Jason D.
Slinker
a,
Ji-Seon
Kim
b,
Samuel
Flores-Torres
c,
Jared H.
Delcamp
a,
Héctor D.
Abruña
c,
Richard H.
Friend
b and
George G.
Malliaras
*a
aDepartment of Materials Science and Engineering, Cornell University, Ithaca, New York, 14853-1501, USA. E-mail: ggm1@cornell.edu; Fax: +1-607-255-2365; Tel: +1-607-255-1956
bCavendish Laboratory, University of Cambridge, Madingley Road, Cambridge, UK CB3 0HE
cDepartment of Chemistry and Chemical Biology, Cornell University, Ithaca, New York, 14853-1301, USA
First published on 27th November 2006
We have used in situ Raman spectroscopy to identify a luminescence quencher formed during organic light-emitting device operation. Raman spectroscopy revealed that oxo-bridged dimerization occurs during the operation of [Ru(bpy)3]2+(PF6−)2 devices, where bpy is 2,2′-bipyridine. Photoluminescence spectroscopy showed that oxo-bridged dimers such as [Ru(bpy)2(H2O)]2O4+(PF6−)4 effectively quench photoluminescence. Comparison of the Raman spectra from devices with the spectra from prepared blended films of [Ru(bpy)3]2+(PF6−)2 and [Ru(bpy)2(H2O)]2O4+(PF6−)4 demonstrated that sufficient dimerization occurs in the device to account for the luminescence quenching observed upon device driving. Dimerization occurred particularly where oxygen and moisture could penetrate the organic film. Dimerization could be a general failure mode of organic electroluminescent devices that incorporate metal complexes. Understanding failure under device-relevant conditions can lead to the development of materials and devices that are intrinsically more resistant to degradation.
Fig. 1 The structure of [Ru(bpy)3]2+. |
Operation of iTMC devices with air stable electrodes enables the study of chemical changes occurring in the complex layer during device operation without interference from electrode degradation. We use in situ Raman spectroscopy22 to identify a luminescence quencher formed during organic light-emitting device operation. [Ru(bpy)3]2+ is found to dimerize to an extent that accounts for loss of luminescence upon device driving. Dimerization occurs particularly where oxygen and moisture can penetrate the organic film. Dimerization could be a general failure mode of organic electroluminescent devices that incorporate metal complexes.16 Understanding failure under device-relevant conditions can lead to the development of materials and devices that are intrinsically more resistant to degradation.
Fig. 2 In situ Raman spectra of a Au/[Ru(bpy)3]2+(PF6−)2/Au device in operation, as well as the spectra for spin-cast films of synthesized [Ru(bpy)3]2+(PF6−)2 and [Ru(bpy)2(H2O)]2O4+(PF6−)4. The spectra are offset along the y-axis for clarity. |
Fig. 3 Images of the photoluminescence of a planar Au/[Ru(bpy)3]2+(PF6−)2/Au device before (upper panel) and after (lower panel) 1 hour of operation. |
In order to elucidate the nature of the chemical changes and luminescence quenching upon running the device, Fig. 2 also presents the Raman spectra for spin-cast films of [Ru(bpy)3]2+(PF6−)2 and [Ru(bpy)2(H2O)]2O4+(PF6−)4. The latter compound, shown in Fig. 4, was recently identified by Soltzberg et al. in post-run [Ru(bpy)3]2+(PF6−)2 devices using mass spectrometry and found to be an effective quencher of luminescence.23 It should be noted that the integration times of the [Ru(bpy)3]2+(PF6−)2 and device spectra are longer (10× and 20×, respectively) than that of the [Ru(bpy)2(H2O)]2O4+(PF6−)4 for clarity. For equivalent integration times, the Raman scattering of the dimer is considerably more intense than [Ru(bpy)3]2+(PF6−)2. Also, the [Ru(bpy)3]2+(PF6−)2 spectrum in Fig. 2 has been adjusted to compensate for the fluorescence background as noted in the Experimental section.
Fig. 4 The structure of [Ru(bpy)2(H2O)]2O4+. |
For the [Ru(bpy)3]2+(PF6−)2 spectrum, all of the Raman peaks are consistent with previous reports and correspond to vibrational modes of the bpy ligand24–34 and the PF6− counterion.35,36 A number of observations arise from comparison of the [Ru(bpy)2(H2O)]2O4+(PF6−)4 spectrum with that of [Ru(bpy)3]2+(PF6−)2. First, all of the Raman peaks occurring in [Ru(bpy)3]2+(PF6−)2 also occur in [Ru(bpy)2(H2O)]2O4+(PF6−)4, which is consistent with the fact that the latter compound also contains bpy and PF6−. Furthermore, the dimer exhibits several additional vibrational peaks around 380 cm−1 and 760 cm−1 arising from the Ru–O–Ru bond. These are the distinctive Raman peaks for RuIII–O–RuIII complexes.37–42 The most significant is the symmetrical Ru–O–Ru stretch37,39 appearing near 380 cm−1. These observations can account for the features seen in the device spectra of Fig. 2. The initial peaks clearly arise from [Ru(bpy)3]2+(PF6−)2. At later times, clusters of peaks emerge about 380 cm−1 and 760 cm−1. These peaks, though broadened by the presence of additional peaks relative to the pure [Ru(bpy)2(H2O)]2O4+, are indicative of the Ru–O–Ru bond. This broadening may suggest the formation of other oxo-bridged dimeric compounds in addition to [Ru(bpy)2(H2O)]2O4+. Oxo-bridged dimers of lower molecular weight similar to [Ru(bpy)2(H2O)]2O4+ were observed in the mass spectra of post-mortem [Ru(bpy)3]2+(PF6−)2 devices,23 though from mass spectrometry alone it was unclear if the additional dimers were the result of device driving or a consequence of the laser measurement. More highly oxidized forms37–44 of [Ru(bpy)2(H2O)]2O4+ could also contribute to the broadening in the Raman spectra observed upon device driving. The RuIII–O–RuIII dimer can reversibly undergo four successive oxidations to RuV–O–RuV, each with a distinctive Raman spectrum.37–42
One final feature distinguishes the Raman scattering of the dimeric [Ru(bpy)2(H2O)]2O4+(PF6−)4 compound from [Ru(bpy)3]2+(PF6−)2. As noted above, the Raman scattering from the dimer is more intense than that of [Ru(bpy)3]2+(PF6−)2. This arises because the absorption of the dimer at the 633 nm excitation wavelength is significantly greater than that of [Ru(bpy)3]2+(PF6−)2, contributing to a stronger resonance Raman effect.37–42 Thus, dimer formation would also account for the increase in the Raman signal observed upon running a device.
The luminescence quenching properties of [Ru(bpy)2(H2O)]2O4+ are revealed in its photophysical characteristics. The absorption of spin-cast films of [Ru(bpy)3]2+(PF6−)2 and [Ru(bpy)2(H2O)]2O4+(PF6−)4, as well as the photoluminescence (PL) spectrum of [Ru(bpy)3]2+(PF6−)2 (450 nm excitation) are given in Fig. 5. We observe no detectable PL of the oxo-bridged dimer within the range of excitation wavelengths from 350 to 800 nm, nor has any PL been reported to date. The absorption spectrum of [Ru(bpy)2(H2O)]2O4+(PF6−)4 is considerably red-shifted relative to that of [Ru(bpy)3]2+(PF6−)2, a feature that contributed to the stronger resonant Raman effect in the dimer for the given 633 nm excitation wavelength, as mentioned earlier. Furthermore, the absorption spectrum of [Ru(bpy)2(H2O)]2O4+(PF6−)4 overlaps the emission spectrum of [Ru(bpy)3]2+(PF6−)2, suggesting that [Ru(bpy)2(H2O)]2O4+(PF6−)4 should serve as an effective photoluminescence quencher, a characteristic trait9,45 in the degradation of [Ru(bpy)3]2+(PF6−)2 devices in general and noted in the device above.
Fig. 5 UV-Vis absorption (top curves) and photoluminescence (lower curve) of [Ru(bpy)3]2+(PF6−)2 (solid lines) and [Ru(bpy)2(H2O)]2O4+(PF6−)4 (dotted line). No photoluminescence was observed from the [Ru(bpy)2(H2O)]2O4+(PF6−)4 dimer. |
The extent of dimerization occurring within the device film can be qualitatively estimated from a comparison of the device spectra with blended films of [Ru(bpy)2(H2O)]2O4+ and [Ru(bpy)3]2+(PF6−)2, the spectra of which are shown in Fig. 6. Note that these spectra were all collected at the same integration time and have not been adjusted for the fluorescence background. In these spectra, dimeric peaks about 380 cm−1 become visible at a 333 : 1 w/w [Ru(bpy)3]2+(PF6−)2-to-dimer ratio and increase with increasing dimer concentration. In order to obtain a quantitative extent of dimerization for the in situ device spectra, the intensities of the spectra from 300 cm−1 to 400 cm−1 were integrated, representing the total signal from a region where the dimer peaks are dominant, and compared to this quantity for the blended films. After 1 hour of operation, the amount of dimerization corresponds to approximately 1% w/w dimer-to-parent complex. This ratio corresponds to a sufficient concentration for efficient energy transfer and thus quenching from a [Ru(bpy)3]2+(PF6−)2 host to a guest, as demonstrated by phosphorescent doping within a [Ru(bpy)3]2+(PF6−)2 matrix.46 This is found to be true, as a blended film of 1% w/w dimer-to-[Ru(bpy)3]2+(PF6−)2 dramatically diminishes the PL quantum yield, Φ, by two orders of magnitude (Φ/Φ0 = 0.01). Quenching of the PL under 380 nm illumination is qualitatively visible in this concentration range in the blended films of Fig. 7. These observations lead to the assertion that oxo-bridged dimer formation occurs during device operation to a sufficient extent to significantly contribute to luminescence quenching.
Fig. 6 Raman spectra of blended films of [Ru(bpy)3]2+(PF6−)2 and [Ru(bpy)2(H2O)]2O4+(PF6−)4, reported in [Ru(bpy)3]2+(PF6−)2-to-dimer w/w ratios. All curves were collected at the same integration time and are offset along the y-axis for clarity. |
Fig. 7 Images of [Ru(bpy)3]2+(PF6−)2 and [Ru(bpy)2(H2O)]2O4+(PF6−)4 blended films (reported in w/w ratios) on quartz substrates under white light (above) and 380 nm UV illumination (below). |
Spatially-resolved studies of sandwich structure ITO/[Ru(bpy)3]2+(PF6−)2/Au devices revealed that dimer formation was much more prevalent near the edges of the top electrode. Bubbles were also found to form near these edges during device operation, consistent with previous reports.45,47 These observations are suggestive that dimer formation is enhanced in regions where oxygen and/or moisture are present, which are possible reactants for the conversion of [Ru(bpy)3]2+(PF6−)2 to [Ru(bpy)2(H2O)]2O4+(PF6−)4. This is also consistent with the observation that device lifetime is diminished with exposure to moisture.45,47 It is also observed that light emission is quenched in the presence of oxygen. In Fig. 8, the response of an ITO/[Ru(bpy)3]2+(PF6−)2/Au device to oxygen exposure is shown. At a 3 V bias, the device is first turned on under vacuum, then alternately run under oxygen-rich and vacuum environments. Each time oxygen is introduced, the radiant flux drops dramatically, while the current remains essentially constant, consistent with the generation of a quencher upon reaction with O2. When vacuum is restored, the current holds steady near the initial state, and, while the radiant flux does exhibit some recovery (in part due to the gradual increase in current), it is significantly diminished with each successive introduction of oxygen.
Fig. 8 The current (solid line) and radiant flux (dotted lines) of an ITO/[Ru(bpy)3]2+(PF6−)2/Au device operating at 3 V under alternating vacuum and oxygen-rich environments. The downward arrows correspond to the introduction of oxygen to the chamber, while the upward arrows indicate where vacuum was restored. |
A mononuclear complex, namely [Ru(bpy)2(H2O)2]2+, has previously been suggested as a quencher,45,47 but neither this nor any other molecule has been directly identified as a decomposition product in a device. It should be noted that the Ru–OH2 bond is not easily detectable37,39 with Raman spectroscopy.
Future work is still necessary to fully understand the degradation of [Ru(bpy)3]2+(PF6−)2 under device relevant conditions. The mechanistic pathway is still not known, though several viable pathways are noted in the literature.39,42,44,48–49 However, the in situ identification of a quencher in this work represents an important step in understanding chemical reactions that take place under device-relevant conditions.
The as-measured Raman spectrum for [Ru(bpy)3]2+(PF6−)2 is complicated by the presence of a fluorescence background which obstructs observation of the intrinsic Raman shifts. For this reason, a polynomial fit was applied to the [Ru(bpy)3]2+(PF6−)2 spectrum, and this fit was numerically subtracted from the as-measured [Ru(bpy)3]2+(PF6−)2 spectrum to provide the adjusted spectrum presented in Fig. 2. In this way, the relative scaling of the intrinsic Raman shifts is preserved. Estimation of the amount of dimerization was made by comparing an integrated portion of the run device spectrum to that from several blended films of [Ru(bpy)3]2+(PF6−)2 and [Ru(bpy)2(H2O)]2O4+. The blended films were prepared on Spectrosil substrates over the dimer-to-monomer concentration range 10% to 0.01% w/w. A polynomial fit was applied to and subtracted from each spectrum to account for the background, and the spectra were numerically integrated in the range 300 to 400 cm−1.
In order to ensure that no changes were occurring in the Raman spectra due to the laser itself, both films and devices of [Ru(bpy)3]2+(PF6−)2 were left under the Raman laser for 2 hours, with spectra recorded immediately before and after exposure. No detectable changes in the Raman shifts were observed. Furthermore, the Raman spectra of thin film samples of [Ru(bpy)3]2+(PF6−)2 were recorded before and after baking for 20 h at 100 °C in air, and no change was observable. In general, the Raman spectra of [Ru(bpy)3]2+(PF6−)2 and [Ru(bpy)2(H2O)]2O4+ films did not change with storage in air up to 1 month.
This journal is © The Royal Society of Chemistry 2007 |