1. Introduction

An efficient Si light source plays a key role in Si photonics and Si-based optoelectronics integration [1–5]. Recently, researchers combined organic materials with Si to realize a Si-based organic light emitting diode (OLED) [6–13] where bulk Si is usually used as the anode and the substrate, and a top emission structure was adopted. The hole injection of a p-Si anode can be adjusted in a very large range by simply changing the p-Si resistivity to match the electron injection in order to achieve maximum efficiency. However, Si has high absorption in the visible light range, which is detrimental to the efficiency of the OLED. In 2005, Zhu et al. first reported a nanometer-thick (~50 nm) polycrystalline-Si anode OLED with the aim of simplifying the technique process of producing active-matrix OLEDs; besides, even though for emitting blue light the maximum luminance efficiency is 23% lower than that of a control ITO anode OLED, for emitting red and green light the maximum luminance efficiencies are 57% and 20% higher than those of the control ones, respectively [14]. In 2006 we reported a 1.54 μm electroluminescence (EL)from the nanometer-thick polycrystalline p-Si (NTPS) anode OLED and demonstrated that its efficiency is about 20 times higher than that of the control bulk Si anode counterpart [15]. In this work, we report an OLED with a bi-layer composite anode of Ni silicide/ NTPS. The composite anode originates from a bi-layer of Ni/ amorphous p-Si. At certain high temperatures, the Ni layer not only will induce Si crystallization but also will form a Ni silicide layer of appropriate thickness. The Ni/ amorphous p-Si bi-layer thus transfers to a Ni silicide/ NTPS composite anode. The NTPS plays a role in injecting holes into the hole transport layer (HTL), and the Ni silicide layer can reduce sheet resistance of the composite anode and is beneficial to the efficiency of the Si-based OLED because of the much lower resistivity (20~200 μΩcm) of Ni silicide than that of polycrystalline Si. By combing the NTPS/ Ni silicide composite anode with a phosphor organic material and adopting other measures, we achieve a highly efficient phosphor OLED.

2. Experimental

In our composite anode OLED, the Ni silicide/ NTPS was fabricated by magnetron sputtering, followed by metal-induced crystallization [16–18]. The Ni silicide/ NTPS composite anode was fabricated as follows. An ultra-thin amorphous p-Si film (~10 nm) was first deposited onto a Corning 1737 glass substrate (130 μm thick) by magnetron sputtering as an isolation layer, which prevents the impure elements in the glass from diffusing into the composite anode and organics in the subsequent processes. Then a Ni layer was deposited by magnetron sputtering with a Ni target. The thicknesses of the Ni layers adopted in sequential experiments were 0.75, 1, 1.5, 2, 3, and 4 nm. Another ultra-thin amorphous p-Si film (~10 nm) was successively deposited onto the Ni layer by magnetron sputtering with a target of 0.01Ωcm p-Si. Then the glass substrate/ Si isolation layer/ Ni/ amorphous p-Si multilayer was annealed at 540°C for 10 min in an N₂ atmosphere. For the annealed multilayer, x-ray photoelectron spectroscopy (XPS) depth profiling was finished (Axis Ultra, Kratos Analytical Ltd.). The sheet resistance of the annealed multilayer was measured at room temperature by the Hall Effect measurement system (Nanometrics HL5500). Raman spectra were measured with a microzone confocal Raman spectroscope (HORIBA Jobin Yvon, LabRam HR 800) in under-focused measure conditions. High-resolution-transmission electron microscopy (HRTEM) observation was carried out with a Tecnai F30 field-emission-transmission electron microscope. At first, a 150 nm Al mirror was deposited onto the back side of the glass substrate of the annealed multilayer. A 5 nm V₂O₅ film was then evaporated onto the front side of annealed multilayer [19]. The HTL was N, N’-bis-(1-naphthl)-diphenyl-1, 1’-biphenyl- 4, 4’- diamine (NPB). The emitter was bis(2-phenylpyridine) iridium(III) acetylacetonate [(ppy)₂Ir(acac)], which was doped into the host material, 4, 4’-N,N’-dicarbazole- biphenyl (CBP) [20]. The electron transport layer (ETL) was 4, 7-diphenyl-1, 10-phenanthroline (BPhen) [21], and a 15 nm Bphen layer doped with Cs₂CO₃ was used as an electron injection layer and a part of the ETL. The top-emission cathode was the Sm/Au bilayer, on which a 2, 9-dimethyl- 4, 7- diphenyl-1, 10-phenanthroline (BCP) layer was deposited to enhance the outcoupling efficiency [22]. The organic materials and the cathode metals were thermally evaporated successively on the composite film in a chamber with a base pressure of 5 × 10⁻⁴ Pa. The typical deposition rate was ~1 Å/s, monitored by quartz crystal oscillators. The finished OLED, referred to as PhOLED, has the structure of an Al mirror 150 nm/ glass substrate 130 μm/ Si isolation layer ~10 nm/ Ni silicide/ NTPS/ V₂O₅ 5 nm/ NPB 40 nm/ CBP: (ppy)₂Ir(acac) (10 wt%) 40 nm/ Bphen 25 nm/ Bphen: Cs₂CO₃ (mass ratio of 1:1) 15 nm/ Sm 5 nm/ Au 15 nm/ BCP 45 nm. The EL spectrum and luminance of the PhOLED was measured in air by a PR705 spectrometer. The molecule structure of (ppy)₂Ir(acac) and the EL spectrum of the PhOLED are shown in Fig. 1(b) . Current voltage characteristics were measured by a computer-controlled Keithley 2400 Sourcemeter.

 

Fig. 1 (a) Schematic structure of the PhOLED, (b) the electroluminescence spectrum of the PhOLED; the inset is the molecule structure of the phosphorescent emitter (ppy)₂Ir(acac).

Download Full Size | PDF

3. Results and Discussion

For the annealed Ni/ amorphous Si bi-layer, XPS depth profiling was carried out in an etching process with an etching rate of 2–3 nm/ min. The probing depth range of the equipment is about 5 nm. Figure 2(a) shows the XPS spectra of Si (2p) for the annealed bi-layer. Before etching, there are two binding energy peaks of 99.2 eV and 103.1 eV, which should originate from polycrystalline p-Si and SiO₂, respectively. After the 120s and 240s etching, the former binding energy peak shifts to 99.7 eV and 99.8 eV, respectively, which can be explained by the existence of Ni silicides. As shown in [21], the Si (2p) binding energy peaks of silicides are about 99.7 eV. Figure 2(b) shows the Ni (2p) XPS spectra for the annealed bi-layer. For the 0–240s etching, the binding energy peaks locate at constant energy of 853.9 eV, which also indicates the existence of Ni silicide. As shown also in [21], the Ni (2p) binding energy peaks of Ni silicides are at about 853.3 eV. The XPS depth profiling results indicate that at the depth range of 0–5 nm of the annealed bi-layer, polycrystalline p-Si exists, and after 120s etching (4–6 nm) Ni silicide is detected. In the PhOLEDs, the NTPS layer injects holes into the HTL; however, its sheet resistance is very large, estimated to be larger than 1 × 10⁶ Ω/□, and will result in large series resistance, high operating voltage, and low power efficiency. By contrast, the Ni silicide layer has low sheet resistance estimated to be about 100 Ω/□ [16,23], which will reduce series resistance of the Ni silicide/ NTPS composite anode and increase power efficiency of the PhOLED.

 

Fig. 2 XPS spectra of (a) Si (2p) and (b) Ni (2p) for the annealed Ni/ amorphous p-Si bi-layer with an Ni layer thickness of 2 nm.

Download Full Size | PDF

The PhOLEDs are optimized by varying the initial Ni layer thickness in the Ni/ amorphous p-Si bi-layer. For the bi-layer annealed at 540°C for 10 min, the characteristics of sheet resistance and transmissivity at 520 nm versus the initial Ni layer thickness are shown in Fig. 3(a) . Increasing the Ni layer thickness from 0.75 to 4 nm and thus the Ni silicide layer thickness, the sheet resistance decreases from 826 to 102 Ω/□, and the transmissivity at 520 nm decreases from 49% to 32%. As shown in Fig. 3(b), both the current and power efficiencies for the PhOLEDs are functions of the initial Ni layer thickness, and both of the efficiencies reach peak values at the same Ni layer thickness of 2.0 nm.

 

Fig. 3 (a) Sheet resistance and transmissivity at 520 nm of Ni silicide/ NTPS bi-layer versus initial Ni layer thickness and (b) PhOLED efficiencies versus initial Ni layer thickness.

Download Full Size | PDF

Figure 4(a) shows the current voltage and luminance voltage characteristics for the efficiency-optimized PhOLED. Figure 4(b) shows the current efficiency and power efficiency voltage characteristics for the same PhOLED. The highest current efficiency of 1.0 × 10² cd/A occurs at a voltage of 6.2 V, a current of 8.5 μA, and a luminance of 1.2 × 10² cd/m²; the highest power efficiency of 60 lm/W occurs at 4.2 V, 1.1 μA, and 12 cd/m², corresponding to a maximum external quantum efficiency of 26% and a maximum power conversion efficiency of 11%, respectively. As far as we know, the external quantum efficiency is the highest one among the Si-based electroluminescence reported in the literature.

 

Fig. 4 (a) Current–voltage and luminance–voltage curves, (b) current-efficiency voltage and power-efficiency voltage curves for the Ni-layer-thickness optimized PhOLED.

Download Full Size | PDF

In order to explain why there is an efficiency peak with the variation in Ni layer thicknesses, we measure the Raman shift spectra of the annealed Ni/ amorphous Si bi-layer. As shown in Fig. 5(a) , when the Ni layers are thinner than 1.5 nm the Raman shift peaks locate at about 480 cm⁻¹, indicating that most of the amorphous Si layer in the annealed bi-layer is not crystallized. The composite anode with an amorphous Si layer at the top results in poor hole injection and large series resistance. For the annealed bi-layer with an initial 1.5 nm Ni layer the Raman shift peak locates at about 503 cm⁻¹, the signal is weak, and the full width at half-maximum is wide, which probably suggests that in addition to the nanoscale Si crystals there is still a amount of amorphous Si in the annealed bi-layer. For the annealed bi-layer with an initial 2 nm Ni layer, the Raman shift peak still locates at 503 cm⁻¹, but the peak signal is stronger and sharper than that of the 1.5 nm Ni layer case, which suggests that a large part of Si has been crystallized to nanoscale Si crystals. A partial HRTEM top-view image for the Ni silicide/ NTPS composite anode with the initial Ni layer thickness of 2 nm is shown in Fig. 5(b), where some nanoscale Si crystals can be seen. When the Ni layer thickness is in the range of 3–4 nm, as shown in Fig. 5(a), the Raman shift peak of 210 cm⁻¹ appears, which can be attribute to Ni silicide [24], and the Raman shift peak of 503 cm⁻¹ vanishes. The Ni silicide is not a suitable anode material for high-efficiency organic light emitting. Thus, either the initial Ni layer that is thinner or thicker than 2 nm leads to PhOLEDs with lower efficiencies compared with that of the Ni-layer thickness-optimized PhOLED.

 

Fig. 5 (a) Raman shift spectra for the Ni silicide/ NTPS bi-layers with different initial Ni layer thicknesses of 0.75, 1, 1.5, 2, 3, and 4 nm; (b) a partial HRTEM top-view image for the Ni silicide/ NTPS composite anode with an initial Ni layer thickness of 2 nm.

Download Full Size | PDF

4. Conclusion

In summary, we have fabricated a high-efficiency phosphorescent organic light-emitting diode with a Ni silicide/ polycrystalline p-Si composite anode. After optimizing the thickness of the initial Ni layer, the phosphorescent organic light-emitting diode achieves a maximum external quantum efficiency of 26% and a maximum power conversion efficiency of 11%, respectively. As far as we know, the external quantum efficiency is the highest one among Si-based electroluminescence reported in the literature.