1. Introduction

Organic solar cells (OSCs) are an alternative to inorganic solar cells because of promising properties of OSCs such as lightweight, ease of production, environmental friendliness, high mechanical flexibility, and low cost. Recent developments in the OSCs have led to a tremendous increase in their certified power conversion efficiency (PCE) of up to 18.2% [1–3]. However, they still lag behind their inorganic (silicon and perovskite) counterparts [4]. The low performance of the OSCs is primarily due to their limited sunlight absorption because of a lack of optimal materials that allow absorption in the whole range of solar spectrum and due to the undesirable energy loss within the OSC. Energetic disorder at the interfaces and in the bulk, including structural disorder, also reduce the OSC performance [4].

The performance of OSCs can be improved by incorporating narrow bandgap organic semiconductor materials, which expand the wavelength range where the efficient absorption occurs up to the near infra-red range. PCE of the OSC is significantly limited due to the large differences between light and charge collection efficiencies of organic semiconductor materials [5,6]. One of the ways to improve the absorption of photons is to increase the thickness of the OSC active layer; however, organic semiconductors usually show low charge-carrier mobility and short diffusion lengths of excitons, which require small layer thicknesses in order to increase the efficiency of charge extraction and suppress charge recombination.

PCE of OSCs with a relatively thin active layer and hence low light absorption can be increased by improving its light capture properties. To increase the PCE, extended research efforts have been made to merge the OSC with nanoplasmonics [7,8]. The key idea behind these efforts is largely to use the light-trapping abilities of metal NPs, which are embedded into the active or buffer layer of the OSC or into both layers, or to increase the light absorption because of the plasmon resonance [9–11]. Experiments with metal NPs of Ag [12] and Au [13] embedded into the active/buffer layers or into both of them demonstrated a visible increase in the PCE of the OSCs, as well as for those with embedded semiconductor NPs of CdSe [14], CdTe [15], PbS [16], PbSe [17], Cu$_2$S [18,19], and ZnO [20]. Also, by introducing optically transparent nanostructures, the dimensions of which are comparable to the wavelengths of the incident light, it is possible to achieve strong light scattering as a result of morphology-dependent light interference. This is known as Mie scattering.

Recent studies of NPs made of dielectric material show that they provide much more control of their radiation pattern (due to the complex interference of dipole-electric and dipole-magnetic, as well as quadrupole contributions) [21] compared to metal NPs. One of the cheap and widely available materials is SiO$_2$, of which NPs can be easily fabricated and which are also well-suited for embedding into the OSC layers. Arrays of NPs such as SiO$_2$ are considered as excellent candidates for such applications because they absorb and reflect very little incident light and can generate intense forward scattering.

The effect of SiO$_2$ NPs of 25, 50, 75, and 100 nm in size embedded in a P3HT:PCBM photoactive layer and a ZnO-based electron transport layer was experimentally studied in Ref. [22]. For NPs embedded in the buffer layer, the authors demonstrated the PCE increase from 3.08% to 3.69% for 50 nm-sized SiO$_2$ NPs and then the PCE decreases for larger embedded NPs (75 and 100 nm). Note that the PCE of 3.08% corresponded to the OSC without NPs.

In this paper, the influence of spherical SiO$_2$ NPs with sizes of 20, 50, and 80 nm in different concentrations embedded in the PEDOT:PSS buffer layer on the efficiency of the OSCs was studied in detail both theoretically and experimentally. Organic layers were deposited using two techniques: spin-coating and doctor blade. Measurements and analysis of the voltage-current characteristics of the fabricated OSC samples were performed, their EQE spectra were obtained, and the surface morphology of the buffer layer with embedded NPs was studied. The obtained results are in agreement with the experimental results published in [22]. In the theoretical part of this work, we calculated the interaction of radiation with a model three-layer system (ITO, buffer layer, active layer) with embedded NPs in the buffer and active layers, analyzed the scattering radiation patterns of various sizes of spherical SiO$_2$ NPs in such a three-layer system. The presence of the organic layers around NPs significantly affects the asymmetry of the scattering patterns, as compared to the radiation pattern of the NPs in free space. A new aspect of this work is also the study of the effect of NPs on the efficiency of OSCs based on small molecules, whereas the polymer solar cells with NPs were also studied previously.

2. Materials and methods

Active layer components. Donor-acceptor bulk-heterjuncion OSCs were studied, their schematic is shown in Fig. 1(a). A fullerene C$_{70}$ derivative was used as the acceptor material: methyl ether [6,6]-phenyl-C$_{71}$-butanoic acid (PC$_{71}$BM, Fig. 1(b)). The star-shaped donor-acceptor oligomer N(Ph-2T-DCV-Et)$_3$ was used as the donor material [23–25] (Fig. 1(c)).

 

Fig. 1. Schematic of OSC (a), structural formula of PC$_{71}$BM (b) and of N(Ph-2T-DCV-Et)$_3$ (c).

Download Full Size | PDF

Materials of the electrodes. As a transparent electrode (anode), glass substrates of 23x23x1.1 mm in size with patterned ITO(indium-tin oxide) layers (XinYan) were used. A polymer complex of poly-3,4-ethylenedioxythiophene and polystyrene-sulfonate (poly-3,4-ethylenedioxythiophene: polystyrenesulfonate, PEDOT: PSS 1:6) in the form of an aqueous suspension (Heraeus) with a concentration of 15 g/L was used as a hole-transport (buffer) transparent layer, which was in direct contact with the active layer. As a cathode electrode, a calcium/aluminum bilayer was deposited in high vacuum.

Fabrication of organic solar cells samples. The process of OSC sample fabrication includes the following steps: preparation of solution for the active layer, cleaning of substrates, deposition of organic layers by wet processing and vacuum deposition of top metal electrodes.

Preparation of solutions for deposition of the active layer. First, dry components (powders) of the active layer with the mass ratio of the donor and acceptor 1:2. were placed in a clean glass vial. After that the solvent was added to the dry component to provide the desired total concentration of the component (typically 24 g/L). As a solvent, we used orthodichlorobenzene (Acros Organics). The solution was stirred on a heated magnetic stirrer at 75$^\circ$ C for 18–24 hours.

Preparation of organic layers with embedded SiO$_2$ NPs. For preparation of an organic electrode layer with embedded NPs, water suspensions of spherical SiO$_2$ NPs with diameters of 20, 50, and 80 nm with concentrations of 5, 10 and 15 g/L(Nanocomposix), respectively, were used. These suspensions were mixed with water suspension PEDOT:PSS (15 g/L) in various proportions by volume: 1:3, 1:5, and 1:7. We will denote a mixture of PEDOT:PSS with SiO$_2$ NPs of 20 nm in size, mixed in a 1:$x$ ratio by volume, as SiO$_2$-20:PEDOT:PSS 1:$x$. The percentage of surface coating with NPs at a given concentration was estimated by using atomic force microscopy.

Deposition of organic layers by spin-coating and doctor blade techniques. Organic layers—PEDOT:PSS: SiO$_2$ and the active layer—were deposited on the surface of the cleaned substrate by two methods: spin-coating (G3, Spin Coating System) and a much more economical method—the doctor blade technique with an adjustable blade speed and a heated table.

Spin-coating technique. The volume of PEDOT:PSS suspension (without NPs) deposited on the substrate for preparation of a buffer film was 0.3 ml, the acceleration time was 4 s, the rotation speed was selected so that the film thickness was about 50 nm, the rotation time at this speed was 120 s. Then a preliminary series of experiments was conducted to select spin-coating parameters for obtaining films of PEDOT:PSS with SiO$_2$ NP with the required thickness and concentration of NPs. Film thicknesses were determined from AFM data. The substrates covered by the buffer layers were annealed on a hotplate at a temperature of 140$^{\circ }$ C for 15 min to remove water residues. Then, the active layer was deposited to the substrates with the buffer layers. The volume of the solution per sample was 0.2 mL, the acceleration time was 5 s, the rotation speed was 600 rpm, and the rotation time was 120 s. The thickness of the active layer with these parameters was $80\pm 5$ nm according to AFM (atomic-force microscope) data. The AFM data were recorded with a scanning probe microscope (NTEGRA Spectra, ND-MDT).

Doctor blade technique. The volume of PEDOT:PSS suspension without or with NPs deposited to the substrate for preparation of a buffer film was $20 \mu$L, the velocity and temperature of the heated table were 15 mm/s and 95$^{\circ }$ C. After applying the suspensions, the substrates were annealed on a hotplate at a temperature of 140$^{\circ }$ C for 15 min. Then, the active layer was deposited to the substrates with the buffer layers. The volume of the solution per sample was $20 \mu$L, the velocity and temperature of the heated table were 15 mm/s and 85$^{\circ }$ C. The thickness of the active layer with these parameters was $80\pm 5$ nm (according to AFM data).

Deposition of metal electrodes. The substrates with the deposited buffer and active layers were moved to an argon-filled glove box integrated with a vacuum chamber (Univex 300, Leybold). Ca/Al bilayer electrodes were deposited at a residual pressure of less than $5\times 10^{-6}$ mbar with the pumping time of about 30 minutes. The total thickness of the Ca/Al bilayer was 100 nm. The electrodes were deposited through shadow masks that allow forming 8 working pixels (i.e., 8 separate solar cells) in the form of circles with a diameter of 2.2 mm with square contact pads.

Measurements of the photovoltaic characteristics and external quantum efficiency. All photoelectric measurements were performed at room temperature in a sealed glove box filled with argon for several hours after fabrication of the samples.

For measuring the volt-ampere characteristics (VAC), the OSC sample was illuminated from the side of the glass substrate through a circular aperture with a diameter of 2 mm (the centers of the aperture and the pixel coincided). The contact pad of the pixel and the contact of the transparent electrode were applied to the probes of a source-meter (SourceMeter 2400, Keithley), which measured both current and voltage. This device was controlled by a computer using a special code allowing measurement of the VAC in different voltage ranges. The VAC were recorded for the OSC samples illuminated with radiation from a solar simulator (Newport 67005) equipped with a special filter (AM1.5G spectrum, Newport). The incident power at the sample was 3.14 mW, which corresponded to the intensity of 100 mW/cm$^2$.

External quantum efficiency (EQE) spectra were measured with the use of a laser-driven light source (LDLS EQ-99X, Energetiq) equipped by a monochromator (CS130-USB-3-MC, Newport), a power sensor (S120UV, Thorlabs) and the source-meter. To avoid higher diffraction orders, additional filters KG3, GG400, OG550 (Newport) were used for 380-500, 480-620 and 600-800 nm spectral ranges, respectively.

3. Results and discussion

In our experiments, we used SiO$_2$ NPs of three different sizes embedded to the buffer (PEDOT:PSS) layer. The average diameter of NPs in each group was 20, 50, and 80 nm, respectively. Figure 2 shows AFM data for PEDOT:PSS with SiO$_2$ NPs films with different volume ratio SiO$_2$:PEDOT:PSS $\nu$: (a) 20 nm, $\nu$=1:1, (b) 50 nm, $\nu$=1:2, (c) 80 nm, $\nu$=1:2, which were prepared by spin-coating. The AFM images of the samples prepared by doctor blade technique were very similar, so they are not shown. Based on the AFM data, Fig. 6(a-c) schematically shows the distribution of NPs in the buffer layers for three sizes of NPs. In all three cases, the NPs protrude above the surface of the buffer layer by an amount approximately equal to their diameter. In the experiment, the conditions of layer deposition were selected so that the thicknesses of the active and buffer layers did not change in all experiments: $55\pm 5$ nm for the buffer layer and $80\pm 5$ nm for the active layer. Thus, for OSC with NPs of 20 and 50 nm in size, the rough surface of the buffer layer with NPs was completely covered by the active layer. The AFM image of the active layer after being applied to the sample 2(b) is available in the Supplement 1 in Fig. S1.

 

Fig. 2. AFM images of the buffer layer with SiO$_2$ NPs deposited by spin-coating. Sizes of SiO$_2$ NPs are 20 nm (a), 50 nm (b), and 80 nm (c).

Download Full Size | PDF

For the NPs of 80 nm in size, our studies did not show any increase in PCE of the OSC samples at any concentrations with the use of the spin-coating or doctor blade technique. This case corresponds to Fig. 2(c), where the NPs protrude above the buffer layer by an amount greater than the thickness of the active layer. Therefore, even at a low NPs concentration used in the experiment, there could be NPs that protrude through the active layer (Fig. 6(c)). In this case, the electrodes were deposited on a rough surface, which in turn could worsen the contact of the active layer with the cathode and lead to an increase in the series resistance, a decrease in the short-circuit current and the fill factor. Moreover, large NPs could increase the likelihood of shunts and leakage currents, which decrease the open-circuit voltage.

In our experiments, an increase in EQE was obtained for samples with NPs of 20 and 50 nm in size. We will further compare the device characteristics for these samples. Figure 3 shows the PCE for OSCs without NPs and with the SiO$_2$ NPs (20 and 50 nm) embedded in the buffer layer in different concentrations (1:3, 1:5, 1:7) for the two deposition techniques. The numerical data for average photovoltaic parameters are given in Table S1 of Supplement 1. Figures. 4 and 5 show VAC and EQE of the OSC samples with and without the embedded NPs prepared by spin-coating and doctor blade techniques, respectively. For each method of film deposition embedding the NPs into the buffer layer with the diameter of 20 and 50 nm at different concentrations leads to a significant shift of the EQE spectra maximums and an increase in PCE.

 

Fig. 3. Maximum and average (over eight devices) values of PCE for OSCs deposited by spin-coating and doctor blade techniques. Size of SiO$_2$ NPs and volume ratio SiO$_2$ : PEDOT:PSS=1:$x$ are indicated below the graphs.

Download Full Size | PDF

 

Fig. 4. VAC of samples fabricated by spin-coating (left) and by doctor blade (right) with and without the embedded NPs.

Download Full Size | PDF

 

Fig. 5. As measured (left column) and normalized (right column) EQE spectra (the spectra were normalized to the maximum) of OSC samples with and without the NPs fabricated by spin-coating (the top row) and doctor blade (the bottom row) techniques.

Download Full Size | PDF

For the spin-coated samples, the PCE increased from 3.57% to 4.15% for NPs of 20 and 50 nm in size. The highest PCE was observed for the volume ratio of 1:5. For the samples prepared by the doctor blade technique, the characteristics of the OSCs are improved compared to the samples prepared by spin-coating: the PCE of the reference sample without NPs increased up to 4.37%. For the OSCs with embedded SiO$_2$ NPs the PCE increased up to 4.98%. The doctor blade technique resulted in the higher PCE because of an increase in the short-circuit current and fill factor.

It is well accepted that in bulk heterojunction solar cells, charge generation and recombination occur mainly in the bulk of the active layer. Since NPs are in the buffer layer, they do not participate directly in charge generation. Although charge recombination can generally occur at the interface of the active layer with the NPs, $V_{\textrm{OC}}$ for our devices practically does not change ($\pm$ 0.02 V) with variation in the NPs concentration, and this is a sign of negligible effect of charge-carrier recombination [26]. We explain the obtained experimental results by Mie scattering on SiO$_2$ NPs in a three-layer medium (ITO, buffer layer, active layer). Numerical simulation was carried out in COMSOL Multiphysics software environment. The calculation method is described in section 2 of Supplement 1.

As discussed above, in the considered three-layer model, a part of the NPs are in the buffer layer, and a part in the active layer (Fig. 6(a)-c). Let us first consider the Mie scattering of NPs in the buffer layer (Fig. 6(d)). Refractive index of SiO$_2$ weakly depends on the wavelength and is approximately $n_{silica}=1.48$ (See Supplement 1, Fig. S2). The refractive index of PEDOT:PSS also weakly depends on the wavelength and is approximately $n_{bl}=1.5$ (See Supplement 1, Fig. S2). Thus, the refractive indices of the NPs and the buffer layer are very close, and hence the scattering cross section in the buffer layer is very low (see Fig. 6(f))). The scattering diagram changes in such a way that back-scattering is suppressed and the radiation is redirected towards the active layer. The refractive index of active layer was taken as $n_{al}=2.0$ based on the data from [27].

 

Fig. 6. Schematics of 20 (a), 50 (b) and 80 nm (c) in diameter spherical SiO$_2$ NPs embedded into the PEDOT: PSS (buffer) layer of the OSCs as shown in detail in panels (d) and (e). Scattering cross section versus the wavelength of incident radiation for the NPs embedded in the buffer (f) and active (g) layers. The upper rows in panels (f) and (g) show 3D scattering diagrams of these NPs at different wavelengths.

Download Full Size | PDF

As follows from our calculations, the scattering cross section of a NP embedded in the active layer by three orders of magnitude larger than that of the NP embedded in the buffer layer (see Fig. 6(g)). This figure also shows the scattered radiation patterns for several wavelengths. Since the shape of the scattered radiation patterns of a spherical particle does not qualitatively changes for the sizes of 20, 50 and 80 nm, we presented the scattered radiation patterns for NPs with a size of 50 nm.

These radiation patterns clearly show that for all sizes of the NPs at the wavelengths from 530 to 700 nm there is an increase in scattering to the sides, where the scattered light interacts with the photosensitive organic semiconductor molecules. The side scattering contributes to additional rescattering of radiation between NPs, which can lead to an increase of the EQE in this wavelength range. Indeed, this effect follows from the EQE spectra: both the values of the EQE and the shape of the EQE spectra changed (Fig. 5). An increase in the EQE values and a red-shifted maximum of the spectra explain the increased short-circuit current and hence PCE. Note that at wavelengths less than 530 nm, scattering perpendicular to the surface of the OSC prevails, resulting mainly in forward scattering.

4. Conclusions

In conclusion, we fabricated the OSCs samples based on a star-shaped donor-acceptor oligomer N(Ph-2T-DCV-Et)$_3$ with SiO$_2$ spherical NPs embedded in the PEDOT: PSS buffer layer. We found a steady increase in the OSC power conversion efficiency for the embedded SiO$_2$ NPs with the diameters of 20 and 50 nm. We explained these results by Mie scattering on SiO$_2$ NPs in a three-layered medium (ITO, buffer layer, active layer). The main contribution to the increase in PCE is made by NPs entering the active layer due to a change of the scattered radiation pattern. In the future, it is of interest to investigate the effect of NPs with a high refractive index on the efficiency of OSC, since for such particles it is possible to precisely control the directional pattern of scattered radiation [28]. The simultaneous introduction of NPs of different sizes, shapes (for example, nucleated particles, which essentially widens the range of the spectrum where absorption is effective) and materials (for instance, using hybrid structures), also promising, especially for the NPs made of high refractive index materials.