1. Introduction

Organic semiconductors are receiving increasing attention in the field of solar energy because of their advantageous properties, including high absorption coefficients, mechanical flexibility, solution processability and low material costs [1–3]. Over the past years, the power conversion efficiencies (PCEs) of organic-based solar cells have been improved rapidly; the current certified world record is as high as 17.5% for organic photovoltaic devices (OPVDs) [4]. Their PCEs, however, still require further improvements to strive with inorganic photovoltaic devices. One of the obstacles to highly efficient OPVDs is the low mobilities of organic semiconductors; the thickness of the photoactive layers is usually limited in order to increase the charge collection efficiency, resulting in incomplete absorption of the photons from solar irradiation. To increase the absorption efficiencies, many optical approaches, such as incorporating diffraction gratings [5] and inserting optical spacers [6,7], have been proposed for improving light harvesting efficiencies. Among these methods, incorporating surface plasmon resonance of metal nanostructures in the OPVDs has been demonstrated as an effective strategy for improving the PCEs [8–14]. Surface plasmon (SP) is collective excitation of free electrons in response to the irradiated electromagnetic (EM) waves. They can be excited either propagating SP polaritons at metal−dielectric interfaces or as confined states in metal nanoparticles (NPs); the latter is also called localized surface plasmon resonance (LSPR) [15]. SPs are capable to enhance the electromagnetic field near the surface, thereby increasing the optical absorption of the photoactive layer [15]. Alternatively, the strong light scattering by the plasmonic nanostructures can elongate the optical path length, thereby increasing the absorption efficiencies.

Most metal nanoparticles used for improving the performance of OPVDs are noble metals, including Ag, Au and Pt [10]. Despite noticeable success has been made using these materials, the high cost impedes their deployment in practical solar modules. Other than the noble metals, NPs made of copper (Cu) also exhibit strong LSPRs in the visible wavelength range. Cu, however, is more abundant and less expensive. In this respect, Cu is an ideal choice for plasmonic materials. Cu NPs indeed have high potential as the substitute for Au and Ag NPs for emerging electronic applications [16]. Unfortunately, Cu NPs are readily oxidized, leading to low chemical stability.

To improve the stability of Cu NPs, Liu et al. employed surface-modified Cu NPs to prevent oxidation and improved the efficiencies of OPVDs by blending them into the buffer layer, poly(3,4-ethylenedioxythiophene) : polystyrenesulfonate (PEDOT:PSS) [17]. Later, Shen et al. prepared uniformly distributed Cu NPs directly on the WO₃ buffer layers of PSCs; the plasmonic effects of the Cu NPs also improved the device performance [18]. Recently, we proposed a fabrication protocol in which the Cu NPs were placed at the cathode interface. Therefore, the fabrication process could be performed in an inert environment, thereby preventing the oxidation of Cu NPs [19]. In this work, we prepared bimetallic alloy NPs made of Au and Cu elements for improving the performance of OPVDs. The addition of Au component in Cu NPs enhanced the stability of the NPs. We, therefore, could directly blend them with the anodic buffer solution of PEDOT:PSS. The OPVDs prepared with the Au-Cu alloy NPs exhibited higher PCE values. Furthermore, indoor photovoltaics (iPVs) have regained increasing attention recently [3,19–27]. Emerging photovoltaics, including OPVDs [3,18–22], perovskite photovoltaic devices [23–25], and dye-sensitized solar cells (DSSCs) [25], exhibit high PCE values under indoor lighting conditions. These highly performed photovoltaics are potential power suppliers for internet-of-things (IoT) applications [25,27]. Therefore, we also found that the OPVDs prepared with the Au-Cu alloy NPs exhibited higher power output under illumination from artificial lighting sources. As a result, we anticipate that the Au-Cu alloy NPs open up a new pathway to improving the performance of OPVDs both for indoor and outdoor applications.

2. Experiments

2.1 Synthesis of Au-Cu alloy nanoparticles

To synthesis Au-Cu alloy NPs, 45 mg of hexadecylamine (HAD) was dissolved into 5 mL of de-ionized (DI) water and was stirred continuously overnight at room temperature [28]. Then, aqueous solutions containing CuCl₂.2H₂O (0.30 mL; 0.1 M), HAuCl₄ (0.28 mL; 1.5M) and glycine (0.3 mL; 1.5 M) were mixed with the above HAD solution and the resulting solution was heated with a hot plate (∼130°C) with continuous stirring for 45 min. We could observe that the color of the solution changed from kelly green to dark brown. After the reaction, the solution was centrifuged (8000 rpm) and washed twice with DI water to remove excess precursor, HDA and glycine. Finally, the alloy NPs were dried through lyophilization.

2.2 Device fabrication

The OPVDs were fabricated on indium-tin-oxide (ITO)-coated glass substrates [11]. The substrates were precleaned using DI water, acetone, and isopropyl alcohol sequentially under sonication and were dried overnight at 100°C. Right before device fabrication, the substrates were exposed to UV ozone treatment for 15 min. An anodic buffer layer of PEDOT:PSS was deposited through spin coating onto the ITO substrates followed by thermal annealing at 150°C for 15 min. The solution of photoactive polymer blends, consisting of poly[4,8-bis(5-(2-ethylhexyl) thiophen-2-yl) benzo[1,2-b;4,5-b’] dithiophene-2,6-diyl-alt-(4-(2-ethylhexyl)-3-fluorothieno[3,4-b] thiophene-)-2-carboxylate-2-6-diyl)] (PTB7-Th) and [6,6]-Phenyl-C71-butyric acid methyl ester (PC₇₁BM), was prepared by dissolving them in chlorobenzene (CB) at a weight ratio of 1:2; the total concentration of the solid contents was 22 mg mL⁻¹. Meanwhile, 1,8-Diiodooctane (DIO; 3% v/v) was also added as the additive for improving the polymer morphology. The solution was spin coated on top of the buffer layer and thermal annealed at 120°C for 10 min, leading to an optimal thickness of ∼105 nm. Finally, device was completed by depositing Ca (30 nm) and Ag (80 nm) through thermal evaporation under a vacuum of 6×10⁻⁶ torr. The active area of the fabricated device was 10 mm². For the devices containing Au-Cu NPs, various amounts of NPs were dispersed into the PEDOT:PSS buffer solutions. The thickness was almost unchanged after the use of the NPs; the thickness of the PEDOT:PSS layer was 30 ± 3 nm for all the devices. Other fabrication procedures for devices containing the alloy NPs were similar to those described above. For each experimental condition, at least five devices were fabricated.

2.3 Device characterizations

The photocurrent response and current density-voltage (J-V) characteristics of the devices were measured under illumination from a Thermal Oriel simulator (AM1.5G) using a Keithley 2400 source meter. The illumination power density was calibrated using a Hamamatsu Si-photodiode equipped a KG-5 filter. To characterize the device performance under the illumination of artifical light sources, a whitle light-emitng diode (WLED) (SY 674, Sheng Yih Technologies), or a fluorescent tube (FT) (TL5, Philips) was used. The illuminance of the artificial light sources was measured using a light meter (TES Electrical Electronic Corp., 1339R). The external quantum efficiency (EQE) spectra were obtained from a measurement system (Enli Technology) equipped with a quartz tungsten halogen (QTH) light source, an optical chopper, a monochromator, and a calibrated Si-photodetector. The Au-Cu NPs were also characterized using a Field Emission Gun Transmission Electron Microscope (FE-TEM) JEOL, JEM-2100F equipped with an energy-dispersive X-ray spectroscopy (EDS) detector.

3. Results and discussion

Au-Cu NPs were synthesized using an aqueous phase synthesis following the previous literature [28]. The molar ratio between Au and Cu was kept at 1:1 and HAD and glycine were used as the capping agent and the reductant, respectively. Figure 1(a) and 1(b) represent the TEM images of the Au-Cu alloy NPs. The NPs were relatively spherical. We observed some NPs aggregated but some NPs were also isolated due to the capping effects of HAD. The high-resolution TEM image showed in Fig. 1(b) revealed a typical single alloy NP. The size was very close to 15.6 nm. The distribution histogram of the size of the NPs is displayed in Fig. 1(c); they are relatively monodispered and the average size was 15 nm.

 

Fig. 1. Characterization of the Au-Cu alloy NPs. (a) the TEM image of the NPs; (b) the high-resolution TEM image of a single alloy NP;(c) the size distribution of the alloy NPs; (d) the absorption spectrum of the Au-Cu alloy NPs dispersed in water.

Download Full Size | PDF

To probe the distribution of Au and Cu atoms in the NPs, elemental mapping was conducted and the EDS spectra of the as-synthesized alloy NPs are shown in Fig. 2(a-c). The images clearly indicated that both elements were successful incorporated into the NPs and the two elements distribute in the NPs evenly. No clear element segregation was observed. Figure 1(d) shows the absorption spectrum of the Au-Cu NPs dispersed in water. We could clearly observe a LSPR peak at 525 nm. Becasue most Au and Cu NPs exhibit LSPR in the visible spectral regime [10,16–19], the peak location of the Au-Cu alloy NPs was reaonable. In order to understand the LSPR properties of the alloy NPs, we simulated the extinction cross sections of the Au and Cu NPs with similar diameters using Mie theory [19,29,30]; the resulting spectra are displayed in Fig. 2(d). Assuming the diameters of both kinds of NPs are 15 nm, the peaks of the simulated spectra located at 510 and 560 nm in water (refractive index, n = 1.33) for Au and Cu NPs, respectively. Because the optical constants for the metal alloys are not systematic and reliable, it is hard to predict the exact LSPR behaviors. However, because the LSPR peak of the alloy NPs (525 nm) located between the two wavelength extremes, the simulation result could confirm the formation of alloy NPs.

 

Fig. 2. The EDS spectra of two representative Au-Cu alloy NPs. (a) the TEM image; (b,c) the corresponding EDS elemental mapping of the NPs; the (b) green and (c) red regions are refer to the elements of Cu and Au, respectively. (d) the calculated extinction cross sections of the Au and Cu NPs with similar diameters using Mie theory; the absorption spectrum of the Au-Cu alloy NPs dispersed in water is also displayed for comparison.

Download Full Size | PDF

To demonstrate the LSPR effects of Au-Cu alloy NPs on the photovoltaic performances, we blended the NPs directly into the PEDOT: PSS buffer layers [11,31]; the results are displayed in Fig. 3. We adopted a polymer blend, consisting of PTB7-Th:PC₇₁BM as the photoactive layer. From the J-V characteristics shown in Fig. 3(a), we can see the reference device, in which the NPs were not embedded, exhibited typical photovoltaic response, with an open-circuit voltage (V_oc) of 0.79 V, a short-circuit current (J_sc) of 12.52 mA cm⁻², and a fill factor (FF) of 0.63, resulting in a PCE value of 6.38%. After the alloy NPs were blended into the buffer layers, the photocurrents were clearly improved. At the optimized concentration of NPs (20%, v/v), J_sc was increased to 12.92 mA cm⁻² and the FF was also slightly increased to 0.67. While the value of V_oc was not changed, the PCE was improved to 7.06%; all the photovoltaic parameters are summarized in Table S1 (Supplement 1). Apparently, the enhancement in the device performance are originate from the LSPR effects of the Au-Cu NPs.

 

Fig. 3. J-V characterization of the PSCs obtained before and after incorporation of the Au-Cu alloy NPs under illumination at 1 sun (100 mW cm⁻¹, AM 1.5G); (b) The corresponding EQE spectra of the PSCs obtained before and after incorporation of Cu NPs. The inset shows the differences in EQE values at various wavelengths between the two spectra (0% and 20%). The absorption spectrum of the NPs is also displayed for comparison.

Download Full Size | PDF

Figure 3(b) represents the corresponding external quantum efficiency (EQE) spectra of the devices revealed in Fig. 3(a). The general shape of the spectra of the plasmonic devices was nearly the same as that of the standard device. The integrated photocurrents calculated form the curves were 10.83, 11.09, 11.64, and 11.05 mA cm⁻² for the reference device and the devices containing 10, 20 and 30% NPs, respectively. The photocurrent resutls generally followed the trend of the J_sc values shown in Fig. 3(a); however, a spectral mismatch existed, resulting in lower integrated photocurrents. We also compared the increases in the EQE values (ΔEQE) after introducing the Au-Cu alloy NPs with the absorption spectrum of the NPs [the inset to Fig. 3(b)]. From the figure, we can see the EQE enhancements cover a wide spectral range. In addition to the spectral range from 400 to 600 nm, which could be originated from the optical LSPR effects, we suspect that electrical effects may also existed, leading to wavelength-independent enhancement [32]. The interface between the active layer and the PEDOT:PSS could be modified so that the route for holes to the anode was changed [32,33]. Further, as shown in Fig. 1(a), some of the NPs are adjoined by each other; such larger aggregates might also lead to stronger scattering, resulting in featureless spectrum of device enhancement. Therefore, the incorporation of the alloy NPs into PSCs somehow results in broadband efficiency enhancement.

To understand the dynamics of the LSPR effects of the alloy NPs, we measured the photocurrent density (J_ph) as a function of effective applied voltage (V_eff) and calculated the maximum exciton generation (G_max) and exciton dissociation probability P(E, T) of the OPVDs; Fig. 4 reveals the data of the results and calculation for the reference device and the one prepared with 20% NP solutions [31,34]. J_ph was equal to the difference between the current density under illumination (J_L) and the current density measured in the dark (J_D) (i.e., J_ph = J_L – J_D). V_eff was determined as V_eff = V₀ − V_a, where V₀ is the voltage at which J_ph = 0 and V_a is the applied bias voltage. For both devices, J_ph increased linearly with V_eff at the lower bias region (<0.3 V) and saturated when V_eff was sufficiently large enough (i.e., V_eff > 0.6 V). At both regimes, the device prepared with alloy NPs exhibited higher J_ph values, implying that more photogenerated excitons in the photoactive layers. We, therefore, considered this could be due to the alloy NPs increased the absorption ability of the OPVD, thereby obtaining more excitons and higher photocurrents.

 

Fig. 4. The properties of photocurrents of OPVDs. (a) Photocurrent density (J_ph) as a function of effective applied voltage (V_eff) for both the reference and plasmonic devices. The inset displays the J_L, J_D and J_ph as a function of bias, whereas the arrows indicate the values of V_oc and the compensation voltage (V₀), respectively, for both devices. (b) Exciton dissociation probability [P(E,T)] as a function of V_eff for both OPVDs.

Download Full Size | PDF

Further, under the saturation condition, all excitons generated in the device could be dissociated and contributed to the photocurrent owing to the high electric potential, which could overcome the high exciton binding energy in organic semiconductors. We noted that the J_ph of the plasmonic device reached saturation regime earlier at a lower bias than the one prepared without alloy NPs, suggesting that more favorable exciton dissociation and charge extraction capability of the plasmonic device. Because the saturated photocurrent (J_sat) is equal to eG_maxL, where e is the electronic charge and L is the thickness of active layer (L = 105 nm). the maximum exciton generation rate (G_max) can be calculated from results of J_sat [Fig. 4(a)] [34]. The values of G_max for the standard and plasmonic device are 1.026×10²⁸ m⁻³ s⁻¹ (J_sat = 156.2 A m⁻²) and 1.052× 10²⁸ m⁻³ s⁻¹ (J_sat = 159.5 A m⁻²), respectively. Because G_max expresses the photon absorption ability, the enhancement in G_max for the plasmonic device implied the maximum absorption of incident photons was increased. On the other hand, J_sat can be also express using Eq. (1),

Therefore, P(E, T) can be calculated from the plot of the normalized photocurrent density (J_ph /J_sat) with respect to V_eff [Fig. 4(b)]. From the figure, we observed the probabilities of exciton dissociation values under short-circuit condition (V_eff = 0) were increased from 77.8% to 87.7% after the use of alloy NPs. The higher probability of exciton dissociation suggested the excitation of LSPR also benefited the dissociation of excitons into free charge carriers [31].

The LSPR effects of the Au-Cu NPs on the indoor performance of the OPVDs were also investigated for evaluating their potential for use under dim-light indoor lighting conditions. Two indoor light sources, a FT and an inorganic WLED with color temperatures of 6500 K and 5000 K, respectively, were used. The J-V curves of the OPVDs measured at 200 lux and 500 lux are displayed in Fig. 5. We notice that the performance of the plasmonic device was enhanced under both indoor lighting conditions. In brief, under illumination from the FT, the values of V_oc, J_sc and FF for the reference device were 0.60 V, 28.1 µA cm⁻² and 0.46, respectively, resulting in a PCE of 12.0% at 200 lux. After integrating NPs in the anodic buffer layer, the J_sc and FF were increased to 30 µA cm⁻² and 0.56 respectively, leading to an improved PCE of 14.9%. At 500 lux, the PCE and FF were increased from 11.6% to 12.9% and 0.53 to 0.58, respectively. Therefore, the PCEs were improved by 24.4% and 11.2% at 200 and 500 lux, respectively. On the other hand, under illumination condition of WLEDs, the standard device exhibited photovoltaic performance with a V_oc of 0.60 V, a J_sc of 32.6 µA cm⁻², and a FF of 0.48, results in a PCE of 13.2%. For the plasmonic device, J_sc and FF were increased to 34.0 µA cm⁻² and 0.54 respectively, resulting in a higher PCE of 15.5% at 200 lux. At 500 lux, PCE and FF were also increased from 11.6% to 13.2% and 0.55 to 0.58, respectively. Their PCEs were increased by 17.5% and 13.1% at 200 lux and 500 lux, respectively. The photovoltaic parameters obtained under illumination from the FT and the WLED are summarized in Table S2 and Table S3, respectively (Supplement 1). These results implied that the alloy NPs could also enhance the device efficiencies under dim-light indoor lighting conditions.

 

Fig. 5. The J-V curves of the OPVDs prepared with and without Au-Cu alloy NPs recorded under illumination from indoor light sources at 200 and 500 lux; the light source was (a) a TL5 fluorescent tube (FT); (b) a white light-emitting diode (WLED).

Download Full Size | PDF

Finally, we also evaluated the stability of the OPVDs prepared with the Au-Cu alloy NPs. The devices were encapsulated and stored in the air; the efficiencies of the devices were measured every day. As reveled in Fig. 6, we cannot observe any significant difference in the device stabilities between the reference device and the one prepared with alloy NPs, suggesting the addition of the NPs would not affect the stability. After 4 storage days, the OPVDs prepared with alloy NPs even exhibited slightly better stability. From such preliminary stability test, we can realize that the Au-Cu alloy particles should be stable enough for use in OPVDs.

 

Fig. 6. The result of stability test for the OPVDs prepared with and without Au-Cu alloy NPs. The devices were measured every day under standard one-sun illumination condition (100 mW cm⁻², AM1.5G).

Download Full Size | PDF

4. Conclusion

In conclusion, Au-Cu alloy NPs are synthesized for triggering the LSPR effects in OPVDs. The addition of Au component in Cu NPs enhances the stability of the NPs. Therefore, as compared with Cu NPs, the alloy plasmonic nanostructures simplify the fabrication processes. For instance, we can anneal the PEDOT:PSS films containing the metal NPs directly in the atmosphere. Meanwhile, the cost of the NPs remains lower than using NPs prepared with pure noble metals. The results in this work clearly imply that the use of the alloy NPs improve the device performance of the OPVDs. In addition to the normal illumination condition from the standard solar irradiation, we also found the LSPR effects improve the efficiencies of the OPVDs under indoor lighting conditions. Finally, the result of the stability test indicates that the incorporation of Au-Cu NPs would not affect the stability of the OPVDs. We anticipate that the results in this study open up a new avenue for plasmonic-enhanced OPVDs featuring low cost, stable nanostructures.