Chapter 3 The materials and fabrication process
3.3 The thermal evaporation system and other instruments
Our thermal coater system is provided by ULVAC, and is showed in the following picture. It is composed of five chambers, each majorly for a specific usage.
Organic layers Cathode
Anode (ITO)
Emission part( The cross section)
Figure 3-2 The ULVAC thermal coater
Chamber 1 evaporates the hole/electrons transportation layers, chamber 2 evaporates the emission layers, and chamber 3 evaporates the metals and p- injection layer materials. Chamber 4 is for ITO sputtering and Chamber 5 is only a pre-treatment chamber.
Parts inside the chamber include:
(1) Holder: The samples are placed on top of it and when the main shutter
& the small shutter are open, the materials can be evaporated to the sample.
(2) Crystal sensors: The deposited film thickness can be monitored by a gold crystal material, which has a frequency of 5MHz.
(3) Main/ Small Shutter: Each chamber has a main shutter and six small shutters.
(4) Crucible: Materials is injected inside the crucible, and when heated to the desired temperature, the deposition will come out.
Chapter 4
positions also lie in the white region. The addition of NPB layer between the red and blue emission layers is essential as it acts as an efficient hole transport material to increase hole concentration for recombination. If NPB was removed, the emission was almost totally red since the holes cannot penetrate further into the other emissive layers deeper inside the structure. For the electron transportation layer, BPhen is chosen other than Alq3 due to its higher electron mobility in order to balance the charge-carriers for recombination. The electron and hole pairs will be confined inside the structure since a p-type/intrinsic/n-type junction is produced. Because of the higher eye sensitivity of our eyes in the green to yellow region near a wavelength of 555 nm, we then introduced a yellow layer tetra-chromatic white OLED device with four emission layers. This layer is added between the blue and green emission layers which match well with the green emission layer without creating another barrier. Our tetra-chromatic device has the following structure.ITO/CFx/50% NPB:50% WO3 (200 Å)/NPB (100 Å)/60% Rb:40%
Alq:1% DCJTB (100 Å)/NPB (40 Å)/ α,α-MADN:7% BpSAB (200
Å)/Alq:3%Rb (40 Å)/ Alq:1% C545T (100 Å) /BPhen (100Å)/ BPhen:
5% Cs2CO3 (200 Å)/Al (1500 Å).
Figure 4-1 The device structures (Set A)
As shown in Figure 4-2, our tetra-chromatic white OLED spectrum showed five peaks near 455, 479 (blue), 519 (green), 555 (yellow) and 608 (red) nm, with an overall FWHM well over 200 nm. Compared with the tri-chromatic spectrum, the nonuniform intensity among the R, G, B peaks are now alleviated with the additional yellow emission. While rubrene can trap charge carriers, further improvement of charge-carrier balance in holes and electrons is possible [62]. From the inset in Figure 4-2, a higher current efficiency is also obtained for the tetra-chromatic white OLED, the final CRI reaches 87. The value is still not very ideal due to the orange-reddish emitter DCJTB which is not saturated in red enough.
Figure 4-2 EL Spectrums of tri- and tetra-chromatic OLED devices under a luminance of 3000 nits. (Inset) The current efficiencies of
tri- & tetra- devices)
Figure 4-3 shows the viewing angle dependence of our device, which shows nearly the same intensity throughout a range of ±60 viewing angle. As this light source is not Lambertian, uniform illumination of the room is possible with our device due to very low angular dependency.
Figure 4-3 Viewing angle dependence of tetra-chromatic WOLED
Figure 4-4 and Figure 4-5 shows a specific advantage of the tetra-chromatic device in which a yellowish color is shown under low drive current density, with CIEx,y near (0.40, 0.44) can be readily obtained.
Similar color for the twilight sky is believed to be important in the regulation of the human circadian rhythm during the post-working duration [63]. Under high current densities for application of illumination in the daytime, a very near white point of CIEx,y (0.33, 0.36) can also be achieved. These CIEx,y points appear to lie close to the locus points that follows the line of a black-body radiator for illumination application.
Figure 4-4 2-D CIEx,y dependence of the devices
Figure 4-5 3-D CIEx,y dependence with the luminance
From Figure 4-4, we can observe that the tetra-chromatic OLED differs from the tri-chromatic OLED in the colors. With a shift of about x=0.05 to the right side. Each of the three dots represents current densities of 0.5, 20 and 200 mA/cm2. By this way, the tetra-chromatic devices will also have CIE positions much closer to the white original point at (0.33, 0.33) under higher current density.
The spectral dependence with the luminance is shown in Figure 4-6, we can see that under lower luminance, the blue emission intensity is weaker, and gets stronger and stronger under higher brightness, thus leads to a more saturated white emission.
Figure 4-6 Spectral dependence with the luminance (Inset: the spectrum of twilight & daylight sky color)
From the efficiency Table 4-1, we can observe that the tetra-chromatic device contributes to the rise of yield from 8.3 to 8.7 cd/A, due to the higher sensitivity of yellow light to the human being’s eye, but while rubrene is a material that will extensively trap electrons [62], the voltage will also rises 0.6 V, so there is no improvement in the power efficiency. Both devices show white light emission with CIE positions of (0.30, 0.39) and (0.45, 0.40) under 20 mA/cm2 .
Table 4-1 EL performance of the devices (Set A) Device Voltage
Figure 4-7 (top) The %NTSC color gamut for device set A (bottom)The devices after color filter transmission(left:
tri-chromatic, right: tetra-chromatic)
The color rendering indexes are very good with values of 87 and 89.
Also good %NTSC color gamut was achieved with values near 70, the tetra-chromatic device have lower %NTSC value due to the insert of an yellow component, which shifts the green coordinate to the right, thus lead to the shrinkage of the %NTSC. Higher %NTSC values can also be reached by selecting a higher quality color filter. But fluorescent devices efficiencies are not enough for white solid state lighting, so we will next focus on using phosphorescent materials inside the OLED device for their capability for producing higher efficiencies. Our next part of research will emphasize on phosphorescent white OLED devices.
4.2 Phosphorescent white OLED
In the case of phosphorescent white OLED, tri-chromatic OLED devices were fabricated more often using the following materials—a green phosphorescent green material Ir(ppy)3. For the blueemission, we
uses a well known greenlish-blue dopant FIrpic. For red emission dopant, we uses Ir(piq)3, a well-known red phosphorescent dopant material which emits pure red color with CIE coordinates of (0.67, 0.33). But unfortunately, Ir(piq)3 has a triplet bandgap of 2.0 eV and its HOMO and LUMO are 5.2 eV and 3.2 eV, respectively, which is rather deep compared with other dopant materials. Although CBP with triplet bandgap of 2.6 eV is commonly used as a host for Ir(piq)3, the large energy gap between CBP and Ir(piq)3 results in charge trapping in the red dopant rather than energy transfer from CBP to Ir(piq)3 [64]. Therefore, a triplet sensitizer which can induce efficient energy transfer from CBP to Ir(piq)3 is necessary to get high performances in Ir(piq)3 devices. Ir(ppy)3
has a triplet bandgap of 2.4 eV compared with 2.0 eV of Ir(piq)3 and it may be suitable as a sensitizer for Ir(piq)3. In addition to energy bandgap, there should be an overlap of emission spectrum of sensitizer and absorption spectrum of Ir(piq)3 to get high efficiency through efficient energy transfer.
So for producing white light, we first exanimated a blue phosphorescent emission layer plus a red dopant Ir(piq)3 doped inside the green emission layer Ir(ppy)3. The structure is shown in Figure 4-8, we devote these set of devices as Set B.
Figure 4-8 A totally phosphorescent white OLED device (Set B)
For the ease of the fabrication process, the whole multi-emission layer have the same host material CBP throughout, and is composed of an FIrpic doped CBP blue EML, an Ir(ppy)3 doped CBP green EML, and an x % Ir(piq)3 co-doped inside the green emission layer. The CBP layer close to NPB is to prevent the emission of NPB, and the layer between the two emission layers is to restrict the flow of the energy transfer from the blue emission layer to the other layers, which quenches the emission of FIrpic. And also a very thin layer hole blocking material BCP is added.
By trapping the excitons, their residence time and recombination probability in the EML were increased, leading to a concomitant increase in OLED efficiency. A 30-nm-thick layer of Alq3 was used to transport and inject electrons into the EML. Although the emission layers are all composed of phosphorescent materials, Ir(piq)3 while having a deep triplet energy, which leads to high driving voltage at low doping concentration, is still not preferred. According to research data, the efficiency drops tremendously and the driving voltage rises heavily for
higher red doping concentrations, which means that more excitons trapped in this deep triplet level is not preferable.
Figure 4-9 The HOMO and LUMO for each material in the device (Set B)
Table 4-2 EL performance of the devices (Set B)
Device Voltage Yield(cd/A) Eff.(lm/W) EQE(%) CIE
1% 8.1 32.3 12.5 10.8 (0.27, 0.55)
1.5% 8.1 24.2 9.4 9.2 (0.29, 0.52)
2% 8.5 20.2 7.4 8.3 (0.32, 0.49)
2.5% 9.4 16.1 5.4 7.8 (0.38, 0.46)
According to Figure 4-10 and Chart 4-2, the efficiency drops from 32.3 cd/A for 1.0% concentration to 16.1 cd/A for 2.5% of Ir(piq)3, while the voltage raises from 8.1 V to 9.4 V under 10 mA/cm2 current density.
Hence, a doping concentration of at least 2.5% is needed to produce acceptable white emission. The color rendering index is fair with a value of 71.
Figure 4-10 The efficiencies changes with the doping concentration of Ir(piq)3 (inset) The EL spectrum
A disadvantage is that the emission of light deviates too much from the white point, which only has a CIE near (0.36, 0.48). This is because the FIrpic is not blue enough having a peak wavelength near 500 nm with CIE position near (0.18, 0.40). So, in order to solve this problem, one is to invent a more saturated blue phosphorescent material, and another is to use a blue fluorescent material with a deeper blue emission
and combine it with phosphorescent materials with other colors.
Figure 4-11 (left)The CIE 1931 positions of the devices (right)The %NTSC color gamut
Figure 4-12 Devices Structures (Set C)
4.3 Phosphorescent and fluorescent combined white OLED
Combining both fluorescent blue emission and phosphorescent green and red emission was then studied. We used the same structure as the totally phosphorescent device and only exchanged the FIrpic doped CBP host layer into a highly efficient sky blue fluorescent layer with MADN
dopant BUBD-1, we called these devices as Set C. The emission of this blue layer has a much bluer color rather than FIrpic which is an advantage in the fabrication of white OLED devices. The emission spectrum and CIE 1931 positions are shown in the Figure 4-13. For the white device, the fluorescent blue-emitting unit is formed close to the ITO anode side in each of the devices because the plasmon-quenching effect from the Al cathode on the blue emission could be avoided with this arrangement in order to obtain a maximal blue emission[65]. Again we evaluated the dopant concentration of Ir(piq)3 inside the Ir(ppy)3 layer, for this layer nearly defines the total efficiencies and CIE positions. We tried 2.0% and 3.0%, and found that the one with a 3.0% shows a too strong red emission thus gives a purple device, and the current efficiency weakens to 7.3 cd/A. The best dopant concentration will be 2.0% in order to enter the white region. This is 0.5% lower than the totally phosphorescent device. This reason can be explained because the sky blue fluorescent dopant BUBD-1 has a deeper blue emission compared with FIrpic, which pulls the CIE positions with lower y coordinates, thus a lower red composition is needed. This is a good effect while the Ir(piq)3
material is not good for the total efficiency of the device.
Figure 4-13 EL spectrum and CIE positions
For its performance, we will point out that the fluorescent blue EL unit is a critical component in determining color gamut and efficiency of the white OLEDs. Our panel showed a peak current efficiency of 14.7 cd/A at a luminance of 147 nits, and under near 1000 nits still owns an efficiency of 12.5 cd/A, power efficiency of 6 lm/W, and EQE near 7%.
The CIE positions are at (0.40, 0.43).
The performance of the RGB OLED as the light source in an RGB color display was evaluated by mathematically applying the transmission curves of color filter arrays (CFA) to the device. The outcome is an RGB system shown in Figure 4-14. Corresponding color coordinates compared with the original NTSC values are also shown, the %NTSC color gamut improved from 56.7% to 62.9% by using the sky blue fluorescent emitter rather than FIrpic. We note that CIEx,y of red and green peaks and especially the red coordinates lies practically on the spectral locus, indicating nearly mono-chromatic light. But the blue peaks still largely
differ from the %NTSC, which means that our blue emitter is still not bluish enough and the color filter is not very well, leading to poor
%NTSC color gamut ratios, so further improvement with deeper blue materials need to be estimated.
Figure 4-14(top) The %NTSC color gamut for the devices (bottom)The devices after color filter transmission(left:
phosphorescent, right: fluorescent plus phosphorescent)
Table 4-3 EL performance of the devices (Set C)
Further improvement by shifting the recombination zone was then performed. Although our device already has good performance, we predicted the efficiency could further be improved by modifying the distance of the emission layers to the anode and cathode. We first modified the thickness of the NPB layer, which acts as the hole transportation layer. From a thickness of 450 nm till 650 nm was studied, the device efficiencies indeed further improved, but at a thickness of 650 nm, the red emission gradually weakens because lesser holes could penetrate so far.
Figure 4-15 The EL spectrum with different NPB thickness
Table 4-4 EL performance of the devices
Another study is to exanimate the effect of phosphorescent sensitizer fluorescent devices, because we considered that this process might further drive down the material cost and fabrication process.
It has been demonstrated that the internal efficiency of fluorescent can be as high as 100% by using a phosphorescent sensitizer to excite a fluorescent dye through resonant energy transfer between triplet excitons in the phosphor and singlets in the fluorescent dye. Provided a device combines this kind of emission red or yellow with a blue emission, it should be a high-efficiency WOLED. Recently, a WOLED with a maximum luminous efficiency of 6.0 cd/A by using phosphor sensitized fluorescence was reported [52]. In this device, DCJTB, Ir(ppy)3, and NPB were used as fluorescent dye, phosphorescent sensitizer, and blue emitter, respectively. In later reports, a WOLED with a higher maximum of 9.22 cd/A is performed by using more efficient materials DPVBi and rubrene to replace NPB and DCJTB as blue and yellow emitters [66]. Lei et al. also fabricated a WOLED with a maximum current efficiency of 9.2 cd/A, in which blue phosphorescent material FIrpic was used as the
Devices (V) (cd/A) (lm/W) EQE (%) CIE 1931 CRI
NPB 450 6.6 12.2 5.8 6.8 (0.40, 0.43) 80 NPB 550 6.2 13.5 6.8 7.4 (0.39, 0.43) 83 NPB 650 6.4 14.7 7.3 6.8 (0.36, 0.46) ---
sensitizer [49]. Performances of this kind of WOLED are further improved by means of introducing a red phosphorescent emitter as well as exchanging the positions of the co-doped and the blue emissive layers [51]. Phosphor sensitization works as follows: By doping a phosphor at high concentrations ~ 5–10 wt% into a conductive host, both singlet and triplet excitons can transfer onto the phosphor molecule. If the phosphor contains a heavy metal atom, spin orbit coupling transfers all excited states on the phosphor to the radiative triplet manifold. These radiative states can then be readily transferred via the dipole–dipole Fo¨rster process to the radiative singlet state of the fluorophore co-doped with both the host and phosphor molecules. By lightly doping ~1% the fluorophore, hopping from the host triplets to the nonradiative triplet state of the fluorescent molecule is discouraged. In principle, therefore, phosphor sensitization can lead to 100% internal quantum efficiency of OLEDs radiating from the singlet manifold of the fluorescent dopant molecules [50].
Figure 4-16 Devices (Set D)
Figure 4-17 (Set D) The EL spectrum with different DCJTB doping concentrations (left: normalized, right: un-normalized
intensity)
Indeed, this provides a good method for producing white light. A special issue that has not been seen in phosphorescent red emitters is that, a red spectral shift of the DCJTB peak from about 560 nm to 580 nm is observed with increasing DCJTB concentration. This effect is also studied in other reports [67].
From the standard intensity spectrum, we can see that by intensifying the red dopant concentration, the red emission tends to maintain its maximum intensity without intensifying any further, but to the decrease of the green emission.
This device shows the best efficiency with a doping of concentration of 1.2%, reaching 16.7 cd/A, and power efficiency of 8.3 lm/W, the CIE shifted slightly to (0.41, 0.47), but still shows a white emission color. But due to the red emitter is not saturated enough; the CRI value is low, R9 which is a strong red reflector is below 0.
Table 4-5 EL performance of the devices (Set D)
Devices (V) (cd/A) (lm/W) (nits) EQE (%) CIE 1931 CRI
0.3% 6.1 23.7 12.0 2357 7.7 (0.33, 0.52) --- 0.6% 6.2 21.8 11.1 2180 7.4 (0.36, 0.50) --- 0.9% 6.3 19.2 9.5 1924 6.5 (0.37, 0.49) --- 1.2% 6.3 16.7 8.3 1665 6.0 (0.41, 0.47) 63
Figure 4-18 shows the comparison of the five kinds of device structures in its power efficiency. The phosphor-sensitized fluorescence obviously shows the highest efficiency.
Figure 4-18 The comparison of the five kinds of device structures in its power efficiency
4.4 OLED with an external color tuning layer
A few papers introduced the idea of using a color emitting OLED in combination with a down-conversion layer to produce white OLED. We also tried to use this easy fabrication process, and improved the CIE positions by selecting an appropriate concentration dye dissolved in PMMA. According to the concept of complementary wavelengths, two colors added in order to produce white color is calculated as in Figure 4-19.
Figure 4-19 The concept of complementary wavelengths
So if our blue devices have emission intensity near 480 nm, we will then need a complementary wavelength near 580 nm in order to produce white light. We exanimate the effect of the doping concentration of the ECTL dye inside polymethylmethacrylate (PMMA) host, which is considered the most transparent thin film material. It is found that the spectrum tends to red-shift at higher doping condition, but the emission peak near 538 nm is nearly not changed. We estimate that the strongest absorption and emission for the dye is between 0.5wt% and 1.0wt%.
Figure 4-20 The PL emission of the ECTL under different concentration
So for our best choice, the doping of this ECTL into PMMA was achieved by mixing 0.5 to 1.0wt% of the dye with PMMA in a common solvent such as toluene. The well-stirred mixture (using ultrasonic mixing for 15 min) was applied on the substrate by using the doctor blade technique then was dried in air.
Blue devices with ECTL
For testing the performances of the color tuning layer, we fabricated three blue devices as shown in Figure 4-21. Device E-A is a fluorescent
blue emitting OLED having the structure of:
ITO/CFx/NPB/MADN:1%BUBD-1/Alq/BPhen:5%Cs2CO3/Al, while Devices E-B and E-C are phosphorescent blue emitting diodes, with either a ungraded or graded emission layer, respectively. Device E-B has the structure: ITO/CFx /NPB/mCP: 8%FIrpic/BPhen/BPhen:
5%Cs2CO3/Al, while Device E-C have the same structure except with a graded emissive layer: ITO/CFx/NPB/mCP: 5%FIrpic/mCP:
15%FIrpic/mCP: 35%FIrpic/BPhen/BPhen:5%Cs2CO3/Al, the average dopant concentration of the emission layer is controlled at 8% .
15%FIrpic/mCP: 35%FIrpic/BPhen/BPhen:5%Cs2CO3/Al, the average dopant concentration of the emission layer is controlled at 8% .