1. INTRODUCTION

The image-sensor market was valued at 15.84 billion USD in 2019, and it is predicted to grow to 24.49 billion by 2025 [1–4] owing to the ongoing development of smartphones, security cameras, self-driving automobiles, and camcorders. A breakthrough in complementary metal-oxide–semiconductor (CMOS) image sensors comprising silicon photodiodes (PDs) was the scaling down of the pixel size to less than 1 µm to obtain high-resolution output images [2,5,6]; however, the small number of photons generated in dark scenes lead to low sensitivity and difficulty in object recognition. To overcome this drawback, image sensors comprising three-dimensional vertically stacked light-absorbing layers have been proposed [7–9]. This approach has been successfully used to enhance the sensitivity of image sensors while retaining the effective area. In our previous studies, we developed organic–silicon hybrid image sensors by superposing green-light-absorbing transparent organic PDs (OPDs) on a Si PD with red/blue filters [10–15]. The green-selective light-absorbing organic layer used in these image sensors considerably reduced cross talk between the pixels used for color recognition. Despite their judicious potential, electrical characteristics of OPDs should be further improved for vivid image recognition, distinct image in darkness, and high frame rate.

Over the past few years, OPDs have gained significant attention owing to their unique features that are promising for various applications, such as blood oxygen saturation testing [16,17], optical switch [18], image sensing [10–15,19], smart displays [20], window-integrated electronics [21], and self-powered wearable sensors [22]. Appropriately designed electrode and organic absorbers [20,23,24] with narrow and wide spectral responsivity, wavelength tunability (including the infrared region), mechanical flexibility, and large-area processability are promising candidates for application in future optoelectronic devices [25]. Among the standard parameters of OPDs, the specific detectivity ($D^*$) is a significant and useful figure of merit; it is directly associated with the signal-to-noise ratio (SNR), accuracy, and sensitivity of OPDs for low light-power detection. Therefore, it is important to develop devices with high external quantum efficiency (EQE; $\eta$) and low dark current density (${J_D}$) with low noise spectral density (${S_n}$).

A solution-processed OPD with a high detectivity of ${3} \times {{10}^{13}}\;{\rm cm}\;{{\rm Hz}^{1/2}}\;{{\rm W}^{- 1}}$ was fabricated by screen printing for large-area production [26]; this detectivity is comparable to those of conventional Si PDs (${6 - 7} \times {{10}^{13}}\;{\rm cm} \cdot {{\rm Hz}^{1/2}} \cdot {{\rm W}^{- 1}}$) [27]. Most solution-processed OPDs are composed of relatively thick (${\gt}{400}\;{\rm nm}$) light-absorbing layers to reduce the dark current density and require a high operational voltage to maximize the photogenerated charges and prevent charge recombination [26,28,29]. To meet the CMOS fab-compatibility benchmark, vacuum-processing of OPDs with solvent-free pixel formation during patterning and electrode deposition processes is desirable as it produces light absorbers with a surface that is pinhole-free, smooth, and has a uniform thickness, which ensures continuous high-quality production. The highest detectivity of a vacuum-processed OPD reported to date is ${8} \times {{10}^{13}}\;{\rm cm}\;{{\rm Hz}^{1/2}}{{\cdot {\rm W}}^{- 1}}$, which was achieved using novel ${\rm donor} {-} \pi {-} {\rm acceptor}$ molecules combined with two fused-type heterocyclic donors and an electron-accepting unit [14].

However, it is difficult to fabricate high-detectivity transparent OPDs with both high EQE and low dark current and noise spectral density (${S_n}$). Although the formation of stepped energy levels at the interface could maximize the charge collection efficiency and minimize the charge recombination, this is generally difficult to achieve [25]. In addition, the buffer layer should have a smooth morphology and an amorphous atomic structure to prevent current leakage and realize high transparency. Therefore, it is imperative to develop a high-quality buffer layer with the desired internal energy level distribution and physicochemical stability to fabricate high-detectivity OPDs suitable for practical applications.

In this study, we developed fab-compatible transparent green-sensitive organic photodiodes (TG-OPDs) with a dark-current-based high light detectivity of ${4.1} \times {{10}^{14}}\;{{ {\rm cm}\cdot {\rm Hz}}^{1/2}}{{\cdot {\rm W}}^{- 1}}$ (at a wavelength of 550 nm under 3 V) and an EQE of ${\gt}{75}\%$ (under 3 V) when applied in a CMOS image sensor. Owing to the introduction of a bathocuproine $({\rm BCP}):{{\rm C}_{60}}$ mixed buffer layer as the electron transporting layer (ETL), the OPDs showed exceptional PD characteristics, with an extremely low dark current (below ${{10}^{- 11}}\;{\rm A} \cdot {{\rm cm}^{- 2}}$) and a high rectification ratio spanning 10 orders of magnitude, which is outstanding results compared to our previous reports without using mixed ETLs [13,14]. The extraordinary electrical characteristics of these OPDs were attributed to the merits of the mixed buffer layer, including its smooth morphology, desirable energy level modulation, and function as an optical spacer. In addition, the OPDs showed robust operational stability under intense heat (temperatures above 150°C) for 2 h and long-term operational stability at 85°C for 30 d. These findings are expected to facilitate the industrialization of OPDs as the main component of imaging modules and optoelectronic sensors for applications such as photoplethysmography, fingerprint recognition, and proximity sensors.

2. EXPERIMENTAL SECTION

A. Material Preparation

First, 2-((5-(10H-phenoselenazin-10-yl)selenophen-2-yl)methylene)-1H-cyclopenta[b]naphthalene-1,3(2H)-dione (PSe-Se-CPND) was synthesized as follows: 2.00 g (4.96 mmol) of 5-(10H-phenoselenazin-10-yl)selenophene-2-carbaldehyde was suspended in ethanol, and 1.07 g (5.46 mmol) of 1H-cyclopenta[b]naphthalene-1,3(2H)-dione was added. The mixture was reacted at 50°C for 2 h. Then, the reaction mixture was cooled to room temperature (27°C) and vacuum filtered. The solid collected by vacuum filtration was purified by column chromatography over silica using methylene chloride and ethyl acetate as the eluent and recrystallized from chloroform/hexane to obtain a dark red solid (yield 83%). ${}^1{\rm H}$ nuclear magnetic resonance ppm (DMSO-$d{6}$, dimethyl sulfoxide-$d{6}$): 8.34 (s, 1H), 8.32 (s,1 H), 8.27 (s, 1H), 8.24–8.16 (m, 3H), 7.98(dd, 2H), 7.88(dd, 2H), 7.71 (m, 2H), 7.61 (t, 2H), 7.45 (t, 2H), 6.61 (d, 1H). The 5-(10H-phenoselenazin-10-yl)selenophene-2-carbaldehyde was synthesized in-house. Cyclopenta[b]naphthalene-1,3(2H)-dione was purchased from Hanchem Corp. Bathocuproine (2,9-dimethyl-4,7-diphenyl-1,10-phenanthroline) and ${{\rm C}_{60}}$ were purchased from Sigma-Aldrich and Frontier Carbon Corp., respectively, and used as received.

B. Device Fabrication

The OPDs with a typical structure were fabricated on indium tin oxide (ITO)-coated glass substrates by sequentially depositing a 100-nm-thick organic bulk heterojunction (BHJ) layer of ${\rm PSe} {-} {\rm Se} {-} {\rm CPND}:{{\rm C}_{60}}$ with different 15-nm-thick ETLs and without an ETL. All film layers were thermally evaporated under high vacuum (${\lt}{{10}^{- 7}}\;{\rm Torr}$). A thin ITO layer was used as the top electrode on the organic BHJ to fabricate the stacked organic CMOS image sensor because of its high transparency (${\gt}{95}\%$) and low sheet resistance (${\sim} 100\;{\Omega / {\rm sq}}$). The anode (150 nm) and cathode (7 nm) ITO electrodes were deposited under different gas pressure conditions and their work-functions exhibit 4.5 and 4.8 eV, confirmed by ultraviolet photoelectron spectroscopy (UPS). To minimize plasma damage to the organic films, a face target sputtering system was used. For encapsulation, a 40-nm-thick ${{\rm Al}_2}{{\rm O}_3}$ layer was deposited by thermal atomic layer deposition; trimethylaluminum was used as the precursor, and de-ionized water was used as the oxidizing agent. The deposition was carried out at 100°C with a growth rate of ${0.55}\;{{{\mathop{\rm A}\limits^\circ}}}{\rm s}^{- 1}$ using Ar as the carrier gas. Finally, the devices were encapsulated in antireflective glass (transmittance ${\gt}{95}\%$). The active pixel size of the PD was ${0.04}\;{{\rm cm}^2}$.

C. Organic Film Characterization and Analysis

Atomic force microscopy (AFM; Veeco) images of the 100-nm-thick as-deposited ETLs were obtained. Cross-sectional transmission electron microscopy (TEM; Tecnai G2ST) images of the organic thin films were obtained using a focused ion beam (Helios NanoLab 400S) milling technique. The highest occupied molecular orbital (HOMO) levels of the organic thin films were measured using an AC-2 photoelectron spectrophotometer (Hitachi High Tech). The lowest unoccupied molecular orbital (LUMO) levels of the films were determined from their optical bandgaps calculated from the edge of their ultraviolet-visible (UV-Vis) absorption spectra (UV-3600 Plus, Shimadzu).

 

Fig. 1. Design of transparent green-sensitive OPDs for CMOS image sensors. (a) Schematic of vertically stacked CMOS image sensors using green organic image sensors. The TG-OPD is placed on the top of silicon PDs with blue and red filters. (b) Device structure of the TG-OPD. (c) Chemical structure of the BHJ and ETL. For the BHJ layer, PSe-Se-CPND was used as the electron-donor (p-type) material, and ${{\rm C}_{60}}$ was used as the electron-acceptor (n-type) material. For the ETL layer, a mixture of BCP and ${{\rm C}_{60}}$ was used. (d) CIE1970 color space used to calculate CRI with TG-OPD. The AM 1.5 G spectrum is included as the reference light source. (e) AM 1.5 G photon flux, transmitted photon fluxes through TG-OPD, and photopic response function $V(\lambda)$. The photon flux was used for AVT measurements.

Download Full Size | PDF

D. Device Characterization and Analysis

The current density–voltage (J–V) characteristics of the OPDs were measured using a semiconductor parameter analyzer (Keithley K4200 with preamp). The bottom ITO electrode (anode) was grounded, and the top ITO electrode (cathode) was biased, respectively. The light was illuminated on the top electrode surface. The photocurrent characteristics of the devices were evaluated under illumination by a green laser diode ($\lambda_{\max} = {553}\;{\rm nm}$) at various light intensities ranging from 0.001 to ${1}\;{\rm mW} \cdot {{\rm cm}^{- 2}}$. The EQE of the devices was measured using an instrumental setup illuminated by monochromatic light generated by an ozone-free xenon lamp with a chopper frequency of 30 Hz. The long-term on/off current and EQE values were measured in a light (AM 1.5G) and humidity (40%) controlled chamber using a multi-channel solar-cell I–V tester (K3600 McScience, Korea) with a current limit of 1 pA. The monochromatic light intensity was calibrated using a Si PD (Hamamatsu S1337). The corresponding $R$ (${\rm A} \cdot {{\rm W}^{- 1}}$), which is defined as the ratio of the generated photocurrent (A) to the incident light power (W), was calculated from the EQE as $R = {\rm EQE}/hv$, where $hv$ is the incident photon energy (eV). The frequency response of the devices was evaluated using a measurement setup consisting of an oscilloscope (MDO3034, Tektronix) and a function generator (LeCroy Wavestation 2012) with a green laser diode (${\lambda _{{\rm max}}} = {553}\;{\rm nm}$, ${\rm power} = {1}\;{{ {\rm mW}\cdot {\rm cm}}^{- 2}}$). The impedance analysis of the devices was carried out on a standard DC potentiostat with electrochemical impedance spectroscopy (EIS) functionality (SP-200). The photodiode noise spectral density was recorded by a power spectrum analyzer (HP35670A) connected to the device through a preamplifier (SRS 570). The device detectivity is $D^* = R{(A)^{0.5}}/{S_n}$, where $A$ is the effective photodetector area (${0.04}\;{{\rm cm}^2}$ for the devices reported here). The SNR10 is defined as the scene illumination level for a signal-to-noise ratio of 10 on the luminance channel calculated on a 18% gray patch under a 3200 K light source after color correction and white balance, with a $f$-number of 2.8 and 15 fps. SNR10 is evaluated from EQE values of Si PD and TG-OPD according to characterization protocols reported previously [30–32].

3. RESULTS AND DISCUSSION

Figure 1(a) shows the multi-stack design of the TG-OPDs in Si hybrid CMOS image sensors. The unique stacked design proposed in this study doubled the photosensitivity of the OPDs positioned on top of a Si PD with blue/red filters and fully transparent ITO electrodes on both sides [Fig. 1(b)]. The light transmitted through the TG-OPD and filters was absorbed by the bottom Si PD to distinguish between the blue and red components. As shown in Fig. 1(c), as a BHJ layer, PSe-Se-CPND was used as the electron-donor (p-type) material, and fullerene (${{\rm C}_{60}}$) was used as the electron-acceptor (n-type) material at a volumetric ratio of 1:1. After formation of the BHJ layer, different ETLs were deposited. BCP is a widely used organic ETL material that has a deep HOMO level of 6.4 eV and a LUMO level of 2.8 eV, which results in a high optical transparency that is responsible for the large energy bandgap and excellent exciton blocking properties. However, devices composed of a single BCP layer have a relatively low device yield owing to the clustering of BCP molecules and their rough surface [33–35]. In addition, ${{\rm C}_{60}}$ efficiently blocks redundant holes because of its deep HOMO level (6.3 eV), while its high electron mobility enhances the charge transport of the generated electrons. However, its narrow energy bandgap (2.3 eV) in the visible region disrupts the light penetration of the BHJ layer. In principle, the electron–hole pairs (excitons) generated by photon absorption diffuse into the donor (PSe-Se-CPND)–acceptor (${{\rm C}_{60}}$) interface, producing charge transfer excitons. To engineer the energy levels suitable for the desired electron charge transport and collection at the electrode, the LUMO level of the ETL (${{\rm LUMO}_{{\rm ETL}}}$) should be lower than that of the BHJ (${{\rm LUMO}_{{\rm BHJ}}}$) layer. Therefore, we modified the ETLs by mixing two materials (${\rm BCP}:{{\rm C}_{60}}$) for application as a single buffer layer (Figs. S1 and S2) to modulate their optical/electrical performance and obtain a ${{\rm LUMO}_{{\rm ETL}}}$ that lies between ${{\rm LUMO}_{{\rm BHJ}}}$ and the work-function of the cathode. To evaluate the potential of the TG-OPDs for application in practical CMOS image sensors, their color rendering indices (CRIs) and average visible transmittances (AVTs) were measured [Figs. 1(d) and 1(e)] according to characterization protocols reported previously [36]. The CRI is an important factor to quantitatively measure the color expression from an object in comparison with a natural light source, and the AVT represents the transparency of the entire device including both the bottom and top electrodes. We obtained u’ and v’ values of 0.264 and 0.420, respectively, and an AVT of ${\sim}{32}\%$, indicating that the fabricated full device had a red-violet color with good transparency. In addition, the device with the mixed buffer layer had higher CRI and AVT values than the other devices investigated here (Table S1).

 

Fig. 2. Physicochemical analysis of the ${\rm BCP}:{{\rm C}_{60}}$ ETL. (a) Absorbance spectra of the BCP (black), BCP:C60 (red), and ${{\rm C}_{60}}$ (blue) layers. The inset shows the Tauc plot obtained from the absorption spectra in (a) to determine the optical energy bandgaps. (b) Photographs of the ETLs under ambient light. (c) Estimated HOMO energy levels of the ETLs. AFM images of the (d) BCP and ${\rm BCP}:{{\rm C}_{60}}$ layers. TEM images of the (e) ${\rm BCP}:{{\rm C}_{60}}$ and ${{\rm C}_{60}}$ layers.

Download Full Size | PDF

To investigate the effect of the mixed layer on the device performance, the physicochemical properties of the different types of ETLs used were analyzed, as shown in Fig. 2. Figure 2(a) shows the UV-Vis spectra of the ETLs. The 100-nm-thick single BCP layer showed no absorptivity; however, the single ${{\rm C}_{60}}$ layer showed strong absorption (${\lambda _{{\rm max}}} = {447}\;{\rm nm}$) in the visible region (400–700 nm). Interestingly, the absorption intensity of the ${\rm BCP}:{{\rm C}_{60}}$ (1:1 volumetric ratio) mixed buffer layer was less than half of that of the ${{\rm C}_{60}}$ single layer. The Tauc plot [inset in Fig. 2(a)] of a material is a useful tool for determining its optical bandgap from the absorption spectrum using the following Tauc relation:

 

Fig. 3. Comparison of the electrical characteristics of the OPDs with different ETLs. (a) Band energy diagram of the OPD with the mixed ETL. The BCP and ${\rm BCP}:{{\rm C}_{60}}$ layers showed the LUMO energies of 2.8 and 3.9 eV, respectively. The anode (150 nm) and cathode (7 nm) electrodes exhibit different work-functions of 4.5 and 4.8 eV due to different process condition and thickness, confirmed by UPS analysis. (b) Current density–voltage ($J - V$) characteristics. The data were collected under dark condition. At 0 V, the currents are not equal to zero due to the real values are under the limit of measurement setup. (c) Nyquist plots of the OPDs in the high-frequency region showing the real (${Z^\prime}$) and imaginary (${Z^{{\prime \prime}}}$) parts of the impedance. (d) EQE of the OPDs at the operating voltage of 3 V. (e) Thermal stability of the OPDs as a function of the annealing time at 150°C. EQE at $\lambda = {550}\;{\rm nm}$ and an operating voltage of 3 V.

Download Full Size | PDF

To investigate the effect of the buffer layers on the electrical characteristics of the fabricated optoelectronic devices, we fabricated OPDs with various ETLs and without an ETL (Fig. 3). The proposed energy band diagram is shown in Fig. 3(a). Most interestingly, the devices with the ${\rm BCP}:{{\rm C}_{60}}$ ETL showed an unprecedented high rectification ratio of ${\sim}{10}$ orders of magnitude (${J_{{\rm on}\:{\rm at}\: - 3\:{\rm V}}}/{J_{{\rm off}\,{\rm at} \,+ 3\,{\rm V}}}$), as shown in Fig. 3(b). The dark currents of OPDs with different ETLs exhibit very low current density as low as ${{10}^{- 11}}\;{\rm A} \cdot {{\rm cm}^{- 2}}$ at 3 V (off state) due to efficient blocking the hole injection from deep HOMO level [39,40] of densified and smooth ETLs (Fig. 2). High-frequency Nyquist plots obtained from EIS analysis of the OPDs revealed that the OPD with the ${\rm BCP}:{{\rm C}_{60}}$ ETL had the lowest charge transfer resistance (${R_{\textit{ct}}}$ of 1.8 kohm) among all the devices investigated [Fig. 3(c)], indicating that charge transfer characteristics compared to without a buffer layer (1.9 kohm). In the forward (negative bias on cathode electrode) region, the OPD with the ${{\rm C}_{60}}$ ETL showed the lowest current density owing to an interfacial energy barrier between the BHJ and ETL (3.7 and 4.0 eV of LUMO level, Figs. S1 and S2). In addition to their excellent $J - V$ and diode characteristics, these devices had a high EQE of ${\sim}{76}\%$ at 550 nm [Fig. 3(d)], which is the highest EQE reported to date for green-sensitive OPDs (Figs. S5 and S6, Table S2). This demonstrates the potential of using a stacked PD device architecture for achieving high S/N ratio and sensitivity. The equivalent absorptance values of the different OPDs at 550 nm are shown in Fig. S7. The high internal quantum efficiency of the OPD with the ${\rm BCP}:{{\rm C}_{60}}$ ETL was attributed to its (i) low interfacial contact resistance (Fig. S8), (ii) conformal contact with both the bottom organic layer and top ITO electrode [Figs. 2(d) and 3(b)], and (iii) desirable stepped energy level alignment (Fig. S2). The other important requirement for the practical application of OPDs is to maintain operational stability during high-temperature annealing because, for full-chip fabrication, the fabrication of OPDs on Si wafers requires sequential thermal fabrication steps, such as passivation layer deposition, microlens formation, and packaging. Figure 3(e) shows the measured EQE of the OPDs at 3 V as a function of the annealing time at 150°C. The OPD with the ${\rm BCP}:{{\rm C}_{60}}$ mixed buffer layer showed good thermal stability up to 150°C for 2 h. In contrast, the EQE of the OPD with the BCP layer first decreased rapidly (within 5 min) at 150°C and then gradually decreased to 46.6% after 2 h of annealing. This was attributed to the crystallization of the BCP layer [34], which degraded the charge transport interface (Figs. S9 and S10). In contrast, the mixed buffer layer maintained a smooth surface, thus suppressing the crystallization of BCP efficiently and increasing the long-term stability of the device (Fig. S11).

Figure 4 shows the results of the electrical analysis of the TG-OPD with the ${\rm BCP}:{{\rm C}_{60}}$ mixed buffer layer and the properties of the OPD in CMOS image sensors. Figure 4(a) shows the $J {-} V$ characteristics of the OPD at different light intensities at a wavelength of 540 nm. The device showed specific light responses and reliable brightness classification under very weak lighting conditions (as low as ${0.1}\;\unicode{x00B5} {\rm W/cm}^2$). Figure 4(b) shows the current density of the OPD as a function of the light intensity at 3 V. The linear dynamic range (LDR) of the TG-OPD was obtained as follows [41]:

 

Fig. 4. Electrical characteristics of the OPDs with the mixed ${\rm BCP}:{{\rm C}_{60}}$ ETL. (a) Current density–voltage ($J {-}s V$) characteristics of OPDs under dark conditions and illumination at $\lambda = {553}\;{\rm nm}$ with light intensities ranging from ${{10}^{- 4}}$ to ${1}\;{\rm mW} \cdot {{\rm cm}^{- 2}}$. (b) Photocurrent density at an operating voltage of 3 V as a function of the light intensity (${J_{\textit{ph}}} - V$ characteristics) extracted from (a). The dotted line indicates a power-law fit with a slope ($\alpha$) of 0.99. (c) Spectral responsivity (black) and dark- and noise-current-based specific detectivity as a function of wavelength for a light intensity of ${100}\;{{ {\rm mW}\cdot {\rm cm} }^{- 2}}$. (d) Normalized photoresponse under modulated light illumination (${\rm light}\;{\rm intensity} = {1}\;{{\rm mW\;cm}^{- 2}}$, wavelength $\lambda = {553}\;{\rm nm}$). The inset shows static behavior of the OPDs under pulsed light illumination (${\rm light}\;{\rm intensity} = {1}\;{\rm mW}\;{{\rm cm}^{- 2}}$, wavelength, $\lambda = {553}\;{\rm nm}$) at 1 kHz. (e) EQE of green OPD and blue and red Si PDs. The EQE of the green-sensitive OPD was evaluated at the operating voltage of 3 V. The EQEs of the blue and red responses were evaluated from those of the underlying Si PDs stacked with a color filter and the OPD.

Download Full Size | PDF

Figure 4(c) shows the spectral responsivity ($R$) and dark and noise-current-based specific detectivity ($D^*$). $D_{{\rm Noise}}^*$ includes flicker or (${1/}f$) noise which can be significant under specific operating conditions. A maximum responsivity of ${0.34}\;{\rm A} \cdot {{\rm W}^{- 1}}$ was obtained under an applied reverse bias of 3 V (Fig. S12). The dark and noise-current-based $D^*$ of the OPD was calculated as follows [3]:

For application in time-resolved image sensors, such as snapshot and video recording sensors, an OPD must have a high-speed transient photoresponse. The ${-}{3}\;{\rm dB}$ cutoff in the photoresponse test indicates that the maximum frequency was obtained when the amplitude of the initial photocurrent decreased to $1/\sqrt {2}$ from the initial value. Interestingly, a faster photoresponse of 35 kHz was observed; this value is more than 3 times higher than that reported previously for a device without a ${\rm BCP}:{{\rm C}_{60}}$ layer [14]. This improved photoresponse of the OPD was attributed to its lowest charge transfer resistance. A sharp increase (${\tau _{{\rm rise}}} = {16}\;{\unicode{x00B5}\rm s}$) and a decrease (${\tau _{{\rm rise}}} = {10}\;{\unicode{x00B5}\rm s}$) in the output signal were observed when the incident light was turned on and off. The results discussed thus far indicate that the TG-OPD fabricated in this study is suitable for application in sequential image-capturing devices, such as mobile phones, camcorders, driving recorders, and surveillance cameras, which require an image-capturing speed of more than 30 fps [45]. Another important factor affecting the practical application of OPDs in image-capturing devices is uniform color detection to provide precise chromaticity and vivid tone. The EQEs of multi-stack organic–Si hybrid CMOS image sensors are shown in Fig. 4(e). The most promising feature of our OPD is its narrow and selective green-light absorption (${\rm FWHM} = {100}\;{\rm nm}$), which provided excellent transmission of blue and red light. Allowing for 10% light reflection from the encapsulation glass, the EQE of red- and blue-sensitive Si PDs approaches 50%. Consequently, the OPD device, including the color filters, had only 40% light attenuation, which is better than the performance of the previously reported X2-structured CMOS image sensors [10]. SNR10 is a key parameter of image sensors for the production of clear, high-quality images under low light conditions [30]; this metric indicates the minimum illuminance (lx) intensity when the detected light output signal is 10 times more intense than the noise signal. The device with the ${\rm BCP}:{{\rm C}_{60}}$ buffer layer showed the lowest SNR10 value of 82.1 lx among all the investigated devices (Fig. S13).

Figure 5 shows the state-of-the-art performance of previously reported OPDs [26,41] and their figures of merit. Except for the OPDs using a reflective metal electrode, where the responsivity was measured in the far visible [47–50] and infrared regions [51] (${\lt}{700}\;{\rm nm}$), our TG-OPDs with a ${\rm BCP}:{{\rm C}_{60}}$ ETL buffer layer are the only devices that simultaneously show both low dark current density (${\lt}{{10}^{- 11}}\;{\rm A} \cdot {{\rm cm}^{- 2}}$) under 3 V and high responsivity (${0.34}\;{\rm A} \cdot {{\rm W}^{- 1}}$) at a wavelength of 550 nm under 3 V [Fig. 5(a)]. The OPDs prepared in this study showed a good EQE of 76% and an extremely high detectivity compared to those of the reported OPDs [Fig. 5(b), Table S2].

 

Fig. 5. State-of-the-art OPD performance. (a) Responsivity versus minimum dark current density curves of the OPDs fabricated in this study and those reported previously. (b) Specific detectivity ($D^*$) versus maximum EQE curves of the OPDs fabricated in this study and those reported previously. For the specific comparison, we included dark- and noise-current-based detectivity: $D_{{\rm Dark}}^*$ (blue), $D_{{\rm Noise}}^*$ (red). See Supplement 1 for details. We also recommend [41,46] for more information.

Download Full Size | PDF

In conclusion, we developed a transparent and fab-compatible green-light-sensitive OPD suitable for organic–silicon hybrid CMOS image sensors. The ${\rm BCP}:{{\rm C}_{60}}$ buffer layer, which was used as the ETL, substantially improved the electrical characteristics of the OPD by (i) minimizing the optical loss (Fig. S14), (ii) modulating the energy level, and (iii) providing a smooth morphology. In addition, it positively affected the thermal and long-term operational stability of the OPD. The results obtained in this study are expected to promote the industrialization of see-through OPDs for various sensor applications, such as wearable devices, fingerprint on displays, and proximity sensors.