High-Efficiency Double Absorber PbS/CdS Heterojunction Solar Cells by Enhanced Charge Collection Using a ZnO Nanorod Array.

The device architecture of solar cells remains critical in achieving high photoconversion efficiency while affordable and scalable routes are being explored. Here, we demonstrate a scalable, low cost, and less toxic synthesis route for the fabrication of PbS/CdS thin-film solar cells with efficiencies as high as ∼5.59%, which is the highest efficiency obtained so far for the PbS-based solar cells not involving quantum dots. The devices use a stack of two band-aligned junctions that facilitates absorption of a wider range of the solar spectrum and an architectural modification of the electron-accepting electrode assembly consisting of a very thin CdS layer (∼10 nm) supported by vertically aligned ZnO nanorods on a ∼50 nm thick ZnO underlayer. Compared to a planar electrode of a 50 nm thick CdS film, the modified electrode assembly enhanced the efficiency by ∼39% primarily due to a significantly higher photon absorption in the PbS layer, as revealed by a detailed three-dimensional finite difference time-domain optoelectronic modeling of the device.

Introduction

Solution-processed PbS-based thin-film solar cells have drawn tremendous research interest in recent years for their potential in large area, high throughput, and affordable solar energy conversion.[1−5] Intensified research has culminated in a remarkable progress in the photoconversion efficiency (PCE) from <1% to better than 8% in a time span of less than a decade.[1−6] The high performing solar cells utilize a two dimensionally constricted assembly of individually surface passivated PbS quantum dots, which assist in harnessing a wide range of the solar spectrum due to their band gap tunability. Notwithstanding the favorable optoelectronic properties, the surface chemistry resulting from the passivation of the quantum dots and their packing have been shown to critically affect the performance of the devices.[7,8] The concerted efforts over the years have witnessed diverse techniques employed to achieve the desired passivation with substantially reduced density of trap states on surface and the close packing of the quantum dots, which is critical for efficient exciton dissociation and charge transfer. Despite the promise and rapid progress of the PbS quantum dot-based solar cells, the complex and elaborate processes of surface passivation and poor stability have raised serious issues on the fabrication of scalable low-cost devices.[9,10]

While seeking an affordable approach to synthesize the PbS absorber layer that would offer the facile tunability of band gap similar to that of the quantum dots without involving multistep chemical processing and/or without ligand engineering, recently, we have shown that the chemical bath deposition (CBD) method could be a viable option.[11,12] For the CBD PbS thin films, the optoelectronic properties, including band gap, are determined by the process parameters, such as temperature, composition, etc., that can be easily tailored. A promising PCE, as high as 4.03%, was obtained for photovoltaic devices employing a stack of a pair of CBD PbS layers of varying band gaps.[12] The architectural modification involving stacking of the band gap-tailored PbS films without any intervening layer facilitated better absorption of a wider range of solar spectrum resulting in better PCE, compared to a single PbS layer. Although the obtained photovoltaic performance is the highest for any PbS thin film-based device not involving colloidal quantum dots,[13,14] we seek to address the need for further improvement in the PCE of the devices.

One of the limiting factors of the performance of the PbS quantum dot-based solar cells has been the inherent compromise between the light absorption and carrier extraction.[4] Although a complete absorption of above band gap photons requires a film thickness of 1 μm, the very short diffusion length of carriers has compelled the thickness of the PbS layer to a smaller length scale of less than 250 nm for their efficient extraction. Recent attempts to enhance the current from this limited thickness of the absorber layer have witnessed the development of depleted p–n heterojunctions, which ensured the transport of excitons through drift compared to the diffusion-only mode by leveraging the built-in electric potential in the depletion region.[15] Although the depleted heterojunction solar cells fared better than their Schottky counterparts, their efficiencies were below that expected from the near-IR (NIR) band gap PbS layers. Concurrent widening of the depletion width and increasing thickness of the absorber layer were suggested to overcome this performance limitation. By extending this concept in an architecture engineering approach, vertically oriented nanowire arrays fully infiltrated with quantum dots, now known as bulk depleted heterojunction, have shown better photovoltaic characteristics.[16−19] These devices enable exciton dissociation and transfer in the nanoscale regime taking advantage of the nanostructured underlayer, wherein the vertically aligned one-dimensional structure provides a higher interfacial area between the active material and efficient electron transport pathways.[16−20] However, the photoconversion efficiency (PCE) is still below that desired for commercialization and scalable production of PbS-based photovoltaics, necessitating exploration of other avenues to improve performance.

Our approach to boost the performance of the cells is based on devising an alternate electron-accepting electrode assembly. So far, the efforts on the bulk depleted heterojunction have concentrated on deposition of PbS quantum dots directly on n-type TiO2[17−19] or ZnO nanorod (NR) arrays (NRAs),[20−22] wherein the uniformity of NR length, diameter, and spacing are critical in that quantum dots must reach the bottom of the rods for efficient coupling. We envisioned that this experimental limitation can be overcome with an alternate electrode assembly of an extremely thin planar n-type layer sandwiched between the PbS film and the NR array. In such a configuration, the PbS layer can be fully photoexploited from its limited thickness of sub-250 nm and can contribute to a higher current collection, all the while avoiding the complex processing of the PbS quantum dots. We show in this work that the performance can be significantly improved by depositing the PbS absorber layer (∼250 nm) on a very thin CdS film (∼10 nm) supported by vertically aligned ZnO NRs (∼200 nm long) on a ∼50 nm thick ZnO underlayer: the PCE increases by a massive 39% compared with the reference device that used only a planar electrode in the form of a 50 nm thick CdS film.

Results and Discussion

Figure a represents the schematics of the ZnO-based solar cell structures, that is, planar and NR-based structures. Solar cells with double PbS absorber layers having optical bandgaps of 0.92 and 1.61 eV were prepared on the fluorine-doped tin oxide (FTO)-coated soda lime silicate glass substrate by solution-based processes. Three different device architectures were studied in this work. The reference device has the structure of Al/PbS(0.92 eV)/PbS(1.61 eV)/CdS/FTO/glass. The other two devices included ZnO layers, namely planar (Al/PbS(0.92 eV)/PbS(1.61 eV)/CdS/ZnO/FTO/glass) and NR-based (Al/PbS(0.92 eV)/PbS(1.61 eV)/CdS/ZnO NR/ZnO/FTO/glass). The length and diameter of the NR were ∼200 and ∼45 nm, respectively.

Figure 1

(a) Schematics of device structures of planar and ZnO NR-based solar cells. (b) Scanning electron microscopy (SEM) micrographs of ZnO NR array and transmission electron microscopy (TEM) images of (c) ZnO NR and (d) PbS/CdS/ZnO NR array/ZnO/FTO/glass solar cells. Inset of (c) shows the lattice fringes corresponding to (0002) planes of wurtzite phase of ZnO. SEM surface images of PbS thin films deposited at (e) 40 and (f) 80 °C. Corresponding fast Fourier transform (FFT) diffraction patterns of PbS thin films deposited at 40 and 80 °C are shown.

Figure b shows a top view SEM image of the array of ZnO NRs grown on the seed layer on the FTO glass substrate. The image clearly shows the growth of near-parallel vertically aligned and uniformly spaced ZnO NRs. The diameter of the NRs was in the range of 25–45 nm, whereas the inter-NR spacing between the ZnO NRs was about 50–100 nm. It is worth noting that the inter-NR spacing is an important factor to obtain pore-free uniformly spaced ZnO NRs. The diameter of the NRs was in the precursor solution between the NRs thereby increasing the junction area.[21,23] The inter-NR spacing between ZnO NRs in this study is assumed to be appropriate for the desirable structure. Figure c depicts the TEM image of the ZnO NRs. The NRs of wurtzite phase were uniform in size without tapering along the growth direction. The d-spacing of 0.26 nm between two adjacent (0002) lattice planes in the NRs was identified, as shown in the inset of Figure c. The X-ray diffraction (XRD) pattern of the ZnO NRs also exhibited preferred c axis orientation, as demonstrated in Figure S1 of the Supporting Information. Figure d shows a cross-sectional TEM image of the PbS/CdS/ZnO NR array/ZnO/FTO structure. Owing to the presumable hydrophilic nature of the precursor solutions, the aqueous sulfide precursor solution is deeply infiltrated into the interior of ZnO NRs and then forms effectively the sulfide layers without pores or voids. Figure e,f shows the surface microstructures of PbS thin films deposited at 40 and 80 °C, respectively, which demonstrates significant difference in the crystallite size. Corresponding fast Fourier transform (FFT) diffraction patterns of the PbS thin films are also shown. The d-spacing values of 0.34 and 0.30 nm were observed for the (111) and (200) planes of PbS films processed at 40 and 80 °C, respectively.

The measured current density–voltage (J–V) curves for devices under AM1.5 G (100 mW/cm2) solar irradiation are shown in Figure a. The representative values of the key parameters, such as VOC, JSC, FF, and η for the devices have been listed in the inset table of Figure . Compared to the reference, that is, the reported device without the ZnO layer and NRs,[12] the devices exhibited noticeable improvements in the performance. The device with the electron-accepting electrode assembly consisting of a 10 nm thick CdS supported by a 50 nm thick ZnO underlayer demonstrated a photoconversion efficiency (PCE) of about 4.35%, with JSC = 23.7 mA/cm2, VOC = 0.332 V, and FF = 55.1%. The marginal improvement in PCE originates from the increase in VOC from 0.276 to 0.332 V. The device with the NR-supported electrode assembly showed superior photovoltaic properties, with a nearly 39% improvement in PCE with respect to the reference device. It eventually resulted in a PCE of 5.59 ± 0.18% with a significant enhancement in JSC. Considering the planar and the NR-based devices, the increase of the junction area might cause an increase of surface recombination and resulted in a decreased VOC and FF in the NR-based device.[21] Even with the decreased VOC and FF values, the significantly improved JSC led to the highest PCE. The origin of the boost in JSC is due to the enhanced exciton dissociation and charge collection facilitated by the NR array.[21,24,25] ZnO NR can create a short and direct pathway for charge transport and increase the p–n junction area between the ZnO NR-based CdS and the PbS absorber, resulting in a higher photocurrent.[26]

Figure 2

(a) J–V characteristics measured under AM1.5 G illumination for representative planar (Al/PbS(0.92 eV)/PbS(1.61 eV)/CdS/ZnO/FTO/glass) and NR-based (Al/PbS(0.92 eV)/PbS(1.61 eV)/CdS/ZnO NR/ZnO/FTO/glass) devices. (b) Schematic cross-sectional view of the NR-based solar cell and (c) the actual energy band diagram across dotted line in (b). The PbS layer represents stack of two PbS layers of band gap of 1.61 and 0.92 eV. Top contact Al is deposited onto the 0.92 eV band gap PbS layer. The NRs provide a direct path of electron transfer to the electrode. The photovoltaic characteristics of the reference sample was reported in ref (12).

The ZnO NR-based device is expected to enhance the absorption range from the visible to near-IR region, which implies effective collection of electrons in the ZnO NR array device that has a longer penetration depth compared to that of the planar ZnO-based device. The proposed mechanism is depicted in Figure b,c, wherein a schematic cross-sectional view and the corresponding energy band diagram are, respectively, given. As shown in Figure b, the vertically aligned ZnO NRs are likely to offer an efficient conduction path by supplying a shorter charge-transport pathway to the electrode. The band alignment between the CdS and the stacked PbS layers has been discussed in detail earlier,[12] wherein we have suggested that the wider band gap PbS layer (band gap of 1.61 eV) being only 100 nm thick is depleted entirely and the splitting of the electron and hole quasi-Fermi levels under illumination strongly favored efficient carrier transport. As presented in Figure c, the corresponding energy level diagram of different layers in the device referenced to the vacuum level was depicted on the basis of the reported energy levels.[11,27] The valence band edge of the CdS layer being far below that of the adjacent PbS layer results in a large barrier for hole injection into the electron-accepting CdS layer, and simultaneously a large conduction band offset of 0.96 eV prevents the back transfer of the photoexcited electrons from the CdS to the PbS layer, leading to the suppression of carrier recombination.[11,28,29] In addition, the energy values of the FTO and ZnO layers indicate strongly favorable band alignment for efficient charge collection. On the other hand, for the PbS (1.61 eV)/PbS (0.92 eV), a staggered type II heterojunction where hole concentration differs by two orders, the electron and hole quasi-Fermi levels split under illumination.[11] Starting from the CdS/PbS (1.61 eV) junction, the splitting gradually increased to ∼0.3 eV and marginally decreased at the PbS (1.61 eV)/PbS (0.92 eV) junction and remained constant at ∼0.25 eV throughout the 0.92 eV PbS layer. The favorable sloping of the quasi-Fermi levels facilitates carrier transport for better photovoltaic performance. While this work presents the highest efficiency of any PbS thin film-based solar cells not involving colloidal quantum dots, the scope for further improvement lies in elucidating the effects of NR dimensions, density, inter-NR spacing, etc.

We validated our experimental results on positive contribution toward enhancement of photovoltaic characteristics through the three-dimensional finite difference time-domain (3D-FDTD) optoelectronic modeling. Figure shows the device architectures used for the modeling and the FDTD results for the electric field intensity (E2) distribution for the reference device and devices with the planar and NR ZnO underlayer. As observed from the color map (Figure b), in the case of NR arrays, an enhancement in the electromagnetic field intensity occurs in the volume in the vicinity of the NRs. This increases effectively the volume from which the photogenerated carriers can be extracted,[17] compared to the case of planar electrode ZnO (50 nm)/CdS (10 nm), indicating the beneficial optical contribution of the ZnO NR array for the PbS-based photovoltaic cells.

Figure 3

(a) Schematic diagrams and (b) simulated electric field intensity distribution (illumination at 600 nm) for PbS/CdS/FTO (reference), PbS/CdS/ZnO film/FTO (planar), and PbS/CdS/ZnO-NRA/CdS/PbS (NR-based) devices.

The FDTD method was further used to theoretically estimate the optical absorption in appropriate portion of the AM1.5 G solar spectrum (i.e., from 300 to 1500 nm) in the PbS layer. The devices shown in Figure a were considered for the simulation. A linearly polarized plane wave that propagates from outside of the FTO-coated glass substrates to the PbS film through the electron-accepting electrode assembly was considered to illuminate the devices. The plots of photon flux Φabs versus wavelength for the reference, planar, and NR cell structures are shown in Figure . The Φabs was obtained by multiplying the calculated absorbance of the PbS layer (Figure S2) in the wavelength region with the corresponding AM1.5 G solar irradiance (Figure S3, taken from ref (30)). The substantial enhancements in the number of photons absorbed in the entire region are clearly shown in case of the ZnO NR-based device. In particular, the increment is significant in the region from 550 to 1200 nm. Although the exact number of carriers generated due to the increased photon absorption could not be estimated due to a nonconstant multiplication yield, the improved light absorption is expected to seriously enhance the efficiency of the devices.[30] To validate the observed high current density for the NR-based devices, we calculated the integrated current density, which typically defines the upper limit of the obtainable short-circuit density. From the simulated photon absorption flux given in Figure , the integrated current density is calculated to be 28.3, 31.3, and 43.0 mA/cm2 for the reference, planar, and NR-based devices, respectively. These values are, as expected, higher than the experimentally obtained ones owing to various losses in the devices. Nevertheless, the higher absorbed photon flux generating more excitons in the absorber layer is likely to be the main contributor to the enhanced PCE value for the ZnO NR-based devices.

Figure 4

Simulated photon absorption flux (Φabs) in the PbS layer for the architectures shown in Figure a. The flux was calculated by multiplying the AM1.5 G solar irradiation with and absorption spectra estimated through the FDTD simulations.

Conclusions

We have reported the highest PCE of about 5.59% for the noncolloidal PbS thin film-based solar cells, where a stack of two band-aligned junctions facilitates absorption of a wider range of the solar spectrum. The achievement was possible by designing an alternate electron-accepting electrode assembly consisting of a very thin CdS layer supported by vertically aligned ZnO NRs prior to the application of double absorption layers. The highest efficiency corresponds to the enhancement of ∼39% compared to a reference device that used only a planar electrode in the form of a 50 nm thick CdS film. Combined with the confirmation of higher photon absorption in the PbS layer by the 3D-FDTD simulation, the substantial enhancement in JSC indicates that the proposed device architecture helps light absorption. This device architecture of solar cell may suggest an alternative processing potential for generating competitive low cost cells.

Experimental Section

The solar cells were prepared on the fluorine-doped tin oxide (FTO)-coated soda lime silicate glass substrate (Pilkinton, U.K.) by solution-based processes. Three different device architectures were studied in this work. The reference device has the structure of Al/PbS(0.92 eV)/PbS(1.61 eV)/CdS/FTO/glass. The other two devices included ZnO layers, namely planar (Al/PbS(0.92 eV)/PbS(1.61 eV)/CdS/ZnO/FTO/glass) and NR-based (Al/PbS(0.92 eV)/PbS(1.61 eV)/CdS/ZnO NR/ZnO/FTO/glass). Figure a represents the schematics of the ZnO-based solar cell structures, that is, planar and NR-based structures. Figures S4 and S5 of the Supporting Information demonstrate the preparation steps of each structure. As a first step for both structures, ZnO thin films were deposited using the spin-coating process. The solution for the ZnO films was prepared in ethanol with 0.5 M zinc acetate (Zn(CH3COO)2·2H2O) and 0.05 M monoethanolamine (C2H7NO). The deposited films were annealed at 400 °C for 10 min in air to obtain a 50 nm thick nanocrystalline ZnO layer. For the growth of ZnO NR, the ZnO film-coated substrates were vertically placed in another aqueous solution bath containing 0.007 M zinc acetate, and 0.05 M hexamethylenetetramine (C6H12N4) at 95 °C for 3 h. The prepared samples were rinsed with distilled water and dried with N2 flow. The length and diameter of the NR were ∼200 and ∼45 nm, respectively. Figures S6 and S7 provide additional information on the effects of three processing parameters, such as concentrations of Zn acetate in spin-coating solution and NR growth solution, NR growth time, with regard to the morphology of NR, and the state of NR arrays.

To fabricate thin-film solar cells on the ZnO NR substrate, an aqueous solution containing 0.03 M cadmium nitrate (Cd(NO3)2·4H2O) and 0.15 M thioacetamide (C2H5NS) was prepared for n-type CdS thin films. The ZnO NR substrate was vertically immersed into the solution at 25 °C for 30 min to obtain ∼10 nm thick CdS thin films. Then, PbS absorber layers were grown onto the CdS/ZnO NR/FTO/glass in an aqueous solution containing 0.05 M lead nitrate (Pb(NO3)2), 0.04 M triethanolamine (C6H15NO3), 0.2 M sodium hydroxide (NaOH), and 0.06 M thiourea (CH4N2S). Double absorber layers were adopted to induce absorption of a wider range of solar spectrum. According to our previous work,[12] stacking of two PbS layers by facing a larger bandgap layer with the incoming light yielded an enhanced efficiency by as much as 30% compared to the single layer approach. In the identical way, here, the samples were first reacted in the PbS precursor solution at 40 °C for 1 h and then in a different bath containing the same precursor solution at 80 °C for 3 min. Bandgaps of the PbS thin films were identified as 1.61 and 0.92 eV, respectively. Total thickness of the double PbS layers was around 250 nm. After the sequential deposition of the PbS layers, an Al electrode of thickness of ∼100 nm was deposited by thermal evaporation.

Microstructures were observed by field emission scanning electron microscopy (JSM-7001F; JEOL) and transmission electron microscopy (TEM, JEM-2100F; JEOL). Optical transmittance and absorbance were measured by a UV–visible NIR spectrophotometer (UV 3101; Shimadzu). Structure of the ZnO NRs was analyzed using an X-ray diffractometer (XRD, Max-2500; Rigaku) under Cu Kα radiation. The depleted heterojunction solar cells with an active area of 0.09 cm2 were measured by using a current–voltage analyzer (IviumStat; Ivium Technologies) and a solar simulator (Sun 2000; ABET Technologies) under an AM1.5 G condition without any mask.