Research Article www.acsami.org
Cite This: ACS Appl. Mater. Interfaces 2017, 9, 40564-40572
Increased Electromer Formation and Charge Trapping in SolutionProcessed versus Vacuum-Deposited Small Molecule Host Materials of Organic Light-Emitting Devices Yong Joo Cho,§ Scott Taylor,† and Hany Aziz*,§ §
Department of Electrical & Computer Engineering and Waterloo Institute for Nanotechnology (WIN) and †Department of Chemistry, University of Waterloo, 200 University Avenue West, Waterloo, Ontario N2G 3G1, Canada S Supporting Information *
ABSTRACT: We investigate and compare between organic light-emitting devices (OLEDs) fabricated by solution-coating versus vacuum-deposition. Electroluminescence, photoluminescence, and chromatographic measurements on typical OLED host materials reveal significant electromer formation in layers fabricated by solution-processing, pointing to stronger intermolecular interactions in these systems. Delayed electroluminescence measurements reveal that solution-processed layers also have increased charge traps. The findings provide insights on the morphological differences between solutionprocessed and vacuum-deposited materials and shed light on the root causes behind the lower electroluminescence stability of solution-processed OLEDs. KEYWORDS: solution-coated OLEDs, stability, electromer, single organic layer device, delayed electroluminescence
■
adversely affecting the quantum efficiency.22−25 Recently, we found that interactions between excitons and polarons during electrical driving can induce aggregation in a wide range of OLED host materials, a process referred to as exciton−polaron induced aggregation (EPIA).26−29 Such process was found to occur more rapidly in case of OLEDs fabricated by solutionprocessing in comparison to those fabricated by vacuumdeposition suggesting that the phenomenon may play a role in the lower stability of solution-processed devices.16 More recently, we found that this aggregation process is associated with the formation of some intermolecular electroplex species that involve both the aggregated host molecules and electron transport material molecules.30 Luminescence from such electroplex causes (or at least contributes to) the red-shift that is often observed in the electroluminescence (EL) spectra of OLEDs after prolonged electrical driving. In this study, we investigate and compare the electrical aging behavior in OLED host materials made by solution-processing versus vacuum-deposition. In order to eliminate confounding factors such as those arising from electrical aging in the CTL materials or from interactions between the host materials and CTLs such as those associated with electroplex species formation as noted above,30 the materials are studied in test devices consisting of only a single organic layer. The following three materials are investigated; 4,4′-bis(N-carbazolyl)-1,1′-
INTRODUCTION Organic light-emitting devices (OLED) are increasingly being utilized in TV and mobile display applications.1−5 OLEDs used in such commercially available products are almost exclusively fabricated by vacuum-deposition techniques. The prospect of fabricating OLEDs via solution-coating techniques has however always attracted significant interest because of its potential cost saving advantage over vacuum-deposition techniques especially for large-area products.6−9 Toward this end, the past few years have witnessed a significant progress in developing high quantum efficiency solution-processed devices utilizing small molecule materials, and efficiencies approaching the theoretical limit have been reported.10−15 Despite the progress in realizing impressive efficiencies, the limited electroluminescence stability of solution-processed OLEDs relative to their vacuumdeposited counterparts made of the same materials is still a major issue.16−19 The underlying factors remain largely unknown. In general, the deterioration in the electroluminescence efficiency of OLEDs is attributed to a number of degradation mechanisms that occur during electrical driving and can lead to a deterioration in device performance over time (often referred to as electrical aging). For example, chemical dissociation can occur to the organic materials under electrical (and exciton) stress, leading to a decrease in their quantum yield over time.20,21 Excitons in the emitting layer (EML) of the OLEDs can also be quenched by collisions with polarons that increasingly accumulate in the adjacent charge transport layers (CTLs) or at the interfaces between the EML and the CTLs, © 2017 American Chemical Society
Received: October 7, 2017 Accepted: November 2, 2017 Published: November 2, 2017 40564
DOI: 10.1021/acsami.7b15190 ACS Appl. Mater. Interfaces 2017, 9, 40564−40572
Research Article
ACS Applied Materials & Interfaces
Figure 1. Normalized electroluminescence spectra of vacuum-deposited (blue line) and solution-coated (red line) devices with (a) CBP, (b) 26DCzPPy, or (c) TCTA.
LiF/Al serve as hole and electron injection contacts, respectively. In order to be able to collect and study their EL spectra, the host material layer is used in a neat (i.e., undoped) form. Our previous studies have shown that molecular aggregation of host materials can occur even when they are doped with a guest emitter at concentrations typical of those used in common OLED EMLs.26,28 On this basis, the use of undoped layers allows more feasible observation of any emission from intermolecular species (which typically have emission bands in the 550−700 nm range and therefore become difficult to detect when a dopant is present), while, at the same time, still representing what happens in typical OLED host/guest EML systems. In one group of devices, the host material layer is made by solution-processing via spin-coating in a glovebox with a nitrogen atmosphere. In the other group, layer was made by thermal deposition in vacuum. Devices from the two groups are denoted to by (Sol) and (Vac), respectively. In both groups, the MoO3, LiF and Al layers are made by thermal deposition in vacuum. (In a previous work, we found that it is possible to spin-coat organic materials on a vacuumdeposited MoO3 layer without compromising its integrity.)16 Figure 1 depicts EL spectra collected from these devices, for different host materials, at a constant current density of 125 mA/cm2 in a nitrogen atmosphere. (This high current density is needed for obtaining adequate EL levels due to the poor charge balance and high hole leakage currents in these singlelayer devices as a result of the higher hole to electron mobility and elevated LUMO of these host materials, which makes electron injection inefficient). The spectra are normalized to the same peak intensity. As the figure shows, while the EL spectra from the vacuum-deposited devices show only one main emission peak in the 400−450 nm range, which can be ascribed to the S1 → S0 singlet transition, as expected for these materials, spectra from the solution-processed devices contained several bands, and for both CBP and 26DCzPPY, the strongest band was, surprisingly, in the 550−700 nm range, with a peak at around 610 nm. The long wavelength bands in the EL spectra of the solution-processed layers indicates that luminescence in these materials is dominated by emission from sub-band gap states or species instead of the intramolecular (and unimolecular or monomer) S1→ S0 singlet transition observed in case of their vacuum-deposited counterparts. Such sub-band gap states or species in case of the solution-processed devices could in general be due to (1) chemical impurities either due to material contamination or chemical decomposition and subsequent formation of byproducts (it should be noted however that both the solution-processed and vacuumdeposited devices were made using the same source material
biphenyl (CBP); 2,6-bis(3-(9H-carbazol-9-yl)phenyl)pyridine (26DCzPPy); and tris(4-carbazoly-9-ylphenyl)amine (TCTA). CBP was selected because it is the mostly widely used host and hole transport material in phosphorescent OLEDs and is often treated as a model compound for other carbazoles; the main class of materials predominantly used in phosphorescent emitter hosts. 26DCzPPy was selected because it is a wide band gap carbazole material and is relevant for blue phosphorescent OLEDs. TCTA was selected because it is a widely used hole transport material that, unlike CBP and 26DCzPPy, contains an amine moiety and, thus, is chemically different from them. This would provide an opportunity to identify the root causes behind the faster EPIA in solutionprocessed host materials and, hence, glean insights into the origins of the lower stability of these devices
■
EXPERIMENTAL SECTION
Materials. CBP, 2,2′,2″-(1,3,5-benzinetriyl-tris(1-phenyl-1-H-benzimidazole) (TPBi) and 26DCzPPy were purchased from Electronic Materials Index Co. TCTA and fac-tris(2-phenylpyridine) iridium (Ir(ppy)3) were purchased from Luminescence Technology Corp. 4,4′-Cyclohexylidenebis[N,N-bis(4-methylphenyl)benzeneamine (TAPC) was purchased from Shanghai Hanfeng Chemical Co. All materials have purity above 99% and were thus used without further purification. Device Fabrication. All devices were fabricated on prepatterned ITO substrates obtained from Kintec. Solution-processed layers: the organic materials were first dissolved in dry toluene to form 5.0 mg/ mL solution and then were coated using a WS-400/500 series spinner from Laurell Technologies Corp. at a spin rate of 500/2000 rpm for 5/ 25 s. The coated films were then annealed at 60 °C on a hot plate in a glovebox. Vacuum-deposited films: all layers were deposited using an Angstrom Engineering EvoVac system. All materials were evaporated at a rate of 0.1−1.0 Å/s at a base pressure of below 5 × 10−6 Torr. Device Measurement. Photoluminescent spectra were collected under monochromatic illumination from a 200 W Hg lamp equipped with an Oriel-77720 monochromator. High-performance liquid chromatography (HPLC) was performed using a C-18 reverse phase column (particle size: 5 μm, inner diameter: 4.6 mm, and length: 250 mm) of Shiseido CAPCELL PAK. The HPLC was a multiple setup system. The detector was a Waters 486 tunable absorbance detector. The controller was a Waters 600 controller. The instrument information on delayed EL measurement setting is as described in our previous work.31,32 All devices were kept and measured in a nitrogen atmosphere at all times.
■
RESULTS AND DISCUSSION Single organic layer test devices of the general structure ITO/ MoO3 (5 nm)/host material (30 nm)/LiF (1 nm)/Al (80 nm), using CBP, 26DCzPPy, or TCTA as the host material, are fabricated and tested. In these devices, the ITO/MoO3 and 40565
DOI: 10.1021/acsami.7b15190 ACS Appl. Mater. Interfaces 2017, 9, 40564−40572
Research Article
ACS Applied Materials & Interfaces
Figure 2. Normalized photoluminscence spectra collected from vacuum-deposited (blue line) and solution-coated (red line) devices with (a,b) CBP, (c,d) 26DCzPPy, or (e,f) TCTA, before and after electrical aging. The electrical aging time was 10 h and 10 min in case of the vacuum-deposited and solution-coated devices, respectively.
the faster aggregation in these systems, becomes particularly relevant for understanding the root causes of the lower stability of these OLEDs. In order to investigate the origins of the long wavelength bands in the solution-processed devices, we carry out photoluminescence (PL) measurements on the devices. Figure 2 shows PL spectra collected from the same devices before and after electrical driving. The vacuum-deposited devices were driven for more than 10 h, while the solution-processed devices were driven for about 10 min owing to their lower stability. As can be seen from the figure, the PL spectra of the solutionprocessed devices do not show the long wavelength bands observed in the EL spectra, not even after electrical driving for the given period of time. Rather, the PL spectra showed the familiar blue fluorescence (peak around ∼410 nm) arising from the S1 → S0 transition of these host molecules and resembled those collected from the vacuum-deposited devices. The results suggest that the long wavelength bands that appear in the EL
batches, so any increased impurity levels in case of the solutionprocessed devices would only be a result of the fabrication process, such as from the solvent or the coating environment) and/or (2) the formation of intermolecular emissive species (such as molecular dimers, electromers, etc.) possibly as a result of morphological factors associated with solution-processing that allow for stronger intermolecular interactions. The formation of electromer species, defined as intermolecular species that exist only during electrical excitation, and with emission bands in the 550−650 nm, has been reported for some OLED materials before, including amines and carbazoles.31−40 Knowing that exciton−polaron interactions bring about faster aggregation in host materials made by solutionprocessing and that the phenomenon plays a role in their lower EL stability in comparison to their vacuum-deposited counterparts,16,26 the question of whether the long wavelength bands in Figure 1 may be due to increased electromer species in case of solution-processed materials, which may be the prerequisite for 40566
DOI: 10.1021/acsami.7b15190 ACS Appl. Mater. Interfaces 2017, 9, 40564−40572
Research Article
ACS Applied Materials & Interfaces
More importantly, increasing the bias brings about a relative increase in the intensity of the 625 nm bands in these devices. This can be seen more easily in the figure insets that show the same spectra after being normalized to the same intensity at 410 nm (the monomer band) to make visualizing the spectral shifts easier. The results clearly show that the intensity of the 625 nm bands becomes increasingly stronger relative to the other bands upon increasing the bias, indicating that the underlying species increases upon increasing the bias. This trend suggests that these long wavelength bands in the solutionprocessed device may indeed be arising from electromer species involving the carbazole moiety.33,37,38 To verify if the long wavelength band indeed corresponds to emission from electromer species, we also fabricated and tested single-layer devices made of TAPC, which is known to exhibit electromer formation even when made by vacuum-deposition. Figure 4e presents data from this device. Clearly, the TAPC device exhibits essentially the same behavior, showing an EL band at 580 nm that has been ascribed before to electromer formation, in addition to the 460 nm monomer S1 → S0 band. Again, the intensity of the 580 nm band increases relative to the monomer band upon increasing the bias. The similar behavior of the 625 nm band in case of the solution-processed CBP and 26DCzPPy with the 620 nm electromer band of TAPC indicates that the bands in CBP and 26DCzPPy are indeed caused by electromers in these materials. In this regard, the weaker electromer band in case of TCTA than in case of CBP or 26DCzPPy in Figure 1 may perhaps have its origins in the differences in their molecular geometries. TCTA has a spherical (or propellershaped) molecular structure, whereas both CBP and 26DCzPPy are generally flatter. Therefore, once can expect CBP and 26DCzPPy molecules to be able to stack more easily, hence interact more strongly, than in the case of TCTA. Because electromers are intermolecular species, one can expect the intermixing of another (second) material with the host material to reduce its formation. This is because the foreign molecules will act as molecular spacers that can separate and, hence, reduce the interactions between the host material molecules. To test for this effect, we investigate the impact of intermixing TPBi with the host materials in the solutionprocessed devices. The device structure was as follows: ITO/ MoO3 (5 nm)/CBP:TPBi or 26DCzPPy:TPBi (30 nm, x %)/LiF (1 nm)/Al (80 nm) where x represents the concentration of TPBi, with concentrations of 0, 5, 10, and 50% (by weight). Figure 5a,b shows the EL spectra of these devices for CBP and DCzPPy, respectively, for x = 0, 5, 10 and 50% (by weight) Clearly, introducing the TPBi leads to a suppression in the 625 nm bands, proving the intermolecular nature of the underlying species, thus ascertaining its electromer origins. Unlike the 625 nm band that decreases in intensity upon increasing the TPBi, one can see another band, at ∼500 nm, that increases in intensity in the same direction. This band corresponds to the host aggregates/TPBi complex reported in our previous work, which, given its nature, naturally intensifies upon increasing the TPBi concentration.30 The above results therefore show that solution-processing leads to increased electromer formation in OLED host materials. Unlike in the case of vacuum-deposition where the organic material thin film, once formed on the substrate, does not get subjected to temperatures well above room temperature, solution-processing typically includes an additional step of baking or curing the formed film at higher temperatures in order to remove residual solvents. Therefore, in order to
spectra of the solution-processed devices are not the result of impurities or chemical decomposition byproducts (that may, for example, be receiving energy from the monomers via an energy transfer process, and thus quenching the 410 nm emission). The fact that these long wavelength bands are visible only in EL suggests they may indeed be arising from an electromer species that exists only during electrical bias.31 We also carried out high-performance liquid chromatography (HPLC) tests to investigate if the solution-processed materials contain higher levels of impurities, possibly as a result of unintentional contamination from the processing environment. For this test, thin films of the host materials are fabricated on clean substrates using vacuum-deposition or solution-coating under the same conditions as those used for the devices above. The organic films are then dissolved off the substrates using acetonitrile, and then the films are collected and tested. Figure 3 shows the HPLC chromatograms for the case of CBP. The
Figure 3. High-performance liquid chromatography (HPLC) chromatograms of CBP powder (black line), vacuum-deposited CBP film (blue line), and solution-coated CBP film (red line). The chromatograms are normalized to the same height. Inset: magnified view of the same spectra.
purity of the original source material (referred to as CBP powder) was also tested for comparison. The tests show that the purity level of CBP after solution-processing was similar to that of its vacuum-deposited counterpart and also to that of the original stock. These results rule out that the long wavelength bands observed in the EL spectra of the solution-processed devices may be associated with chemical contamination of the materials during device fabrication. The results therefore suggest that these bands are indeed the result of morphological (rather than chemical) factors in the solution-processed films that alter their EL spectra from their PL spectra, suggesting the phenomenon may possibly be associated with the formation of electromer species. It is known that the formation of electromer species depends strongly on electric fields, causing the intensity of their bands in spectra to increase upon increasing bias.33−37 Therefore, to determine if the long wavelength bands observed in our devices may indeed be due to electromers, we study changes in the EL spectra with applied bias. Figure 4 shows EL spectra of singlelayer devices made of CBP and 26DCzPPy made by vacuumdeposition or solution-coating, tested under different bias levels. From the figure, it can be seen that increasing the bias brings about only very small changes in the EL spectra of the vacuum-deposited devices. In contrast, the changes in case of the solution-processed devices were much more significant. 40567
DOI: 10.1021/acsami.7b15190 ACS Appl. Mater. Interfaces 2017, 9, 40564−40572
Research Article
ACS Applied Materials & Interfaces
Figure 4. Electroluminescence spectra of vacuum-deposited and solution-coated devices with (a,b) CBP, (c,d) 26DCzPPy, or (e) TAPC, under different bias voltages. Insets: the same spectra after being normalized to the same height at the monomer band (except in (d), which shows spectra at the higher voltage).
Figure 5. Electroluminescence spectra of solution-coated devices with (a) CBP or (b) 26DCzPPy, containing 0, 10, 25, and 50% TPBi. The spectra are normalized to the same monomer band height.
investigate if the increased electromer formation in the case of the solution-processing might be due to thermally induced
morphological changes, we test the effect of subjecting the vacuum-deposited CBP films to the same temperatures and 40568
DOI: 10.1021/acsami.7b15190 ACS Appl. Mater. Interfaces 2017, 9, 40564−40572
Research Article
ACS Applied Materials & Interfaces
Figure 6. (a) Electroluminescence (EL) spectra of CBP vacuum-deposited devices before (continuous line) and after (dotted line) thermal annealing at 60 °C. (b) EL spectra collected from the annealed device after different time intervals of continuous electric driving at 125 mA/cm2.
The formation of electromers in solution-processed devices suggests that these films may have increased morphological and structural defects relative to their vacuum-deposited counterparts, for example, due to variations in the extent of intermolecular interactions and/or molecular packing density from one location to another within the film. Indeed, Kwon et al. have reported a correlation between increased electromer formation and charge trap density.41 Therefore, in order to investigate this possibility, we study and compare the delayed EL characteristics of single-layer devices made by solutionprocessing versus vacuum-deposition. In these devices, a phosphorescent emitter is used as a guest into the host materials. The purpose of using the guest is two folded: (1) it helps enhance the delayed EL signal and thus make the measurements possible, and (2) it allows the test devices to closely mimic the situation in actual OLED EMLs. In delayed EL measurements, a device is driven using a square pulse driving scheme with a pulse width of 0.5 ms (the pulse is long enough for prompt EL to reach its steady-state intensity). An optical shutter opens to collect delayed EL around 0.1−0.3 ms after the end of the forward bias pulse, which is significantly longer than the lifetime of Ir(ppy)3 triplet state lifetime (
