Figure 1b displays the current density versus voltage (J-V) characteristics in the dark and under illumination of the Al2O3-PM-OPD and control. It is witnessed that the dark current density is reduced significantly under either a forward bias or a reverse bias after the ALD Al2O3 treatment. In particular, the dark J of the Al2O3-PM-OPD operated under a forward bias of 19 V is five orders of magnitude lower than that of the control device, and that of the Al2O3-PM-OPD operated under a reverse bias of -19 V is one order of magnitude lower than that of the control device. As noted from the photocurrent density versus voltage curves, the introduction of ALD Al2O3 weakly reduces the photocurrent under a reverse bias, while the photocurrent response under a forward bias is higher than that under a reverse bias. The light-to-dark-current density ratio (Jlight/Jdark) of the Al2O3-PM-OPD reaches 5.30×104 under 19 V bias. This is undoubtedly phenomenal as it is around two magnitudes superior to that of control measured at -19 V bias (5.77×102).
To understand the mechanism of improvement in photodetection performance of PM-OPDs, we carry out the energy band diagram analysis firstly. Figure 1c shows the energy level diagram of the control PM-OPD which has already been clarified in literatures.24 The control device embraces a cathode Schottky junction and an anode Ohmic contact, because its dark J-V exhibits a unilateral conduction property. We measured the work function of the HTL and the donor P3HT film from the average surface potentials by the Kelvin Probe Force Microscopy (KPFM) technique (see methods in supporting information). The work function of P3HT was measured to be 5.000 eV, close to that of PEDOT:PSS (5.043 eV), which produces an Ohmic contact at the anode side and further smooth hole injection under a forward bias, as displayed in Supplementary Fig. 1a. Therefore, the control device failed to invoke any distinguishable photocurrent responses at 19 V, as displayed in Fig. 1b. Adjacent to the Al cathode, a downward band bending Schottky junction is built up, so that the hole injection is blocked under a reverse bias; see Supplementary Fig. 1b. Accordingly, the accumulation of photogenerated electrons close to the cathode can narrow the cathode Schottky barrier, so that plenty of holes can now tunnel into the control circuit as illustrated in Supplementary Fig. 1c. For the control device, it is disappointing that the anode Ohmic contact leads thermally activated carriers along with the interfacial carriers flow smoothly from the active layer into the external circuit with bias, inducing a relatively large dark current density, thus somewhat low Jlight/Jdark. For the same reason, the detectivity along with the weak-light detection ability of the control device is limited.
Unlike the control device, the Al2O3-PM-OPD possesses a significantly suppressed dark current under forward bias, suggesting that the anode contact appears to be a Schottky junction. To verify this suggestion, the work functions of PEDOT:PSS surfaces treated by ALD Al2O3 deposition of different thicknesses were measured, as displayed in Fig. 2a. With the increase of the Al2O3 thickness from 0 to 1.2 nm, the work function of the treated PEDOT:PSS film first decreases and then increases, and the lowest work function of 4.4 eV is produced at 0.8 nm Al2O3. It verifies that the incorporation of a 0.8 nm thick ALD Al2O3 interlayer leads to a downward band bending Schottky junction in the vicinity of the anode side (referred to anode Schottky junction) as shown in Fig. 1d. In addition to the cathode Schottky junction, Al2O3-PM-OPDs can be regarded as back-to-back dual Schottky junction device. The newly formed anode Schottky junction is effective in rejecting holes from the external circuit into the device under a forward bias, as seen in Fig. 1e. According to Fig. 2a, the barrier height of anode Schottky junction with a 0.8 nm thick Al2O3 reaches the highest value, therefore the corresponding dark current is suppressed to the most extent, as shown in Supplementary Fig. 2.
The variation of the work function of PEDOT:PSS/Al2O3 surface indicates that certain chemical reaction must take place at the PEDOT:PSS surface after ALD Al2O3 treatment. First, Raman spectroscopy measurements were carried out to verify our conjecture. It can be seen clearly from Fig. 2b that the Raman spectrum of PEDOT:PSS surface has experienced obvious change after 0.8 nm Al2O3 incorporation. Besides of the shift of Raman peaks, a series of additional peaks are produced. For instance, close to the major Raman peak at 1432 cm-1 (peak 4), peaks locating at 1425 cm-1 (peak 3), 1448 cm-1 (peak 5), and 1471 cm-1 (peak 6) are observed according to the envelope fitting results shown in Supplementary Fig. 3. Then, the high-resolution X-ray photoelectron spectroscopy (XPS) measurements of various PEDOT:PSS surfaces were conducted to identify the change in binding energy spectrum of particular elements. As clearly seen from Fig. 2c, the PEDOT:PSS surface presents apparent envelope changes in its C1s XPS peak after introducing 0.8 nm ALD Al2O3. It is obvious that the C1s peak of PEDOT:PSS locates at 284.6 eV, which is ascribed to the C–C bond; whereas the C1s peak of PEDOT:PSS/Al2O3 (x = 0.8 nm) situates at 285.2 eV, which can be assigned to the C = C bond.25,26 In addition, the shoulder at 286.5 eV in PEDOT:PSS that represents the C-O-C group gets broadened, indicating that the carbon atoms have been slightly oxidized, possibly forming C-O-Al complexes.27 The new peak at 289 eV in the C1s spectrum may be due to the formation of carbonyl/carboxyl groups on PEDOT arising from Al2O3 oxidation 28. Supplementary Fig. 4 further indicates that when the ALD Al2O3 thickness either decreases or increases from the optimal value of 0.8 nm, the differences between the C1s XPS major peak position and width of the treated PEDOT:PSS surface and its pristine become less apparent. This provides evidence that we have undertaken the atomic-level manipulation of chemical reaction in PM-OPDs. We have also measured the Fourier transform infrared spectroscopy (FT-IR) spectra of PEDOT:PSS surfaces without or with ALD Al2O3 (x = 0.8 nm) treatment as shown in Fig. 2d. It is seen that FT-IR peaks of the functional groups, including C-O-C, C-C, and C = C groups 29–31, suffer from noticable changes either in peak positions or peak envelopes. The Raman, XPS along with FT-IR spectroscopies all denote that chemical bond variations take place at the PEDOT:PSS surface after ALD Al2O3 treatment. The observed bond variations origin from the chemical reaction between the PEDOT:PSS chemical groups and Al2O3 bottom surface. On the one hand, the Al2O3 product of ALD deposition can react with PEDOT:PSS chemical groups.32 On the other hand, it is also possible that the feed reactants of H2O and Al(CH3)3 in the process of ALD deposition join in the reaction with PEDOT:PSS chemical groups. By analyzing the C1s XPS peak of Al2O3 layers with different thicknesses prepared on Si wafer, as shown in Supplementary Fig. 5, we found that there is C element residue in ALD Al2O3, confirming that the reaction between Al(CH3)3 and H2O during ALD deposition is incomplete.
Upon Raman spectroscopy, we also confirm that there are interfacial chemical reactions between PEDOT:PSS and PCBM acceptor. As shown in Fig. 2e (middle and bottom subplots), drastic changes of characteristic peaks appear in the Raman spectrum of PEDOT:PSS/PCBM relative to that of individual PCBM. Instead, there are negligible changes observed for the Raman spectrum of PEDOT:PSS/P3HT relative to that of individual P3HT (Supplementary Fig. 6).33 It is believed that the reaction between PEDOT:PSS and PCBM can induce some dangling bonds and interfacial defects, which exacerbates the leakage current. From the top subplot in Fig. 2e, it is seen that, with the Al2O3 interlayer, the Raman spectrum of PEDOT:PSS/PCBM is very close to that of individual PCBM, manifesting that the Al2O3 interlayer also plays the role of inhibiting the chemical reaction between PEDOT:PSS and PCBM. Through the extinction spectrum characterizations (Supplementary Fig. 7), it is also witnessed that more remarkable differences between films of PEDOT:PSS/P3HT:PCBM with and without Al2O3 are witnessed with the acceptor-rich photoactive blend, relative to the donor-rich photoactive blend. This further reveals that the interfacial reaction takes place between PEDOT:PSS and PCBM, and the introduction of Al2O3 affords an isolation among them, yielding an approximately ideal interface with negligible reaction-induced surface defects.
Besides, the effect of Al2O3 interlayer on surface defects of PM-OPDs could be further addressed by capacitance characteristic in the dark. The capacitance versus voltage (C-V) and capacitance versus frequency (C-F) characteristics, as shown in Supplementary Fig. 8, demonstrate apparent decreases of capacitances after Al2O3 incorporation, indicating that the accumulation of carriers at PEDOT:PSS/P3HT:PCBM interface has been significantly attenuated. Furthermore, the calculated trap density of state (tDOS) versus frequency or demarcation energy (Eω) of different PM-OPDs are displayed in Fig. 2f (see methods in Supporting Information). The low, moderate and high frequency regimes are labelled by Regime I, II and III, respectively; among them, the interface-defect-induced trap states locate in Regime II.34 Fig. 2f reveals that the amount of interface defects of PM-OPD reduces significantly with 0.8 nm Al2O3 interlayer, confirming the suppression of interfacial chemical reaction between PEDOT:PSS and PCBM acceptor. The reduced interface-defect-induced trap states provide a greatly increased shunt resistance of the device, as shown by the frequency dependent impedances of Al2O3-PM-OPDs and control in Supplementary Fig. 9. All these evidence jointly explain the origin of dark current suppression of the proposed Al2O3-PM-OPDs with respect to the control. Additionally, it is found that the role of Al2O3 incorporation for suppressing dark current by constructing Schottky junction was also effective in improving the performance of OPDs comprising of other photoactive blends (see Supplementary Fig. 10).
To elucidate the working mechanism of Al2O3-PM-OPDs under illumination, the energy band diagrams of the device are discussed. Under a forward bias, the downward band bending of the anode Schottky junction functioned as a switch for hole tunneling. The hole tunneling results from the accumulation of photogenerated electrons towards the Al2O3/P3HT:PC71BM interface, and thus the narrowing of anode Schottky junction, as seen in Fig. 1f. Since the applied bias also lowers the cathode Schottky junction, the injected carriers under illumination can flow smoothly from the blend towards the cathode.35,36 It is seen from Fig. 1b that the photocurrent response of Al2O3-PM-OPD under a forward bias of 19 V was much higher than its response under a reverse bias of -19 V. Moreover, compared with the control device that can only work under a reverse bias, Al2O3-PM-OPD shows much superior responses under a forward bias. In Fig. 3a, the responsivity spectrum characteristic of Al2O3-PM-OPD under 19 V bias further manifests improvement of around one magnitude over a broad wavelength range from 300 nm to 700 nm, compared with that of control device operated under a reverse of -19 V. The wavelength dependent power densities (Pin) of illumination are shown in Supplementary Fig. 11. The improvement in photo responsivity must be related to the accumulated photogenerated electrons close to the electrode, which is positively correlated with the quantity of trapped electrons responsible for inducing the following carrier injection. Figure 3b shows the simulated distribution of photogenerated carriers versus wavelength for Al2O3-PM-OPD. It is noted that a 0.8 nm thick Al2O3 is too thin to induce any change in absorption distribution, so the simulated distribution of control is the same as that of Al2O3-PM-OPD. As seen from Fig. 3b, light absorption in the P3HT:PC71BM region close to the anode is much stronger than that adjacent to the cathode. This explains why Al2O3-PM-OPD demonstrates superior photocurrent response under a forward bias relative to a reverse bias. Besides, the control PM-OPD that can only respond to photo under a reverse bias produced inferior performances because of the same reason. Overall, the Al2O3 interlayer not only played a decisive role in suppressing the dark current, it also shifted hole injection to the electrode that possesses increased photogenerated carriers in its vicinity for strengthening the photo responsivity.
One also sees from Fig. 3a that the photo responsivities in the ultraviolet regime increase more significantly than those in the visible regime. This phenomenon might arise from the difference in photo energies at different wavelengths. The ultraviolet photos have energies higher than twice of the bandgap energy of P3HT, so one absorbed ultraviolet photon can give rise to more than one photogenerated electron at high applied bias. Under this premise, the peak responsivity takes place at 315 nm, abiding by the rule of optical resonance. As shown by the photogenerated carrier distribution in Fig. 3b, at the wavelength of 315 nm, the optical cavity resonance induces enhanced absorption close to the anode (at position range from 150 nm to 210 nm), facilitating the production of trapped electrons and the subsequent hole injection from the anode. Besides, both Al2O3-PM-OPD and control can respond stably to pulsed illumination as shown in Fig. 3c. The pulsed illumination was supplied by 532 nm laser with a Pin of 6.4 µW/cm2. It can be seen clearly that Al2O3-PM-OPD displays much stronger signal relative to control, in agreement with the afore-displayed responsivity characteristics in Fig. 3a. The derived response speeds of Al2O3-PM-OPD and control are close to each other, as shown in Fig. 3d. The rise time and fall time of Al2O3-PM-OPD are respectively 187 ms and 375 ms, in consistent with the performances of other reported PM-OPDs with P3HT:PCBM (100:1) blend.36
As the dark current of PM-OPDs was reduced significantly with Al2O3 incorporation, it was expected that the dynamic range (DR) would also be favorably improved. DR is derived from the ratio of the highest and the lowest detectable illumination power densities (Phigh/Plow) within the sublinear response range, where the slope (β) of the light intensity-dependent photocurrent double logarithmic plot is lower than 1 due to its photomultiplication property (see methods in supporting information). 37–39 DRs of Al2O3-PM-OPD (19 V) and control (-19 V) were acquired under illumination by a 532 nm continuous laser with varied power densities (Pin) (Fig. 4a-b). As expected, we successfully broadened the DR of PM-OPD from 60 dB to 116 dB by incorporating a 0.8 nm thick Al2O3 interlayer. Figure 4c indicates that with the decrease of Pin, the EQE and responsivity of Al2O3-PM-OPD continues to increase; the weak-light detection limit of Al2O3-PM-OPD reaches a remarkable level of 2.5 nW/cm2. At 2.5 nW/cm2, Al2O3-PM-OPD possesses record high EQE and responsivity respectively of 4.31×108% and 1.85×106 A/W, outperforming all other polymer-based PM-OPDs (see Supplementary Table 1). By contrast, the weak-light detection limit of control is 25 times inferior, that is around 64 nW/cm2, as shown in Fig. 4c, and the corresponding EQE and responsivity are respectively 5.24×105% and 2.25 ×103 A/W, more than 800 times inferior to those of Al2O3-PM-OPD under the same illumination.
Besides, it is also seen from Fig. 4c that EQE (or responsivity) of control keeps almost constant with the decrease of Pin, in accordance with the results in other reports.23 Such a difference between Al2O3-PM-OPD and control might arise from their differences in trapped states, which play the role of trapping electrons for inducing the holes injection effect. It is deduced that the control device might have more trapped states close to cathode that are not fully filled under illumination at its DR, the ratio of trapped electrons to the incident photos is unchanged with the decease of Pin, thereby resulting in a constant EQE. Differently, in Al2O3-PM-OPD, the trapped states close to the anode might be fully filled under illumination throughout its DR, the ratio of trapped electrons to the incident photos could increase with the decrease of Pin, and consequently its EQE increases.
Based on the noise current characteristics, the detectivity (D*) versus Pin of different PM-OPDs were estimated. Supplementary Fig. 12a-b show the measured noise currents (in) of Al2O3-PM-OPD (19 V) and control (-19 V) (see methods in supporting information). The results suggest that the noise currents of the fabricated PM-OPDs markedly depend on frequency, indicating that in are mainly dominated by frequency-dependent noise current. The power density dependent detectivity are determined from the corresponding responsivity characteristic and the measured in, as displayed in Fig. 4d. The detectivity of Al2O3-PM-OPD also increases with the decrease of Pin, similar to what happens to EQE and responsivity. At 2.5 nW/cm2, i.e., the lowest detectable illumination power density, Al2O3-PM-OPD possesses a detectivity as high as 1.13×1015 Jones, which is more than two orders of magnitude higher than that of control (2.28×1012 Jones, at Pin of 64 nW/cm2). It is mentioned that the detectivity of Al2O3-PM-OPD is comparable to the champion detectivity (3.4×1015 Jones) of all reported polymer-based PM-OPDs which belonged to the device made of PVK:ZnO blend22 (Supplementary Table 1). Nevertheless, the responsivities of the reported PVK:ZnO based PM-OPD were ~ 103 A/W, close to the characteristic of our control, more than two orders of magnitude lower than that of the Al2O3-PM-OPDs. When both responsivity and detectivity characteristics are under evaluation, the demonstrated Al2O3-PM-OPD takes the possession of the overall record high detection performances.
In order to further verify the improvement of weak-light detection performance of Al2O3-PM-OPD, we constructed the corresponding image sensor with the setup of scanning process that is schematically represented in Fig. 5a. A bias was applied to the bottom anode line and top cathode line of a selected pixel, and its photocurrent was then readout. The layout design of crossed electrode lines is shown in Fig. 5b, and the physical map of the top cathode lines can be found in Fig. 5c. The photoactive area of the overlap between the ITO anode and Al cathode was 2 mm×2 mm. A “C” shape shadow mask was subsequently placed in contact with the image sensor for the following imaging process. Figure 5d shows the image of the Al2O3-PM-OPD (19 V) under 532 nm laser illumination with varied Pin (in subplots from left to right, Pin are 6400 nW/cm2, 640 nW/cm2, 64 nW/cm2, and 2.5 nW/cm2, respectively). It is witnessed that the “C” symbol can be fairly resolved by Al2O3-PM-OPD on the condition of either high power density or low power density. The image captured by sensor made of Al2O3-PM-OPD at Pin of 2.5 nW/cm2 is so well resolved that the contrast is even better than that of control at Pin of 6400 nW/cm2 (Fig. 5e, the left most subplot, bias of -19 V). As it is seen from Fig. 5e, the image property of control sensor is very disappointing with the contrast getting inferior with the decrease of Pin. Under illumination of 2.5 nW/cm2, the “C” symbol completely fades away in the output electric signal. All these outcomes attest to the outstanding weak-light imaging capability of the proposed Al2O3-PM-OPD.