STM-excited luminescence on organic materials

STM-excited luminescence on organic materials

ELSEVIER Synthetic Merals 91 ( 1997) 69-72 STM-excited luminescence on organic materials S.F. Alvarado *> L. Libioulle, P.F. Seidler IBM Research...

432KB Sizes 4 Downloads 80 Views

ELSEVIER

Synthetic

Merals

91 ( 1997)

69-72

STM-excited luminescence on organic materials S.F. Alvarado *> L. Libioulle, P.F. Seidler IBM Research

Di~~ision, Ziirich

Rescurch Ldmrntoq.

Siiwwutmssr

4, CH-8803

Riischlikor~,

S~vit:erlntui

Abstract A scanning tunneling microscope (STM) is used to generate electroluminescence from thin films of tris( 8-hydroxyquinolato) aluminum (Alq?) deposited on Au( 11 1) substrates. The emission spectra are highly dependent on the local structural features of the thin film. Furthermore, the intensity distribution is modulated by the collective charge-carrier excitations of the tip and the substrate. This is manifested as a red-shifted emission spectrum characterized by relatively narrow emission lines from the organic material. The luminescence signal appears to be a linear superposition of IWO different kinds of Alq,. We show also that the energy of the lowest unoccupied molecular orbital (LUMO) can be determined with the aid of measurements of the threshold for tunneling-excited electroluminescence. For Alq, we find E,,,, = 1.57 * 0.15 eV above the Fermi level. 0 1997 Published by Elsevier Science S.A. K~JNYI~~S:

Luminescence:

Scanning

tunnelling

microscopy

1. Introduction Scanning tunneling microscopy techniquesare employed to study the structural and spectroscopicproperties of thin organic films with nanometer resolution. Electroluminescence is excited via injection of electrons tunneling from the tip of the scanning

tunneling

microscope

(STM)

into the

organic material. This allows us to characterize the local electronic propertiesof the thin film and to attempt to understandtheir relationshipwith structural featuresobservedconcomitantly by standard STM imaging. The STM-excited luminescence(STLj technique allows us to simulate the operation of an organic light-emitting diode (OLED). in which the cathode is a metal-vacuum-organic material tunneling junction. Therefore, it is possiblewith this technique to mapthe electroluminescenceefficiency of a device-equivalent structure at nanometer dimensions.Experiments performed on thin tris( 8-hydroxyquinolato)aluminum (Alq,) films of thickness ranging from about 2 to about 4 nm are presentedhere.

2. Experimental Our experimentswere performed using an STM mounted inside an ultrahigh vacuum apparatus(base pressurein the lower lo-‘” mbar range). Alq, films were grown in situ on * Corresponding 0379-6779/97/$17.00 PllSO379-6779(

author 0 1997 Published 97 )03977-5

by Elsevier

Au( 111) substratesat room temperature.The substrates,prepared in a separatevacuum chamberby evaporating Au onto mica, were cleaned in situ by Ne-ion bombardment and recrystallized by heating, prior to organic layer deposition. The Alq, effusion cell was calibrated using a quartz balance. The thickness and homogeneity of the thin films were checked with the STM as discussedbelow. The following STL-based measurementswere performed: (i) spatially resolvedluminescencemaps,for which the wavelength-integrated luminescenceintensity is recordedwhile scanningthe tip acrossthe sample; (ii) wavelength-dispersedluminescence spectroscopy; (iii) wavelength-integrated luminescence intensity versus tunneling voltage (Z, versus VT) measurements. The latter techniqueis a kind of spectroscopy allowing, for example, the determination of the threshold energy for light emission.In all casesthe luminescencesignal is detectedfrom the free surfaceof the sampleat apolar angle of approximately 35” with respectto the plane of the surface with the substrateat room temperature.Both STM and STL measurementswere performed in the constant-current tunneling modeat currents in the rangeof 5 to 800 pA (negative tip bias) usingpolycrystalline tips madeof W, Ir, or PtIr. The highest current density flowing through the organic layer at the tunneling spot (the size of which is in the 1-nmrange) is estimatedto be of the order of 10’ A/cm”. Similarly high current densitieshave beendemonstratedin tunneling experiments on poly( phenylene vinylene) (PPV) [ 1] and on a solublePPV derivative [ 21. Further experimental detailsare given in [ 11.

Science S.A. All rights reserved

3. Results Topographic images collected by the STM reveal that the Alq, completely wets the Au surface. The thickness of the polymer film was checked by imaging its surface, and then reducing the bias voltage to V, = 0.1 V. well below the energy of the lowest unoccupied molecular orbital (LUMO), so that the STM tip penetrated the film and imaged the Au( 111) substrate instead. We find that the tip begins to penetrate the organic layer at tunneling voltages below a threshold value of vtll,, = 1.4 & 0.1 V, which, as shown below, is approximately the position of the LUMO level. Although rough Alq, regions can be found, we also observe domains in which the underlying atomically flat terraces, single-atom steps, and in some cases even the surface reconstruction of the Au( 111) surface is mimicked on the surface of the organic layer. This indicates that, locally, the thickness of tile thin film is quite uniform. We find that the first Alq, layer exhibits an apparent thickness of about 0.46 nm. This value may at first seemlow, particularly when one considers that the distance from the central Al atom to the edge of the hydroxyquinolate ligand in a free Alq, molecule is already approximately 0.6 nm. The molecule could however assume a variety of orientations with respect to the metal surface, and interaction with the surface may modify the shape of the molecule [3]. Furthermore, there might also be a contribution owing to electronic effects, i.e., the difference in conductance of the Au substrate and the organic material. Above this first layer the molecules can form smooth layers with a characteristic thickness of 0.52 * 0.01 nm. Fig. 1 shows an example of the surface stiucture of a region a few molecular layers thick. Fig. 1(a) is a region of the film that exhibits very smooth, flat domains with a r.m.s. roughness of 0.11 nm, and Fig. 1(b j shows a region in which the surface roughness is higher by a factor of2.5. The thin films appear to be amorphous, but in some cases the smoother domains exhibit parallelepiped-like features about 10 nm wide and 100 nm long. Nevertheless, no clear evidence of inplane crystalline order at the molecular level has been found in the flat regions of our samples.This observation is consistent with the fact that it is easy to make flat amorphousfilms, which is important for fabrication of efficient devices. The luminescenceintensity distribution is quite homogeneousfor the smootherregions and lessso for the rougher regionsof a thin film. Preliminary results indicate that luminescence intensity tendsto decreasewith film thickness,and measurementson even thicker films arecurrently being performed. Luminescence excited by electrons tunnel-injected from the STM tip into the organic material yields highly locationdependentspectra.The spectraare red-shiftedwith respectto the emissionfrom Alq, in OLED devices, which normally peaks at about 2.3 eV [4]? and relatively narrow emission lines are observed.As an example, Fig. 2(a) showsspectra collected for a tip biasof - 3.0 V and a tunneling current of 200 pA. In very rough regions even broader spectracan be found. Film regionscharacterizedby smoothsurfaces exhibit

Fig. 1. STM topographs of an Alq,-coated Au{ 111) surface showing (a) domains of smooth terraces mimicking the underlying substrate structural features, and (b) a region of the surface exhibiting a higher degree of roughness.

spectra with a dominant peak at Izv = 1.8 + 0.03 eV, and a weakerpeak at h Y= I .98 i 0.03 eV, the intensity of the latter transition increasingwith film thickness.Actually the spectra appearto be a linear superpositionof spectralfeaturesarising from two different kinds of Alq,. Both peakshave a linewidth of about 100 meV (FWHM). Additional features of the spectra are a weaker. broader (about 160 meV) feature hv = 1.55+ 0.06 eV and a weak transition at Izv = 2.1 eV. We note that the energy of all the transitions is nearly invariant with tunnel voltage in the range 2.5 < Y,-4 6 V and a maximum in total luminescenceintensity occurs for 3 I VTI4 V dependingon tip and measuringconditions.As thismaximum occurs within the tunneling voltage range observed for different metallic surfaces[ 51, we can deducethat the collective

F s 8 ‘5: .z s E .-

1 .a 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 1.2

1.4

1.6 1.8 2.0

2.2

2.4

2.6 2.8

3.0

3.2

1.2 60

1.5

1.8 tunnel

z-

E 50 v)

2.1 voltage

2.4

2.7

3.0

(V)

Fig. 3. Integrated luminescence yield vs. tunneling voltage measurement used to estimate the threshold of luminescence of a thin film of Alq, on Au(ll1).

.=,” 30 5 20 .E IO 6

400 I 1.2 1.4

1.6

1.8 2.0

2.2 2.4

2.6

2.8

3.0

3.2

photon energy (eV) Fig. 2. (a) Spectra of STM-induced luminescence from a region of Lmooth terraces I wlid circle5 ) and from a roughtx region (solid line). ( b 1 Fluorescence spectrum from the clean Au( I I 1 ) subaxts excited with ;i PtIr tip. Tunnel parameters: I’, = 2.5 V i nsgntive tip hias) and I= 100 pA.

charge-carrier excitations ( plasmons) [S] involving the tip and the subxtrate play an important role in the excitation of light emissionin our samples. The apparent red shift of luminescencefrom the organic material appearsto be related to the intrinsic fluorescenceof the tip-Au( 1I 1) tunnelingjunction. Fig. 2(b) showsthe fluorescencespectrum of the clean ALI( I II ) surface excited with a pristine PtIr tip. Owing to the enhancedradiative decay of collective excitations involving the tip and the metallic surface, the maximum intensity of the emission occurs at 2.5 I \‘,I 3.5 V, which is in relatively good agreementwith the resultsin [S]. The shapeof the emissionband doesnot changesignificantly over the range 2.5 I V, I 3.5 V. A comparison of the spectrain Fig. 2(a) and (b) suggests that the spectrumof the clean Au( 1111 surfacedefines the spectral rangeof observableluminescencefor thin Alq, layers within the tunneling junction. We find that the maximum emission intensity of the clean Au( 111) surface is at least a factor of two to threehigher than that of the Alq,-covered sul-face.This is surprising becausewe expect relatively intenseemission from Alq,. Quenching of the Alq, emissioncausedby the proximity of the metal surfacesmay thus play a role. Fig. 3 showsan I, versus V, spectrum.Averaging several measurements,we deducethat the thresholdfor STM-excited luminescenceof Alq, thin films is V,,,.,- = 1.5k 0.1 eV. This measurementallows us to estimate the energy of the first LUMO level. Taking into account the broadeningof the elec-

tronic transitionsfound in the wavelength-resolvedluminescence spectra, namely, 100< AE < 160 meV (see Fig. 2), we estimate the energy of the LUMO to be at ELUMo= 1.57i 0.15 meV relative to the Fermi level. This result suggeststhat the STL at hv = 1.6 eV is related to a transition involving the LUMO level of Alq, andthe Fermi level of the Au( 1I 1) substrate.

4. Conclusions The STM imagesreveal that thin films of Alq, deposited on Au( 111) substratesexhibit domainsof different structural order. STM-excited luminescencemeasurementshave been usedto study the local electronic excitations of the thin Alq, layers. We find that the spectralfeaturesof the luminescence depend on the structural details of the film, i.e., thickness, uniformity. androughness.The measurements alsoshowthat, owing to the proximity of the metallic surface,the emission spectrumof the organic moleculesis red-shiftedandthe luminescenceintensity appearsto be strongly coupled to the collective charge-canier excitations of the tip-metal substrate tunneling junction. The maximum intensity of the STL from the organic moleculesoccursat approximately the sametunneling voltage asfor the clean metal surfaces.We have also shown how STL techniquescan be usedto estimatethe position of the LUMO of an organic material.

Acknowledgements The authorsthank E. Delamarchefor making the Au( 111) substratesusedin theseexperiments.Thanks are also due to A. Curioni and W. Andreoni for many interestingandenlightening discussions,and to F.K. Reinhart for lending us very useful equipment.

72

S.F. Ai~umdn

et 01. /Synthetic

References

[ I] S.F. Alvarado, W. Riel3. P.F. &idler and P. Strohriegl, Phys. Rev. B, in press. [2] D.G. Lidzey, D.D.C. Bradley, S.F. Alvarado and P.F. &idler, Nature, 386 (1997) 135.

Metals

91 (1997)

69-72

[3] Ab initio calculations of Aly, interacting with merallic surfaces as well as of the electronic transitions are currently underway: A. Curioni and W. Andreoni, to be published. [4] See, for instance: P.E. Burrows. Z. Shen, V. Bulovic, D.M. McCarty, S.R. Forrest, J.A. CroninandM.E.Thompson, J. Appl. Phys.,79 (1996) 7991 and Refs. therein. [S] R. Berndt and J.K. Gimzewski, Ann. Phys., 2 (1993) 133 and Refs. therein.