Article pubs.acs.org/JPCC
Doping Mechanism and Electronic Structure of Alkali Metal Doped Tris(8-hydroxyquinoline) Aluminum Kisoo Kim, Kihyon Hong, Sungjun Kim, and Jong-Lam Lee* Division of Advanced Materials Science and Department of Materials Science and Engineering, Pohang University of Science and Technology (POSTECH), Pohang, Kyungbuk 789-784, Korea ABSTRACT: We investigated the electronic structures of alkali metal doped Alq3 molecules prepared by codeposition of Na and tris(8-hydroxyquinolinato)aluminum (Alq3) (Na:Alq3) using in situ synchrotron radiation photoelectron spectroscopy. The doping ratio of Na to Alq3 (RNa) was 0, 0.3, 0.6, 0.8, 2.7, 3.4, or 3.9. The work function, calculated from photoemission spectra, remained at 3.6 eV ± 0.03 eV for all samples, while the energy of the highest occupied molecular orbital increased from 2.2 to 2.85 eV as the doping ratio increased from RNa = 0 to RNa = 2.7. The work function and valence band spectra indicated that there is no band bending or surface work function change due to an alkali doping effect, in contrast to the findings of previous reports. The doped layer was composed of an n-type organometallic complex according to the analysis of O 1s and N 1s spectra. The N-type doping effects shown in the N 1s spectra of coevaporation samples were reflected in the schematic band diagram, so the energy difference between the Fermi level (EF) and the lowest unoccupied molecular orbital (LUMO) decreased by 0.64 eV. The schematic band diagram demonstrates that a monotonic shift of the LUMO toward EF was observed with increasing doping, which is in contrast to general n-type doping effects in inorganic semiconductors. Also, we experimentally observed increased electron transport characteristics of alkali metal doped Alq3. The operating voltage at 100 mA/cm2 decreased from 11.9 V (RNa = 0) to 9.6 V, and the luminance at 10.5 V increased from 3575 cd/m2 (RNa = 0) to 9675 cd/m2 when the Na:Alq3 film (RNa = 2.7) was inserted between Alq3 and LiF/Al.
1. INTRODUCTION Metal doped organic materials have attracted extensive research interest due to their unique magnetic,1,2 electric,3−6 and physical properties,7 which are unobtainable in pure synthetic organic materials, as well as their potential applications. For example, our previous research has shown that a cobalt doped Alq3 layer designed for use in spintronic devices displays roomtemperature ferromagnetism due to bound magnetic polarons.1 Alkali-doped fullerides have been observed to be superconducting materials and to transition between a metallic and an insulating state.3−6 Lithium doping in metal−organic framework materials designed for gas storage enhances gas uptake by using the redox-active ligands as struts.7 Among metal doped organics, alkali metal (or alkali earth metal) doped Alq3 systems, like that of the fullerides, are some of the most intensively studied, not only as conducting films, but also as a representative model to analyze the evolution of electronic structure through interaction between alkali metals and organic materials.8−18 Alkali doped Alq3 has been studied generally as an electron injection layer or an electron transport layer for OLEDs.11−13 Recently, alkali doped layers have been used as essential connecting units to generate charges for tandem OLEDs.14 Interest in the fundamental properties of metal−organic reactions is therefore considerable. Li, Mg, K, Ca, Cs, Sm, and their derivatives are analyzed as doping materials for Alq3, and two kinds of results are reported to explain the effect of alkali metal doping in most studies.8−11 First, the appearance of new components in N 1s, O 1s, and VB © 2012 American Chemical Society
spectra represent the formation of alkali metal−organic complexes. Second, the shift of all energy levels induced by alkali metal doping reflects the band bending. However, there has been no clear explanation as to why the band bending occurs and why the results for the evolution of band bending in alkali metal doped organic materials have so far been in discord with each other. According to DFT calculations, a reaction between an alkali metal and Alq3 does not bring about a core-level shift as shown in the case of C atoms.16a,b The DFT calculation is carried out under the premise that the evolution and band bending are unrelated.16 This means that the vacuum level shifts (i.e., the band bending) have no relation to the formation of the alkali metal−Alq3 complex. In addition, the variation of the band bending with doping concentration is reported to have a semilogarithmic form,8 a certain constant value,9,10 or an approximate zero17,18 in previous experimental results. Despite numerous studies,8−18 this discordance in the results of band bending means that, until now, there has been no clear consensus even on the band diagram of alkali metal doped small molecules. Moreover, a direct relationship between the electronic structure of alkali doped Alq3 and the electrical properties of OLEDs has not been clearly reported, and an optimum doping concentration based on a quantitative analysis of the relationship has not been Received: October 26, 2011 Revised: March 17, 2012 Published: March 27, 2012 9158
dx.doi.org/10.1021/jp2102918 | J. Phys. Chem. C 2012, 116, 9158−9165
The Journal of Physical Chemistry C
Article
Figure 1. Structure of alkali metal-doped Alq3: (a) alkali metal deposition onto Alq3 (Na/Alq3); (b) coevaporation of alkali metal and Alq3(Na:Alq3). Schematic energy band structure of (c) Na/Alq3 and (d) Na:Alq3.
suggested for device applications.8−19 Thus, it is necessary to clarify the origin of band bending and to determine the band structure of alkali-doped organic materials to elucidate the nature of alkali metal doping. It has been implicitly agreed that alkali metal deposition on organic film is similar to codeposition of an alkali metal and an organic material8−14 because the alkali metal easily diffuses into the organic material.8a,b,9,10c,13,20 Accordingly, most alkali doped films8−14 have been prepared by alkali metal evaporation on top of an organic layer to analyze the electronic structure and to form alkali-doped organic devices, with the exception of a few studies.18,19 However, a recent study21 reports that the degree of alkali metal penetration and clustering (Figure 1a) on organic films could be altered by reaction between the alkali metals and organic materials. Therefore, it is desirable for accurate experiments to use a codeposition method, evaporating alkali metals and organic materials simultaneously (Figure 1b). In the present work, an Au/Na:Alq3 sample was prepared by codeposition of Na and Alq3 (Na:Alq3) on an Au film and compared with an Au/Alq3/Na sample that was prepared by the deposition of Na on an Au/Alq3 substrate (Alq3/Na). Through the comparison of these two structures using in situ measurements of synchrotron radiation photoelectron spectroscopy (SRPES) and ultraviolet photoelectron spectroscopy (UPS),22 we elucidated the origin of band bending, the band
structure, and the quantitative relationship of electronic structure and electrical properties in an alkali doped small molecular system (Figure 1).
2. EXPERIMENTAL SECTION 2.1. Photoemission Study. To investigate the energy band structure of alkali doped Alq3, Si wafers coated with a 50 nm layer of Au were loaded into a vacuum chamber with an electron analyzer in the 4B1 beamline of the Pohang light source. The electron energy analyzer was a VG scienta R3000 analyzer. An incident photon energy of 650 eV was used to obtain C 1s, O 1s, N 1s, Al 2p, and Na 2p spectra. Valence band spectra and secondary cut-offs were acquired using an ultraviolet source (Omicron; He I line, 21.2 eV). The pass energies of SRPES and UPS were 50 and 5 eV. Both SRPES and UPS spectra were monitored after in situ codeposition of Alq3 and Na on Au film (Figure 1b). The evaporation for material deposition was performed in a connected preparation chamber at a base pressure of 8 × 10−9 Torr and core level spectra were obtained in the main chamber at a base pressure of 8 × 10−10 Torr. Therefore, the results of in situ measurements using SRPES and UPS were little affected by oxidation conditions. The concentration of alkali metals in the Alq3 films was obtained by comparing the relative intensities of the alkali metal core-level and the Al 2p core-level.28,29 The doping ratio (R) of Na to Alq3 (RNa) was 0, 0.3, 0.6, 0.8, 2.7, 3.4, or 3.9. 9159
dx.doi.org/10.1021/jp2102918 | J. Phys. Chem. C 2012, 116, 9158−9165
The Journal of Physical Chemistry C
Article
Figure 2. (a) WF spectra of Na:Alq3 structure and (b) VB spectra of Na:Alq3 structure with Na concentration obtained using in situ UPS measurements. Up arrow: zenith of HOMO. Down arrow: σ bond backbone of C 2p component in Alq3 molecules.
water, then dried using high-purity N2 gas. The ITO was treated using O2 plasma (power, 150 W) for 1 min under an air pressure of 0.1 Torr. For the traditional doping method, samples were loaded into a thermal evaporator and layers of 4′bis[N-(1-naphtyl)-N-phenyl-amino]biphenyl (α-NPD) (70 nm thick) and Alq3 (60 nm thick) were deposited on them. The Na-doped Alq3 layer was prepared by sequential deposition of Alq3 (30 nm thick) and Na (4, 10, or 30 Å thick) (Alq3/Na). Finally, LiF (1 nm thick) and Al (150 nm thick) were deposited as a cathode. For codeposition, samples were loaded into a thermal evaporator and α-NPD (70 nm thick) and Alq3 (50 nm thick) were deposited. To control the Na content in the Alq3 layer, the evaporation rates of Na and Alq3 were measured separately. To deposit Na-doped Alq3 layers, Na and Alq3 were coevaporated from separate sources and deposited to a total thickness of 10 nm. RNa was 0, 0.7, 1.5, 2.7, or 4.0. Finally, LiF (1 nm thick) and Al (150 nm thick) were deposited (Figure 1d). During deposition, the base pressure of the chamber was maintained as low as 1 × 10−6 Torr. The active area of the device was 2.5 mm × 2.5 mm.
To examine the formation of the interface between the alkali metal and Alq3, Na layers of thickness of 2, 10, or 30 Å were deposited on Alq3 (Figure 1a). As an Na source, we used 99.99% metallic Na. We confirmed that the deposited Na was not contaminated because the intensity of the O 1s spectra was equal to the background level after adequate pre-evaporation in preparatory experiments. The thickness of the Alq3 film was 100 Å to prevent charging during photoemission. Fluxes of thermally evaporated Alq3 molecules were fixed at 3 Å/s by using a thickness monitor, which was calibrated by using a surface profilometer (Alpha-step, Tencor Instruments). The thickness of Na was checked by using the surface profilometer with a 120 nm thick Ag cap layer to prevent oxidation from occurring during the thickness measurement in the preparatory experiments. The measured Na films did not oxidize. If there was oxidation of the Na film, the degradation of the Ag cap layer was easily observed through optical microscopy. The total thickness or doping concentration during Alq3 and Na deposition was controlled by choosing the evaporation times for the photoemission study and device characterization. The work function (WF) is given by the equation, WF = Ek(Vac) − [Ek(EF) − hv], where hv is the photon energy, and Ek(EF) and Ek(Vac) are the kinetic energies (Ek) of photoelectrons emitted from the Fermi level and secondary electrons at a low-energy cutoff, respectively. Because we used a UPS source with an He I excitation line, hv is 21.2 eV. Ek(EF) was measured from the photoelectron spectrum of the Au coated Si substrate (metallic EF level). The onset of the secondary electron binding energy was determined by extrapolating solid lines from the background and from the straight onset in the UPS spectra. The onset of photoemission, corresponding to the vacuum level at the alkali metal doped Alq3 surface, was measured with a negative bias (−20 V) on the sample to avoid the detector work function. 2.2. OLED Fabrication. Soda-lime glass coated with indium tin oxide (ITO, 150 nm thick, ∼ 20 Ω/□) was used as the starting substrate. The surface of the glass was cleaned sequentially with acetone, isopropyl alcohol, and deionized
3. RESULTS AND DISCUSSION 3.1. UPS Measurements of Na-Doped Alq3. The WF of undoped Alq3 was 3.6 eV. WF spectra of Na:Alq3 samples remained at 3.6 eV ± 0.03 eV for 0 ≤ RNa ≤ 2.7 and shifted to lower values when RNa > 2.7. At RNa = 3.9, the WF was ∼0.6 eV less than when RNa ≤ 2.7 (Figure 2a). In the VB spectrum of undoped Alq3 (Figure 2b), the energy difference between the highest occupied molecular orbital (HOMO) and EF was 2.2 eV. In the VB spectra of Na:Alq3 samples (Figure 2b), the energy of the HOMO increased from 2.2 to 2.85 eV as the doping ratio increased from RNa = 0 to RNa = 2.7. The shift of the filled LUMO was the same as that of the HOMO. The energy difference between the filled LUMO and the HOMO was 1.65 ± 0.1 eV. The changes in the WF and VB spectra of the Na:Alq3 samples indicated that alkali metal doping did not induce interface dipole and band bending, but the LUMO moved 9160
dx.doi.org/10.1021/jp2102918 | J. Phys. Chem. C 2012, 116, 9158−9165
The Journal of Physical Chemistry C
Article
Figure 3. (a) N 1s spectra, (b) O 1s spectra, and (c) Na 2s spectra with Na concentration obtained using SRPES in situ measurements; (d) Na:Alq3 structure.
Figure 4. Energy level shift relative to undoped Alq3 film and HOMO position of (a) Na:Alq3 structure and (b) band diagram of Na:Alq3 structure with Na concentration. Energy level shift relative to undoped Alq3 film and HOMO position of (c) Na/Alq3 structure and (d) band diagram of Na:Alq3 structure with Na thickness using in situ SRPES and UPS measurements.
toward EF as in K-doped fullerides.6 Generally, if a change of the film’s WF23 or an interfacial dipole24 occurs, the vacuum level and LUMO shift toward lower binding energies and the HOMO shifts toward a higher binding energy with respect to EF (Figure 1b). However, the vacuum level, as shown by the WFs (Figure 2a) and the σ bonds,15a remained at 9.9 eV (Figure 2b, inset) for 0 ≤ RNa ≤ 2.7 in Na:Alq3 samples. The band gap of Na:Alq3 was not changed by alkali metal doping (not shown here).18 The energy of the HOMO shifted monotonically away from EF with increasing doping concen-
traion;17 thus, the gradually increased energy level of the HOMO meant that the LUMO moved toward EF (Figure 1d). Changes in the shape and energy levels of spectra in the vicinity of 10 eV when RNa = 3.9 showed that conformational changes occurred for highly doped Alq3 (RNa > 3). Generally, the backbone of Alq3 should be moved without alteration by alkali dopants because the σ bonds of organic matter are strong covalent chemical bonds.10c The σ bonds in organic molecules could be changed by severe structural alterations like distortion, internal rotation, or dissociation of bonds to reduce the total 9161
dx.doi.org/10.1021/jp2102918 | J. Phys. Chem. C 2012, 116, 9158−9165
The Journal of Physical Chemistry C
Article
Figure 5. Current density−voltage of (a) Na:Alq3 structure and (b) Na/Alq3 structure; luminance−voltage characteristics of (c) Na:Alq3 structure and (d) Na/Alq3 structure. Inset of panel c: luminance−current density of Na:Alq3 structure OLEDs with doping concentration.
did not form) for RNa < 3. We cannot observe any oxidation related spectra for either the Alq3:Na or the Na/Alq3 sample. The Na 2s spectrum of Na:Alq3 (Figure 3d) shows a chemical shift of 0.9 eV in the Na−Alq3 bond with respect to the metallic Na bond. The single peak at 65 eV corresponded to the Na− Alq3 bond for 0.3 ≤ RNa ≤ 2.7. Therefore, the codeposition of Na and Alq3 did not cause the formation of Na clusters or shift the vacuum level by atomic doping of Na when RNa < 3. When RNa > 3, the fwhm of the Na 2s spectrum broadened from 2.69 to 2.98 eV; this broadening indicates that the Na 2s spectrum consists of two components: the peak centered at 64.6 eV corresponds to the Na−Na bond and the peak at 65.5 eV corresponds to the Na−Alq3 bond (cf. the fwhm of the Na 2s spectrum of Na on Alq3 films is 3.26 eV). For RNa > 3, Na 2s core levels also shifted by about 0.6 eV toward higher binding energies, as did the WF, VB, and other core levels. As RNa increased, the intensity of the metallic Na bond increased. The presence of the Na−Na bond in the broad Na 2s spectra is evidence of Na metal clusters in highly doped Alq3 films. 3.3. Energy Level Shift and Band Diagram of NaDoped Alq3. When Na and Alq3 were codeposited, the energy of the HOMO increased from EF by 0.65 eV without significant shifts in the core-level energy18 or the WF (