The role of defects in organic image sensors for green photodiode

Controlling defect states in a buffer layer for organic photo devices is one of the vital factors which have great influence on the device performance. Defect states in silicon oxynitride (SiOxNy) buffer layer for organic photo devices can be controlled by introducing appropriate dopant materials. We performed ab initio simulations to identify the effect on doping SiOxNy with carbon (C), boron (B), and phosphorous (P) atoms. The results unveil that hole defects in the SiOxNy layer diminish with the phosphorous doping. Based on the simulation results, we fabricate the small molecule organic photodetector (OPD) including the phosphorous-doped SiOxNy buffer layer and the active film of blended naphthalene-based donor and C60 acceptor molecules, which shows excellent enhancement in the external quantum efficiency (EQE). The results of our charge-based deep level transient spectroscopy (Q-DLTS) measurements confirmed that the EQE enhancement originates from the decrease of the hole traps induced by the reduced hole defects. The method of controlling the defect states in SiOxNy buffer layers by the doping can be used to improve the performance in various organic photo devices.

develop new buffer layers whose band structure can be controlled with ease. In our previous paper, the silicon oxynitride (SiO x N y , SiON) layer has been introduced as a good buffer layer for OPDs because its band structure can be freely tuned by adjusting the atomic concentration of oxygen (x) and nitrogen (y) 19 . By controlling the x and y, the band structure of SiO x N y buffer layer can be aligned well to HOMO and LUMO levels of bulk heterojuction (BHJ) films. In the OPD devices using the SiO x N y buffer layer, the detectivity and the thermal stability were improved as well as the DC decreased 19 . Here, we propose a novel approach to improve the OPD performance by controlling the defect states within the energy band gap. Our ab initio simulations demonstrate that the defect states of SiO x N y buffer layer are controlled by doping with various atoms such as C, B, and P. In addition, we could achieve excellent enhancement in EQE of our OPD device including the SiO x N y buffer layer which had reduced defect states by doping with P atoms. The energy band gap of the buffer layer is a crucial factor which has great influence on the device performance, especially, EQE and DC. As shown in the simulation results ( Fig. 1) using the SETFOS software tools, we calculated the device performance (EQE, DC) depending on the characteristics of buffer layers, electron blocking layer (EBL) and hole blocking layer (HBL), in the device structure of ITO/EBL/OPD/HBL/ITO. As shown in Fig. 1a, EQE increases as the hole barrier in EBL or the electron barrier in HBL gets lower, i.e. HOMO of EBL or LUMO of HBL decreases. In addition, as shown in Fig. 1b, DC decreases as the electron blocking of EBL increases, i.e. LUMO of EBL decreases, while it is not significantly influenced on the hole blocking of HBL, i.e. HOMO of HBL.
The other key factor for the OPD device performance is to adjust the density of the defect states within the band gap of the buffer layer 10 . In most cases, defect states can act as a trap for electrons or holes, which interrupts their transport and decreases EQE. Therefore, one of the crucial issues for developing high performance OPDs is to control the defect states within the bandgap of the buffer layer. To verify the change of defect states within the bandgap of SiO x N y by the doping, we carried out ab initio calculations using the Vienna ab initio Simulation Package (VASP) and molecular dynamics simulations for amorphous phase modeling 20 . The atomic and the electronic structure of amorphous phase were obtained by calculating the structural properties such as radial distribution function and angle distribution function by the melt-quench method of molecular dynamics simulations and the projector-augmented-wave (PAW)-generalized gradient approximation (GGA) potential 21,22 The electrical properties including inverse participation ratio, density of states, wave function and so on were also calculated. The energy cutoff was chosen to be 400 eV and 2 × 2 × 1 Monkhorst-Pack k-point meshes were used.  www.nature.com/scientificreports/ The possible defect states in amorphous SiN were also calculated. The results indicate that k-defects (Si-dangling bond), Si-Si homopolar bonds, under-coordinated N atoms, and distorted bonds are the main defects in SiN. Particularly, dangling bonds of Si or N, and unstable Si-Si bonds induce localized defect states near the valence band. The distorted bonds create the defect states near the conduction band, having an influence on the leakage current.
In addition, over-coordinate Si atoms (5-fold Si, 6-fold Si) make defect levels near the valance band and over-coordinated N atoms (4 fold N) make localized defect levels near the conduction band ( Supplementary  Fig. S1).
As shown in Fig. 2, to verify the doping effects on SiON, we analyzed the geometric and the electronic structure of SiON with changing the dopant materials (C, B, P). First, in the case of changing the C content from 2 to 6 at.%, C atoms reduced the Si-Si bonds by forming the bond with Si and reduced also the distorted bonds by inducing tetrahedral bonds. As a result, the addition of C atoms of 2~6 at.% induced the reduction of defects. However, the addition of C atoms more than 6% increased defects by forming C-C clusters which induced the defect states in the band gap. In addition, we performed the calculations for adding B and P atoms of 6 at.%. In the case of B atoms, they formed the bond with N and induced the reduced C-Si or N-Si bonds, which resulted in the increased Si-Si homopolar bonds. This increased the band gap and defect states. In the case of P atoms, they mainly formed the bonds with Si atoms and contributed to the reduction of Si-dangling bonds in the system, which induced the reduction of defects. The results of these calculations for the three kinds of atoms (C, B, and P) confirmed that P-doped SiON has the lowest defect density, while B doping increases the defect density at maximum level (Fig. 2, Table 1).
Based on the simulation results, doping of the SiON films with C and P atoms were carried out and their electrical properties were characterized. (Supplementary Fig. S2).
As shown in Fig. S2, the leakage current increases in the C-doped sample. However, the P-doped sample shows the lower breakdown voltage by comparing with the reference un-doped SiON sample. The results indicate the reduction of defects by the P-doping.

Results and Discussion
To verify the effect on the OPD device performance by P-doping in the SiO x N y buffer layer, we fabricated the OPD devices with the P-doped SiO x Ny buffer layers. The cross-sectional transmission electron microscopy (TEM) images for our OPD devices and their elemental maps by energy-dispersive X-ray spectroscopy (EDS) are shown in Fig. 3. Each OPD device consist of a buffer layer of SiO x N y (7 nm) and an active layer (160 nm) of 1:1 blended naphthalene-based donor and C60 acceptor molecules, as depicted in Fig. 4a-c. Two kinds of buffer layers,  (Fig. 4b) has a pure SiO x N y buffer layer without doping, while the P-doping sample (Fig. 4c) was applied for the SiO x N y buffer layer of the doped sample.
After the preparation of pure and P-doped SiO x N y buffer layers, respectively, a 160-nm-thick organic bulk hetero junction (BHJ) layer, which is the 1:1 blend film of 'naphthalene based donor' and C60, was formed on the SiO x N y layers, followed by the deposition of a 7-nm-thick ITO capping layer. The 'naphthalene based donor' , shown in Fig. 4a, is a novel push-pull-structured organic semiconducting material, which is 5-(naphthalen-1-yl (phenyl) amino) selenophen-2-yl) methylene) -1H-indene-1,3 (2 H) -dione, with the absorption properties selective to green-light and it has a LUMO and HOMO values as a 3.6 eV and 5.5 eV as respectively.
The device performances of the OPD samples such as EQE and DC were characterized and were shown in Fig. 4d,e. The EQE of the sample with a P-doped SiO x N y buffer layer increased by ~3.4%, while the DC characteristic of the P-doped sample is similar to that of the reference one. In Fig. 4e, it is notable that both samples with SiO x N y layers show the improved DC performance regardless of P-doping, in comparison with the OPD sample without a SiO x N y buffer layer.
To elucidate the mechanism of enhanced EQE characteristic by the P-doping of SiO x N y buffer layer, we performed the charge based deep level transient spectroscopy (Q-DLTS) measurements for the two OPD samples, in which the temperature is kept constant while the frequency is scanned. This method has been particularly successful to measure the current transient for organic materials [23][24][25] .
In good agreement with our simulation results, the hole defect states near the energy level, which is 0.3 eV higher than the valence band energy level, remarkably decreases by the P-doping of SiO x N y buffer layer, as shown in Fig. 5a. Figure 5b depicts the mechanism for the enhanced EQE by the reduction of hole defect states. The holes and electrons separated from excitons should be transported to the ITO electrodes without being disturbed to obtain the high performance in EQE. However, the hole defect states in the SiO x N y buffer layer can act as the hole traps which interrupt the transport of holes, as shown in Fig. 5b.

Conclusions
In this study, for the purpose of controlling the defect states of SiO x N y buffer layer for OPDs, we performed the ab initio simulations for the doping of SiO x N y with C, B, and P. It was revealed that the hole defects in SiO x N y could be reduced by P doping. In the device performance measurements, the OPD with a P-doped SiO x N y buffer layer shows the improved EQE. Based on our QDLTS measurements, it was experimentally confirmed that the EQE enhancement originates from the reduced hole traps induced by the reduced hole defects. The method to control the defect states in SiO x N y buffer layers by the doping can be used to improve the performance in various organic photo devices.

Methods
Characterizations. Transmission electron microscopy (TEM) imaging and energy-dispersive X-ray (EDX) analysis were performed on a FEI Tecnai Osiris at 200 kV. This microscope was equipped with a high brightness X-FEG gun and silicon drift Super-X EDX detectors and Bruker Esprit acquisition software.
The composition analysis and valance band offset measurements was carried out by X-ray photoelectron spectroscopy (XPS, PHI Quantera II Scanning XPS Microprobe).
The HR-REELS investigation was conducted by Auger electron spectroscopy (AES; PHI 4700 system) with attached monochromatic electron gun (LK 1000). The primary energy of monochromatic electrons for the HR-REELS measurement was 0.3 keV and the full width at half maximum (FWHM) of the elastic peaks of 0.3 keV was 50 meV.
Q-DLTS measurements were performed with a PhysTech FT 1030 DLTS system. Temperature scans were made between 300 and 350 K, at a heating rate of 5 K/min. Samples were placed in the liquid helium cryostat. The pulse height, filling pulse width, and pulse period width were 0.4 V, 10 ms, and 10 ms, respectively. Here, reverse bias is −0.8 V to −0.4 V, and forward bias is 0.4 V to 0.8 V, respectively. The activation energy, capture cross-section, and concentration of traps were calculated using an Arrhenius plot.