 3,  I  – V   characteristics of ITO / P3OT / Al  sam-ples exhibit a considerable repeatability and diode operationwas very stable in terms of current flow and emitted powerfor voltages up to  8V , corresponding to a dissipated elec-trical power of the order of   300mW . However, the emittedoptical power decayed abruptly above  8 V  (see the inset of Fig. 3b) and some current instability was observed. Devicebreakdown occurred at bias voltages of around  15V , corres-ponding to a dissipated power of  ≈ 2 W . Typical breakdownvaluesforITO / P3OT / Ag + Al sampleswerelower( 10V biasand  600mW  dissipated power), apparently due to the pres-ence of   Ag  that, although advantageous for SERS experi-ments, quenches the efficiency of the diodes and anticipatesthebreakdown[20].Ingeneral,wedidnotobtainanyevidenceoftheformationof black spots, even under on-line optical microscope inspec-tion (up to 40X magnification). Only in the case of one agedITO / P3OT / Al  sample, did we observe black spots and bub- FIGURE 4  Operation of an agedITO / P3OT / Al device.  a  7V bias aftertens of ON / OFF cycles: the activearea is partially illuminated and thereis no evidence of black spots;  b  7 Vbias at the beginning of the last cyclebefore breakdown: the active area ispartially illuminated and three blackspots are well visible in the brightestregion;  c  8 V bias: the emission in-tensity of the initially bright area (A)has grown, the initially dark area (B)has started to emit and the black spotshave diffused throughout the emissiveregion;  d  after several minutes opera-tion at 9 . 5 V bias: some sparks start toappear at the edge of region A;  e  afterseveral minutes operation at 13 . 5 Vbias: area A has switched OFF, whilearea B is still emitting;  f   microscopeview of the bubble indicated by thearrow in  d , taken with 200X magnifi-cation FIGURE 3  I  – V   characteristics of four ITO / P3OT / Al devices and of oneITO / P3OT / Ag + Al device ( crosses ). The  inset   shows optical power-voltagecharacteristics of one ITO / P3OT / Al sample bles. After a three-month-long storage time under  10 − 2 Torr vacuum, the sample was operated for several hours with re-peated ON / OFF cycles. Its behavior, both in terms of   I  – V  characteristics and of surface illumination, was the same asthat observed for fresh samples. However, during the operat-ing cycle described by the sequence of pictures of Fig. 4a–e,  606 Applied Physics B – Lasers and Optics weobserved the formation of several black spots. Eventually,after some more minutes of operation at  13 . 5V  (Fig. 4e), theOLED broke down and bubbles were clearly visible to thenaked eye. Their position corresponded to the black spots ob-servedduringOLEDemission. Figure4f showsamicroscopepictureofthebubbleindicated by thearrowinFig. 4d.Apart from the presence of black spots and bubbles, thesequence of Fig. 4a–e also represents the typical evolution of ITO / P3OT / Al  samples, in terms of surface illumination anduniformity of the emission. In particular, when OLEDs areswitched ON, some macroscopic regions of the surface mayappear dark (Fig. 4a,b). The morphology of bright and darkareas does not change with repeated switching ON and OFFof the devices and seems to be related with the uniformityof the active layers. Indeed, it is well known that OLED effi-ciency is strongly dependent on the value of the electric field(more than bias voltage) and, consequently, on the thicknessof the active layer [21]. Above  8V , when device instabili-ties and power decay begin, the initially bright areas (markedwith A in Fig. 4c) gradually loose permanently their perma-nent efficiency, while the initially dark areas (marked with Bin Fig. 4c) switch ON (Fig. 4e). 3 Raman experiments Raman spectra were recorded using the 514.5 and 457 . 9nm  lines of an  Ar + laser or the 647.1 and the  406 . 7nm lines of a  Kr + laser. Samples were irradiated by using  40 ÷ 150mW laserpowerandthelaserbeamwasdefocussedtoim-pair thermal effects or quenching of the Raman signal. Powerdensity measurements were performed with a power meter(model 362; Scientech) and gave an accuracy of   ∼ 5%  inthe  300 – 1000nm  spectral range. The scattered light was col-lected at  90 ◦ with respect to the plane of the samples and de-tected byaJobin-YvonHG2Smonochromator equipped witha cooled RCA-C31034A photomultiplier and a data acquisi-tion facility. Spectra were usually registered on-line, duringOLED operation, and at different bias voltages. The sampleswerecontinuouslykeptunder 10 − 2 Torr vacuum,bymeansof arotative pump(Fig. 2).Due to the well-known SERS efficiency of silver nano-clusters,weperformed the firstRaman tests with OLED sam-ples having  Ag + Al  cathodes. We started with the  514 . 5nm exciting linethat, asshownbytheelectronic spectraof Fig. 1,inducesaresonantRaman effectonP3OT.Figure5showstheRaman spectra of an ITO / P3OT / Al + Ag  sample, recordedat different bias voltages and, for comparison, the spectrumof a powder sample of P3OT. In this case, due to the reson-ance, the Raman spectra were mainly produced by the bulkresponse of the polymeric layer and the SERS effect at thepolymer / metal interface was masked. No appreciable differ-encewasobservedamongspectraregistered beforeswitchingON ( V   = 0 )  and during OLED operation, apart from an in-creaseofthebackgroundsignal,duetotheenhancedlumines-cenceofthedevice, associated withcurrentflow.Figure 6 shows the Raman spectra of the same device of Fig. 5 and of a powder sample, registered with the  406 . 7nm exciting line. In this case, the resonant contribution was con-siderablyreduced (Fig. 1) andthe Raman signaloftheOLEDwasmainly associatedwithaSERSeffect.Allthemainbands FIGURE 5  Raman spectra of an ITO / P3OT / Ag + Al sample recorded atdifferent bias voltages. The spectrum of a P3OT powder sample is alsoreported for comparison. Exciting wavelength  λ = 514 . 5 nm, laser power40 mW FIGURE 6  SERS spectra of the same device of Fig. 5 recorded at differentbias voltages. The spectrum of a P3OT powder sample is also reported forcomparison. Exciting wavelength  λ = 406 . 7 nm, laser power 40mW occurring at  I   = 0  are attributed to the vibrations of the thio-phene ring [22,23]. In particular, the bands at  1525cm − 1 and 1472cm − 1 are associated with antisymmetric and symmet-ric  C = C  stretching , respectively. The band at  1380cm − 1 isassociated with the  C − C  ring stretching, while the band at 726cm − 1 is associated with  C − S − C  bending. In general,theband at 1472cm − 1 is knownastheamplitude mode.The most intense band of the spectrum of Fig. 6( 1472cm − 1 ) was considerably shifted  ( 1455 → 1472cm − 1 ) with respect to the spectra of Fig. 5. A minor shift was alsodetected for the band at  1525cm − 1 ( 1520 → 1525cm − 1 ) .A similar Raman dispersion, which at a first glance couldbe due to a chemical adsorption of the polymer on themetal layer, was already described for the case of poly(3-decylthiophene) [24] and for P3OT films on dielectric sub-strates[22].Thedispersionobservedin[22]regardedonlythe C = C  symmetric ring stretching band and was attributed tothecoexistenceofpolymericchainshavingdifferentconjuga-tion lengths and, hence, different Raman resonances. Indeed,by using a standard deconvolution procedure (Fig. 7), we  GIORGETTI et al. A study of the degradation of P3OT-based OLEDs by SERS 607 FIGURE 7  Deconvolution of the C = C symmetric stretching bands of Figs. 5, 6 and IR spectrum of P3OT in KBr pellets found that the amplitude mode of Fig. 6 exhibits a shoul-der at  1455cm − 1 , corresponding to the band observed withthe green exciting line. At the same time, the band observedat  1455cm − 1 in the spectra of Fig. 5 exhibits a shoulder at 1475cm − 1 . To clarify this point, we examined the IR spec-trumofP3OTina KBr pellet. Thisspectrum,shownin Fig. 7,permits observation of the  C = C  stretching in absence of the Raman resonance, and consists of a wide band from 1440cm − 1 to  1470cm − 1 , with a maximum at  1455cm − 1 .Moreover, we also observed that the dispersion of the Ramanbands associated with  C = C  stretching not only occurs inthe spectra of P3OT powders and films, but also in samplesconsisting of P3OT in  Ag  colloids. For this purpose, stablesilversolswerepreparedbyreductionof  AgNO 3 (Aldrich,pu-rity  99 . 998% ) with excess  NaBH 4  (Aldrich, purity  99% ) [25]and aged a week to prevent the formation of reduction prod-ucts [26]. The typical pH value of the aqueous suspensionwas ≈ 9 . SERS spectra of P3OT in  Ag  hydrosol are shown inFig. 8: also in this case, Raman bands associated with  C = C stretching move to higher frequencies with decreasing excit-ing wavelength. In light of all these observations, we could FIGURE 8  SERS spectra of P3OT suspended in Ag colloids. Excitingwavelengths  λ = : 406.7, 457.9 and 514 . 5 nm definitely exclude the chemisorption of P3OTon the  Ag + Al layer and attribute the SERS signal of Fig. 6 to a purely elec-tromagnetic effect and the dispersion of the Raman signal totheexistenceofawidedistributionofconjugationlengthsthatresonatewith differentexciting laserlines.Figure6also showsSERS spectrarecorded during OLEDoperation. At first, no significant changes were observed be-tween0and 5V bias(spectraregisteredbeforeswitchONandat  I   = 20mA  were identical). However, after several minutesoperation at  5V  (corresponding to a dissipated power of only 100mW ) and some ON / OFF cycles, the OLED appeared tobeseriouslydamaged. Thesamebiasvoltageof  5V produceda current flow of only  1mA , although the SERS spectrum ap-peared unchanged. An attempt to increase the current flow bybiasing the device up to  10V  produced a fast decrease of cur-rent flow(down to  0 . 07mA ) and a sudden decay of the SERSsignal, with nochangeinthespectralposition ofthemain Ra-man bands. This behaviour is consistent with adetachment of the activelayer fromthecathode, probably dueto ahugefieldintensification and consequent local increase of temperatureat silversites.SERS tests at  406 . 7nm  were repeated with  Ag -free sam-ples. In this case, a reduced Raman signal was expected, dueto the lower SERS activity of   Al . However, the absence of  Ag  in the OLED structure improved both efficiency and sta-bility of the devices and moved breakdown threshold towardshigher levels of bias voltage. The results of Raman tests onaITO / P3OT / Al samplearereportedinFig. 9.Inordertobet-ter identify the positions of the bands, all spectra have under-gone the same smoothing procedure by the Savitsky–Golaymethod. In this case, the SERS effect could not be attributedentirely to an electromagnetic mechanism, because the ini-tial( V   = 0 ) positionoftheamplitudemodewasdown-shiftedwith respect to that of P3OT powders (see Figs. 6 and 7).Thisprovidesindicationofaweakinteraction between Al andpolymerattheinterface.SERSspectraofFig. 9wererecordedat differentbias voltages. Formoderate bias voltage (between0 and  8V ), OLED operation was stable, although with non-uniformsurfaceillumination(Fig. 4).Noappreciablechangesof SERS spectra were observed within this voltage range,apartfromaminorgrowthofthebackgroundsignalduetothe FIGURE 9  SERS spectra of an ITO / P3OT / Al device recorded at differentbias voltages. Exciting wavelength  λ = 406 . 7 nm, laser power 150 mW
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks